Beyond Prediction: Reinforcement Learning as the Defining Leap in Healthcare AI

Daniel J. Tan; Dilruk Perera; Erik Cambria; Gousia Habib; Kai He; Mengling Feng; Qianyi Xu

REVIEW 5 major objections 7 minor 1 cited by

Beyond Prediction: Reinforcement Learning as the Defining Leap in Healthcare AI

T0 review · 5 major / 7 minor · reviewed 2026-08-05 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A survey argues reinforcement learning marks a fundamental shift in healthcare AI: instead of only predicting outcomes, systems learn to choose actions that optimize long-term patient goals, making RL a prescriptive, agentive framework rath

desk verdict A broad but sloppy survey; the 'defining leap' framing outruns the evidence, and the fabricated reference makes it hard to trust. read the letter →

arxiv 2508.21101 v1 pith:B4FIGQV6 submitted 2025-08-28 cs.LG cs.AI

Dilruk Perera , Gousia Habib , Qianyi Xu , Daniel J. Tan , Kai He , Erik Cambria , Mengling Feng This is my paper

classification cs.LGcs.AI

keywords reinforcementlearninghealthcareAIclinicaldecision-makingdynamictreatmentregimesofflineinformationfusionrewarddesignoff-policyevaluation

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Reinforcement learning, the paper argues, is not just another predictive model applied to medicine. It changes the question healthcare AI is allowed to ask: instead of "what will happen?", the system learns "what should be done?" by optimizing a sequence of decisions toward long-term outcomes such as survival or glucose control. The survey reviews 246 studies and organizes them by RL method, clinical task, information-fusion architecture, and deployment challenge. Its intended contribution is a technical roadmap and a critical reflection: RL is promising, but safe deployment depends on reward design, conservative offline algorithms, off-policy evaluation, and human-aligned integration. The reader should care because, if this framing is right, the field's success metric shifts from static prediction accuracy to the quality and safety of interventions over time.

What carries the argument

The central object is the Markov decision process (MDP)—with state, action, transition probability, reward, and discount factor—and its partially observable extension (POMDP) in which the agent sees multi-modal observations instead of the true state. The policy learned by reward optimization is the unit that carries the argument from prediction to action. Supporting machinery includes conservative offline RL methods (BCQ, CQL, BRAC, IQL), off-policy evaluation (DM, IS, WIS, CWPDIS, DR), and information-fusion architectures that combine vitals, labs, notes, imaging, and telemetry at data, feature, or decision level.

What would settle it

Run a prospective randomized comparison of an offline RL policy (for example, a CQL-based mechanical-ventilation policy or a sepsis fluid/vasopressor policy trained on ICU data) against standard care, using 90-day mortality as the endpoint. If the policy does not beat usual care despite favorable off-policy estimates, the survey's defining-leap claim loses its empirical footing; a cheaper check is a silent pilot in which RL recommendations are logged against clinician actions and outcomes without being executed.

Watch

Extended reading notes

Core claim

The paper's central claim is that the maturation of offline and model-based reinforcement learning turns clinical AI from retrospective risk estimation into agentive decision-making. Using the MDP/POMDP formalism, it treats the patient as a partially observable state, clinician actions as interventions, and a designed reward as the clinical goal; the learned policy is the object that recommends next actions. Surveying applications from sepsis management, mechanical-ventilation weaning, sedation and anticoagulant dosing, diabetes and HIV therapy, diagnostics, and robotic surgery, the paper contends that RL systems have already shown they can recommend actions optimized for long-term outcomes

Load-bearing premise

The load-bearing premise is that policies trained on past patient records and judged by statistical estimates from those same records will actually improve care when used at the bedside; the paper itself concedes that most systems are trained on static retrospective data and lack prospective validation.

Editorial extensions

If this is right

If RL is the correct frame, clinical AI products must be evaluated on sequential decision quality—mortality, readmission, time-in-range—and not just on one-step accuracy, because the policy's value is the long-horizon return.
Conservative offline methods become the default safety layer: BCQ, CQL, BRAC, and IQL exist specifically to prevent out-of-distribution actions that inflate Q-values and recommend unsafe treatments.
Reward design moves to the center of clinical AI: proxy rewards like in-hospital mortality or SOFA scores are acknowledged as imperfect, and inverse RL, Bayesian IRL, and clinician-in-the-loop reward shaping are the main remedies.
Deployment depends on integration infrastructure: EHR interoperability, multimodal fusion, federated learning, and edge computing are the paper's routes to real-time, privacy-preserving policy learning.
Off-policy evaluation must be standardized; current comparisons across studies are inconsistent, and OPE is the only substitute for online trials when real-time experimentation is unethical.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

One direct test of the paradigm claim: run silent prospective pilots where the RL policy's recommendations are logged but not executed, then compare its decisions and estimated outcomes against actual clinician decisions. Success would support the leap; mismatch would expose the evaluation gap the paper concedes.
The 'defining leap' framing implies that many supervised models in dynamic treatment settings have been evaluated on the wrong outcome; re-benchmarking them as static policies against RL in sequential simulations would show where the paradigm gap is genuinely large.
The paper's taxonomy suggests that partial observability, not algorithm choice, may be the dominant source of policy failure; comparing POMDP belief-tracking against MDP simplifications on the same retrospective cohort would isolate that factor.
If federated RL matures, multi-hospital policy learning could become the norm, but the paper leaves open whether cross-site transportability is solved by federated training or only masked by aggregation.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Two things to know. First, this is a big survey of RL in healthcare with a genuinely useful organizing idea: looking at RL through an information-fusion lens and mapping clinical applications onto data/feature/decision fusion levels. Second, don't cite it until the reference list is cleaned up, because it contains a fabricated DOI and placeholder author names.

What's actually good: the taxonomy is broad, the coverage since 2020 is extensive (246 papers), and the sections on reward misspecification, off-policy evaluation, and deployment challenges are reasonable and would help a newcomer navigate the field. The authors do try to be critical about limitations, including the absence of prospective validation.

The soft spots are serious. The fabricated DOI is in reference [13] (Baucum et al.), with authors 'Doe, J., Smith, J.' and DOI 10.1145/1234567.1234568—that is not a typo, it's a placeholder. There's also a placeholder email in the author list. The promised full search strings in Figure 4 are not there; the figure shows terminology, not a reproducible search. The paper describes Ventai as tabular Q-learning in the text, but Table 5 lists FQI for that study. And the same Luther et al. finding appears twice with two reference numbers. For a survey, whose entire value is the reliability of its citations and methods, these are disqualifying.

The central claim is another issue. Calling RL 'the defining leap' is an interpretation, not a result, and the paper's own Section 8 concedes that most systems are trained on retrospective data and lack prospective validation. Worse, the paper cites Nauka et al. (2025) showing that the AI Clinician—the flagship sepsis RL model—fails to transport to the emergency department, with less than 1% of patients having complete data by hour 6. The survey never squares that counterexample with its paradigm-shift framing. That is a gap in the argument, not a minor quibble.

So who is this for? A reader who wants a checklist of RL papers in healthcare, with a critical angle, could get some value if the integrity issues were fixed. But as-is, the paper is not ready for peer review. I'd desk-reject it and invite the authors to resubmit after fixing the reference list, making the methodology reproducible, and tempering the 'defining leap' language to what the evidence supports.

Referee Report

5 major / 7 minor

Summary. This paper is a survey of reinforcement learning (RL) in healthcare (2020–2025) that argues RL represents a fundamental shift in healthcare AI—from passive prediction to active, long-horizon decision-making—and positions this as a 'defining leap' and a 'foundational framework for building intelligent, autonomous healthcare systems.' The paper claims to have systematically selected 246 papers from a structured search, and organizes the field into foundational RL concepts, clinical applications (critical care, chronic disease, diagnostics, surgery, operations), technical/ethical challenges, design trade-offs, and emerging frontiers (GEN-RL, MARL, FED-RL, EDGE-RL, GHE-RL, HITL). It uses an information-fusion lens throughout and includes a critical discussion of reward misalignment, offline policy evaluation, privacy, and explainability.

Significance. The paper covers a broad and timely range of topics and contains several correct critical observations: reward misspecification is a central obstacle; offline RL methods such as BCQ and CQL do not solve the safety problem; off-policy evaluation is not standardized; and the paper explicitly acknowledges the lack of prospective validation and the risks of training on retrospective, static data. These are useful messages for the RL-in-healthcare community. However, the significance of the paper as a research contribution is undermined by the mismatch between its strong 'defining leap' thesis and the evidence it itself presents, and by internal inconsistencies and reference-integrity problems. The paper does not provide a parameter-free derivation, reproducible code, or a falsifiable prediction; its value would lie in a trustworthy synthesis, which is not currently achieved.

major comments (5)

[§8; §5.6; Tables 4–5] The central claim that RL is a 'defining leap' in healthcare AI is not supported by the surveyed evidence. Section 8 concedes that 'most systems are trained on retrospective, static data, risking overfitting and unsafe generalization' and that 'current solutions also lack prospective validation.' Section 5.6 states that OPE methods trade off bias and variance and that evaluation 'remains largely inconsistent, urgently requiring a standardized benchmarking framework.' Tables 4 and 5 repeatedly label results as 'estimated' or 'offline-only evaluation; not prospectively validated' (e.g., Table 4, row [115]; Table 5, row [115]). The paper also includes reference [133], which documents that the AI Clinician—the most-cited sepsis RL system—fails to transport to emergency care (<1% complete data by hour 6; 49% of patients received no treatment), yet this counterexample is not reconciled with th
[§4.1.1; Table 5] The reporting of the Ventai/Peine et al. study is internally contradictory. The text in §4.1.1 says Peine et al. used tabular Q-learning and demonstrated a '42.6% improvement in estimated performance return' over clinicians. Table 5, however, lists the algorithm as FQI, the dataset as MIMIC-III/eICU, and the outcome as '42.6% reduced mortality, 83.3% performance return.' These are not compatible descriptions of the same study. A systematic survey's data extraction must be internally consistent; this discrepancy calls into question the reliability of the survey's headline claims for a key ICU application.
[References [13] and author contact block] The reference list contains an unverifiable citation: reference [13] is 'Baucum, M., Doe, J., Smith, J.' in ACM TIST with DOI 10.1145/1234567.1234568. This DOI is a placeholder format, and 'Doe, J.' is a placeholder author name. Similarly, the corresponding-author email 'wjh@example.org' is a placeholder. A submitted manuscript cannot contain such citations; this is a research-integrity concern and prevents readers from checking a cited study.
[§4.2.2 and §4.2.5; Refs [120]–[121]] The same Luther et al. psychosis-risk study is described twice with nearly identical limitations (n=46, limited generalizability) in §4.2.2 and §4.2.5, and references [120] and [121] are duplicate entries for the same article. This double-counting inflates the apparent coverage of the survey and indicates that the thematic synthesis was not carefully deduplicated.
[§2.1; Figure 4] The methodology section promises that 'the full search string combinations are detailed in Appendix A Figure 4.' Figure 4 is a PRISMA-style diagram containing a short list of phrases (e.g., 'Rl in Healthcare', 'RL in Drug Dosing and Sedation') and a list of database names; it does not contain the full Boolean search strings needed for reproducibility. Since the paper claims to be the most comprehensive systematic survey since 2020, this is a material gap in the methodology as reported.

minor comments (7)

[Abstract] 'This survey explore RL's rise' should be 'This survey explores RL's rise.'
[Abbreviations table] 'SER Safe Expeeriece Reshaping' contains a typo: 'Expeeriece' should be 'Experience.'
[§3.3] Equation (5) is used twice, once for value-based RL and again for actor-critic regression, creating duplicate equation numbering.
[§5.4] Typo: 'soltuio' should be 'solution.' Also in §1 'systmatic' should be 'systematic.'
[§8] Typo: 'healtcare' should be 'healthcare' in 'a cornerstone of healtcare AI culture.'
[References] Several references are duplicated with different numbers (e.g., [49]/[50], [186]/[187], [198]/[199], [175]/[176]). These should be deduplicated and consistently cited in the text.
[§2.2] The text says 16 additional papers are 'mentioned in Section 6,' but Section 6 does not explicitly list or discuss these 16 exceptions; please clarify where they are identified.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the survey's paradigm claim is an interpretive synthesis, and the few self-citations are illustrative rather than load-bearing.

full rationale

This paper is a literature survey, not a derivation. The central claim that RL is a 'defining leap' from prediction to action is an interpretive position supported by cited external studies; it is not derived from an equation fitted to its own outputs. Technical components such as the MDP, POMDP, Bellman objectives, and OPE methods are presented as textbook definitions and standard tools, not as results generated by the survey itself. The paper repeatedly and explicitly concedes the evidence gap: Section 8 states 'Current solutions also lack prospective validation' and 'Most systems are trained on retrospective, static data, risking overfitting and unsafe generalization,' and Table 5 marks several outcomes as 'estimated data' or 'offline-only evaluation; not prospectively validated.' These are honest limitations and constitute a correctness/overstatement concern, not circularity. The authors do cite their own prior work ([146] Smart Imitator, [212] Meddreamer, [177] HMARL, [65], [111]), but these citations function as examples of techniques or emerging trends, not as load-bearing proofs of the paradigm claim; removing them would not change the logical structure of the conclusion. Likewise, the survey includes [133] (Nauka et al.) documenting the AI Clinician's transportability failure in emergency care, which undercuts rather than props up the central transferability assumption. No fitted input is renamed as a prediction, no uniqueness theorem is imported from the authors' prior work, and no ansatz is smuggled in via self-citation. The evidentiary weakness is real, but under the specified circularity criteria the derivation chain is not circular; the proper finding is 'no significant circularity.'

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

No new physical or technical entities are introduced; the paper coins acronyms (GEN-RL, FED-RL, EDGE-RL, GHE-RL) for existing research directions, not postulates with independent evidence requirements.

assumptions (3)

standard math MDP/POMDP formalism captures clinical decision-making (Markov property, Bellman backup, partial observability)
Used throughout Section 3 as the mathematical backbone; standard model but an idealization of clinical reality.
ad hoc to paper The surveyed corpus (246 papers + 16 exceptions) is representative of RL-in-healthcare research 2020-2025
The comprehensiveness claim in Section 1 and conclusions depend on this selection; search strings and exception list are not fully disclosed.
domain assumption Off-policy evaluation on retrospective EHR data is a valid proxy for clinical benefit
Section 5.6 and Section 8 explicitly rely on OPE because online deployment is infeasible; the paper itself flags lack of prospective validation.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Beyond Prediction: Reinforcement Learning as the Defining Leap in Healthcare AI." pith.science (2026). https://pith.science/paper/B4FIGQV6

@misc{pith2026250821101,
  author       = {Pith},
  title        = {Pith review of: Beyond Prediction: Reinforcement Learning as the Defining Leap in Healthcare AI},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/B4FIGQV6}},
  note         = {Machine review of arXiv:2508.21101}
}

read the original abstract

Reinforcement learning (RL) marks a fundamental shift in how artificial intelligence is applied in healthcare. Instead of merely predicting outcomes, RL actively decides interventions with long term goals. Unlike traditional models that operate on fixed associations, RL systems learn through trial, feedback, and long-term reward optimization, introducing transformative possibilities and new risks. From an information fusion lens, healthcare RL typically integrates multi-source signals such as vitals, labs clinical notes, imaging and device telemetry using temporal and decision-level mechanisms. These systems can operate within centralized, federated, or edge architectures to meet real-time clinical constraints, and naturally span data, features and decision fusion levels. This survey explore RL's rise in healthcare as more than a set of tools, rather a shift toward agentive intelligence in clinical environments. We first structure the landscape of RL techniques including model-based and model-free methods, offline and batch-constrained approaches, and emerging strategies for reward specification and uncertainty calibration through the lens of healthcare constraints. We then comprehensively analyze RL applications spanning critical care, chronic disease, mental health, diagnostics, and robotic assistance, identifying their trends, gaps, and translational bottlenecks. In contrast to prior reviews, we critically analyze RL's ethical, deployment, and reward design challenges, and synthesize lessons for safe, human-aligned policy learning. This paper serves as both a a technical roadmap and a critical reflection of RL's emerging transformative role in healthcare AI not as prediction machinery, but as agentive clinical intelligence.

Figures

Figures reproduced from arXiv: 2508.21101 by the authors.

**Figure 1.** Taxonomy of Reinforcement Learning in Healthcare. The figure illustrates the core branches covered in this survey, including RL foundations, clinical and operational applications, technical and ethical challenges, and emerging frontiers such as generative RL, federated learning, and multi-agent systems. This taxonomy reflects RL’s evolution from prediction to action and provides a conceptual roadmap for the rest of … view at source ↗

**Figure 2.** RL in Healthcare: A Mapping by Application Type. and the management of critical conditions such as sepsis and mechanical ventilation (MV). Specifically, we highlight key studies and milestones demonstrating how RL is transforming clinical decision-making and improving patient outcomes across various medical domains. Recent research demonstrated that RL can be effectively utilized to support clinical decision-making… view at source ↗

**Figure 3.** Overview of Emerging Frontiers in RL Despite its success, MARL faces challenges in sample efficiency and explainability, which are critical for clinical trust. Future research should focus on improving these areas for better real-time performance in emergency healthcare contexts. 7.3. FED-RL Federated RL (FED-RL) is gaining traction in healthcare for its ability to preserve data privacy while enabling collaborativ… view at source ↗

Figures from the paper (1 more)

**Figure 4.** Figure 4: PRISMA Framework: A structured approach to conducting systematic reviews, detailing the stages from study identification to inclusion and data synthesis. Appendix B [PITH_FULL_IMAGE:figures/full_fig_p030_4.png]

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

CSPO: Alleviating Reward Ambiguity for Structured Table-to-LaTeX Generation
cs.AI 2026-04 unverdicted novelty 6.0 of 10

CSPO disentangles reinforcement learning rewards across LaTeX table components to reduce ambiguity and improve structured table image to LaTeX generation.

Reference graph

Works this paper leans on

281 extracted references · 55 canonical work pages · cited by 1 Pith paper

[13]

Optimizing substance use treatment selection using reinforcement learning

Baucum, M., Doe, J., Smith, J., 2023. Optimizing substance use treatment selection using reinforcement learning. ACM Transac- tionsonIntelligentSystemsandTechnology14,45–67.URL: https: //doi.org/10.1145/1234567.1234568, doi:10.1145/1234567.1234568

work page arXiv 2023
[115]

Reinforcement learning to optimize ventilator settings for patients on invasive mechanical ventilation: Retrospective study

Liu, S., Xu, Q., Xu, Z., Liu, Z., Sun, X., Xie, G., See, K.C., 2024b. Reinforcement learning to optimize ventilator settings for patients on invasive mechanical ventilation: Retrospective study. Journal of Medical Internet Research 26
[133]

Challenges with reinforcement learning model transportability for sepsis treatment in emergency care

Nauka, P.C., Kennedy, J.N., Brant, E.B., Komorowski, M., Pirrac- chio, R., Angus, D.C., Seymour, C.W., 2025. Challenges with reinforcement learning model transportability for sepsis treatment in emergency care. npj Digital Medicine 8, 1–5

2025
[120]

Probabalistic rein- forcement learning impairments predict negative symptom severity and risk for conversion in youth at clinical high-risk for psychosis

Luther, L., Raugh, I.M., Strauss, G.P., 2025a. Probabalistic rein- forcement learning impairments predict negative symptom severity and risk for conversion in youth at clinical high-risk for psychosis. Psychological Medicine 55, e28
[121]

Probabilistic rein- forcement learning impairments predict negative symptom severity and risk for conversion in youth at clinical high-risk for psychosis

Luther, L., Raugh, I.M., Strauss, G.P., 2025b. Probabilistic rein- forcement learning impairments predict negative symptom severity and risk for conversion in youth at clinical high-risk for psychosis. Psychological Medicine 55, e28
[1]

Apprenticeshiplearningviainverserein- forcement learning, in: Proceedings of the twenty-first international conference on Machine learning, p

Abbeel,P.,Ng,A.Y.,2004. Apprenticeshiplearningviainverserein- forcement learning, in: Proceedings of the twenty-first international conference on Machine learning, p. 1

2004
[2]

A prehospital triage system to detect traumatic intracranial hemorrhage using machine learning algorithms

Abe, D., Inaji, M., Hase, T., Takahashi, S., Sakai, R., Ayabe, F., Tanaka, Y., Otomo, Y., Maehara, T., 2022. A prehospital triage system to detect traumatic intracranial hemorrhage using machine learning algorithms. JAMA Network Open 5, e2216393. URL: https://jamanetworkopen.jamanetwork.com/article.aspx?doi= 10.1001/jamanetworkopen.2022.16393, doi:10.1001...

work page arXiv 2022
[3]

Machine learning in clinical decision making

Adlung, L., Cohen, Y., Mor, U., Elinav, E., 2021. Machine learning in clinical decision making. Med 2, 642–665

2021

Show all 281 references

[4]

Aguilera,A.,ArévaloAvalos,M.,Xu,J.,Chakraborty,B.,Figueroa, C., Garcia, F., Rosales, K., Hernandez-Ramos, R., Karr, C., Williams, J., et al., 2024. Effectiveness of a digital health inter- ventionleveragingreinforcementlearning:Resultsfromthediabetes and mental health adaptive...

2024
[5]

Optimized blockchain-basedhealthcareframeworkempoweredbymixedmulti- agent reinforcement learning

Al-Marridi, A.Z., Mohamed, A., Erbad, A., 2024. Optimized blockchain-basedhealthcareframeworkempoweredbymixedmulti- agent reinforcement learning. Journal of Network and Computer Applications 224, 103834

2024
[6]

A systematic review of trustworthy and explain- able artificial intelligence in healthcare: Assessment of qual- ity, bias risk, and data fusion

Albahri, A., Duhaim, A.M., Fadhel, M.A., Alnoor, A., Baqer, N.S., Alzubaidi, L., Albahri, O., Alamoodi, A., Bai, J., Salhi, A., Santamaría, J., Ouyang, C., Gupta, A., Gu, Y., Deveci, M., 2023. A systematic review of trustworthy and explain- able artificial intelligence in heal...

2023
[7]

Asynchronousadvantageactor-criticagentfor starcraft ii

Alghanem,B.,2018. Asynchronousadvantageactor-criticagentfor starcraft ii. arXiv preprint arXiv:1807.08217

2018 arXiv
[8]

Reinforcement learning in healthcare: optimizing treatmentstrategies,dynamicresourceallocation,andadaptiveclin- ical decision-making

Ali, H., 2022. Reinforcement learning in healthcare: optimizing treatmentstrategies,dynamicresourceallocation,andadaptiveclin- ical decision-making. Int J Comput Appl Technol Res 11, 88–104

2022
[9]

Explainabilityforartificialintelligenceinhealthcare: a multidisciplinary perspective

Amann,J.,Blasimme,A.,Vayena,E.,Frey,D.,Madai,V.I.,Consor- tium,P.,2020. Explainabilityforartificialintelligenceinhealthcare: a multidisciplinary perspective. BMC medical informatics and decision making 20, 1–9

2020
[10]

Amritanjali, Gupta, R., 2025. Federated learning for privacy pre- serving intelligent healthcare application to breast cancer detection, in: Proceedings of the 26th International Conference on Distributed Computing and Networking, pp. 302–306

2025
[11]

A comprehensive review on machine learning in healthcare industry: classification, restrictions, opportunities and challenges

An, Q., Rahman, S., Zhou, J., Kang, J.J., 2023. A comprehensive review on machine learning in healthcare industry: classification, restrictions, opportunities and challenges. Sensors 23, 4178

2023
[12]

On the challenges of using reinforcement learning in precision drug dosing:Delayandprolongednessofactioneffects,in:Proceedingsof the AAAI Conference on Artificial Intelligence, pp

Basu,S.,Legault,M.A.,Romero-Soriano,A.,Precup,D.,2023. On the challenges of using reinforcement learning in precision drug dosing:Delayandprolongednessofactioneffects,in:Proceedingsof the AAAI Conference on Artificial Intelligence, pp. 14102–14109

2023
[14]

Using artificial intelligence to learn optimal regimen plan for alzheimer’s disease

Bhattarai, K., Rajaganapathy, S., Das, T., Kim, Y., Chen, Y., Ini- tiative, A.D.N., Zong, N., 2023. Using artificial intelligence to learn optimal regimen plan for alzheimer’s disease. Journal of the American Medical Informatics Association 30, 1645–1656

2023
[15]

Reinvent 2.0: an ai tool for de novo drug design

Blaschke, T., Arús-Pous, J., Chen, H., Margreitter, C., Tyrchan, C., Engkvist, O., Papadopoulos, K., Patronov, A., 2020. Reinvent 2.0: an ai tool for de novo drug design. Journal of chemical information and modeling 60, 5918–5922

2020
[16]

Trans- forming health care with artificial intelligence: redefining medical documentation

Bongurala, A.R., Save, D., Virmani, A., Kashyap, R., 2024. Trans- forming health care with artificial intelligence: redefining medical documentation. Mayo Clinic Proceedings: Digital Health 2, 342– 347

2024
[17]

Open problems and fundamental limitations of reinforcement learning from human feedback

Casper,S.,Davies,X.,Shi,C.,Gilbert,T.K.,Scheurer,J.,Rando,J., Freedman,R., Korbak,T.,Lindner, D.,Freire,P., etal.,2023. Open problems and fundamental limitations of reinforcement learning from human feedback. arXiv preprint arXiv:2307.15217

2023 arXiv
[18]

Implementing machine learninginhealthcare—addressingethicalchallenges.NewEngland Journal of Medicine 378, 981–983

Char, D.S., Shah, N.H., Magnus, D., 2018. Implementing machine learninginhealthcare—addressingethicalchallenges.NewEngland Journal of Medicine 378, 981–983

2018
[19]

Rlhf de- ciphered: A critical analysis of reinforcement learning from human feedback for llms

Chaudhari, S., Aggarwal, P., Murahari, V., Rajpurohit, T., Kalyan, A., Narasimhan, K., Deshpande, A., da Silva, B.C., 2024. Rlhf de- ciphered: A critical analysis of reinforcement learning from human feedback for llms. arXiv preprint arXiv:2404.08555

2024 arXiv
[20]

Chen,L.,Paleja,R.,Gombolay,M.,2021.Learningfromsuboptimal demonstrationviaself-supervisedrewardregression,in:Conference on robot learning, PMLR. pp. 1262–1277

2021
[21]

Event-triggered drug dosage control strategy of immune systems via safe integral reinforcement learning

Chen, L., Zhang, Y., Yang, P., Jin, X., 2025. Event-triggered drug dosage control strategy of immune systems via safe integral reinforcement learning. European Journal of Control , 101201

2025
[22]

Reinforcement learning based diagnosis and prediction for covid-19 by optimizing a mixed cost function from ct images

Chen, S., Liu, M., Deng, P., Deng, J., Yuan, Y., Cheng, X., Liu, M., 2022a. Reinforcement learning based diagnosis and prediction for covid-19 by optimizing a mixed cost function from ct images. IEEEJournalofBiomedicalandHealthInformatics26,5344–5354. doi:10.1109/JBHI.2022.3180495

2022
[23]

A model-based hybrid soft actor-critic deep reinforcement learning algorithm for optimal ventilator settings

Chen, S., Qiu, X., Tan, X., Fang, Z., Jin, Y., 2022b. A model-based hybrid soft actor-critic deep reinforcement learning algorithm for optimal ventilator settings. Information Sciences 611, 47–64
[24]

Teleconsultation dynamic scheduling with a deep reinforcement learning approach

Chen, W., Li, J., 2024. Teleconsultation dynamic scheduling with a deep reinforcement learning approach. Artificial Intelligence in Medicine 149, 102806. doi:10.1016/j.artmed.2024.102806

2024
[25]

An online trajectory guidance framework via imitation learning and interactive feedback in robot-assisted surgery

Chen, Z., Fan, K., 2025. An online trajectory guidance framework via imitation learning and interactive feedback in robot-assisted surgery. NeuralNetworks185,107197. doi: 10.1016/j.neunet.2025. 107197

2025 doi
[26]

A survey on explainable deep reinforcement learning

Cheng, Z., Yu, J., Xing, X., 2025. A survey on explainable deep reinforcement learning. arXiv preprint arXiv:2502.06869

2025 arXiv
[27]

Deep reinforcement learning extracts the optimal sepsis treatment policy from treatment records

Choi, Y., Oh, S., Huh, J.W., Joo, H.T., Lee, H., You, W., Bae, C.m., Choi, J.H., Kim, K.J., 2024. Deep reinforcement learning extracts the optimal sepsis treatment policy from treatment records. Communications Medicine 4, 245

2024
[28]

Molgan: An implicit generative model for small molecular graphs

De Cao, N., Kipf, T., 2018. Molgan: An implicit generative model for small molecular graphs. arXiv preprint arXiv:1805.11973

2018 arXiv
[29]

Advances and applications in inverse reinforcement learning: a comprehensive review

Deshpande,S.,Walambe,R.,Kotecha,K.,Selvachandran,G.,Abra- ham, A., 2025. Advances and applications in inverse reinforcement learning: a comprehensive review. Neural Computing and Applica- tions , 1–53. D. Perera et al.:Preprint submitted to Elsevier Page 22 of 39 Reinforcement...

2025
[30]

Desman, J.M., Hong, Z.W., Sabounchi, M., Sawant, A.S., Gill, J., Costa,A.C.,Kumar,G.,Sharma,R.,Gupta,A.,McCarthy,P.,etal.,
[31]

Dikmen, M., Burns, C.M., 2016. Autonomous driving in the real world: Experiences with tesla autopilot and summon, in: Proceed- ingsofthe8thinternationalconferenceonautomotiveuserinterfaces and interactive vehicular applications, pp. 225–228

2016
[32]

Situation-dependentcausalinfluence-basedcooperativemulti-agent reinforcementlearning,in:ProceedingsoftheAAAIConferenceon Artificial Intelligence, pp

Du, X., Ye, Y., Zhang, P., Yang, Y., Chen, M., Wang, T., 2024. Situation-dependentcausalinfluence-basedcooperativemulti-agent reinforcementlearning,in:ProceedingsoftheAAAIConferenceon Artificial Intelligence, pp. 17362–17370

2024
[33]

Patient-specific sedation management via deep reinforcement learning

Eghbali, N., Alhanai, T., Ghassemi, M.M., 2021. Patient-specific sedation management via deep reinforcement learning. Frontiers in Digital Health 3, 608893

2021
[34]

Eghbali, N., Alhanai, T., Ghassemi, M.M., 2023. Reinforcement learning approach to sedation and delirium management in the intensive care unit, in: 2023 IEEE EMBS International Conference on Biomedical and Health Informatics (BHI), IEEE. pp. 1–5

2023
[35]

Learn from safe experience: Safe reinforcement learning for task automation of surgical robots

Fan, K., Chen, Z., Ferrigno, G., Momi, E.D., 2024. Learn from safe experience: Safe reinforcement learning for task automation of surgical robots. IEEE Transactions on Artificial Intelligence 5, 3374–3384. doi:10.1109/TAI.2024.3351797

2024
[36]

Sim-to-real surgical robot learningandautonomousplanningforinternaltissuepointsmanipu- lationusingreinforcementlearning

Fan, K., Ou, Y., Tavakoli, M., 2023. Sim-to-real surgical robot learningandautonomousplanningforinternaltissuepointsmanipu- lationusingreinforcementlearning. IEEERoboticsandAutomation Letters 8, 2502–2510. doi:10.1109/LRA.2023.3254860

2023
[37]

Provably robust federated reinforcement learning

Fang, M., Wang, X., Gong, N.Z., 2025. Provably robust federated reinforcement learning. arXiv preprint arXiv:2502.08123

2025 arXiv
[38]

Towards trustworthy automatic diagnosis systems by emulating doctors’ reasoning with deep reinforcement learning, in:AdvancesinNeuralInformationProcessingSystems,pp.24502– 24515

FansiTchango,A.,Goel,R.,Martel,J.,Wen,Z.,MarceauCaron,G., Ghosn, J., 2022. Towards trustworthy automatic diagnosis systems by emulating doctors’ reasoning with deep reinforcement learning, in:AdvancesinNeuralInformationProcessingSystems,pp.24502– 24515

2022
[39]

More robust doubly robust off-policy evaluation, in: International Conference on Machine Learning, PMLR

Farajtabar, M., Chow, Y., Ghavamzadeh, M., 2018. More robust doubly robust off-policy evaluation, in: International Conference on Machine Learning, PMLR. pp. 1447–1456

2018
[40]

Semi-markov offline reinforce- ment learning for healthcare, in: Conference on Health, Inference, and Learning, PMLR

Fatemi, M., Wu, M., Petch, J., Nelson, W., Connolly, S.J., Benz, A., Carnicelli, A., Ghassemi, M., 2022. Semi-markov offline reinforce- ment learning for healthcare, in: Conference on Health, Inference, and Learning, PMLR. pp. 119–137

2022
[41]

arXiv preprint arXiv:2502.04408

Ferrante,M.,Carosi,A.,Angelillo,R.M.D.,Toschi,N.,2025.Trans- forming multimodal models into action models for radiotherapy. arXiv preprint arXiv:2502.04408

2025 arXiv
[42]

Off-policy deep rein- forcementlearningwithoutexploration,in:Internationalconference on machine learning, PMLR

Fujimoto, S., Meger, D., Precup, D., 2019. Off-policy deep rein- forcementlearningwithoutexploration,in:Internationalconference on machine learning, PMLR. pp. 2052–2062

2019
[43]

Popcorn:Partially observed prediction constrained reinforcement learning

Futoma,J.,Hughes,M.C.,Doshi-Velez,F.,2020. Popcorn:Partially observed prediction constrained reinforcement learning. arXiv preprint arXiv:2001.04032

2020 arXiv
[44]

Learningtotreatsepsiswithmulti-outputgaus- sian process deep recurrent q-networks

Futoma,J.,Lin,A.,Sendak,M.,Bedoya,A.,Clement,M.,O’Brien, C.,Heller,K.,2018. Learningtotreatsepsiswithmulti-outputgaus- sian process deep recurrent q-networks. URL:https://openreview. net/forum?id=SyxCqGbRZ

2018
[45]

Ai-assisted warfarin dose optimisation with curate

Gan, T.R.X., Tan, L.W., Egermark, M., Truong, A.T., Kumar, K., Tan,S.B.,Tang,S.,Blasiak,A.,Goh,B.C.,Ngiam,K.Y.,etal.,2025. Ai-assisted warfarin dose optimisation with curate. ai for clinical impact:Retrospectivedataanalysis. Bioengineering&Translational Medicine , e10757

2025
[46]

Causal inference from cross-sectional earth system data with geographical convergent cross mapping

Gao, B., Yang, J., Chen, Z., Sugihara, G., Li, M., Stein, A., Kwan, M.P., Wang, J., 2023a. Causal inference from cross-sectional earth system data with geographical convergent cross mapping. nature communications 14, 5875
[47]

Gao, J., Lu, Y., Ashrafi, N., Domingo, I., Alaei, K., Pishgar, M.,
[48]

Off-policy evaluation for human feedback

Gao, Q., Gao, G., Dong, J., Tarokh, V., Chi, M., Pajic, M., 2023b. Off-policy evaluation for human feedback. Proceedings of NeurIPS URL: https://arxiv.org/abs/2310.07123

arXiv
[49]

Personalized decision making for coronary artery disease treatmentusingofflinereinforcementlearning

Ghasemi,P.,Greenberg,M.,Southern,D.A.,Li,B.,White,J.A.,Lee, J., 2025a. Personalized decision making for coronary artery disease treatmentusingofflinereinforcementlearning. npjDigitalMedicine 8, 99. doi:10.1038/s41746-025-00789-4

doi
[50]

Personalizeddecisionmakingforcoronaryarterydisease treatmentusingofflinereinforcementlearning

Ghasemi,P.,Greenberg,M.,Southern,D.A.,Li,B.,White,J.A.,Lee, J.,2025b. Personalizeddecisionmakingforcoronaryarterydisease treatmentusingofflinereinforcementlearning. npjDigitalMedicine 8, 99
[51]

The false hope of current approaches to explainable artificial intelligence in health care

Ghassemi, M., Oakden-Rayner, L., Beam, A.L., 2021. The false hope of current approaches to explainable artificial intelligence in health care. The Lancet Digital Health 3, e745–e750

2021
[52]

Medications for analgesia and sedation in the intensive care unit: an overview

Gommers, D., Bakker, J., 2008. Medications for analgesia and sedation in the intensive care unit: an overview. Critical Care 12, 1–5

2008
[53]

Guidelines for reinforcement learning in healthcare

Gottesman, O., Johansson, F., Komorowski, M., Faisal, A., Sontag, D., Doshi-Velez, F., Celi, L.A., 2019. Guidelines for reinforcement learning in healthcare. Nature medicine 25, 16–18

2019
[54]

Gu,X.,Deligianni,F.,Han,J.,Liu,X.,Chen,W.,Yang,G.Z.,Lo,B.,
[55]

Machine learning with optimization to create medicine intake schedules for parkin- son’s disease patients

Gutowski, T., Antkiewicz, R., Szlufik, S., 2023. Machine learning with optimization to create medicine intake schedules for parkin- son’s disease patients. PLOS ONE 18, e0293123

2023
[56]

Extubation after breathing trials with automatictubecompensation,t-tube,orpressuresupportventilation

Haberthür, C., et al., 2002. Extubation after breathing trials with automatictubecompensation,t-tube,orpressuresupportventilation. Acta Anaesthesiologica Scandinavica 46, 973–979

2002
[57]

Sim2real rope cutting with a surgical robot using vision-based reinforcement learning

Haiderbhai, M., Gondokaryono, R., Wu, A., Kahrs, L.A., 2024. Sim2real rope cutting with a surgical robot using vision-based reinforcement learning. IEEE Transactions on Automation Science and Engineering

2024
[58]

Evaluating the diagnostic performance of symptom checkers: Clinical vignette study

Hammoud, M., Douglas, S., Darmach, M., Alawneh, S., Sanyal, S., Kanbour, Y., et al., 2024. Evaluating the diagnostic performance of symptom checkers: Clinical vignette study. JMIR AI 3, e46875

2024
[59]

Han, S., Choi, W., 2025. Development of a large language model- based multi-agent clinical decision support system for korean triage andacuityscale(ktas)-basedtriageandtreatmentplanninginemer- gency departments. advances in artificial intelligence and machine learning. 2025; 5...

2025
[60]

Longi- tudinalchangesindiagnosticaccuracyofadifferentialdiagnosislist developed by an ai-based symptom checker: retrospective observa- tional study

Harada, Y., Sakamoto, T., Sugimoto, S., Shimizu, T., 2024. Longi- tudinalchangesindiagnosticaccuracyofadifferentialdiagnosislist developed by an ai-based symptom checker: retrospective observa- tional study. JMIR Formative Research 8, e53985

2024
[61]

Investigatingthe potential of reinforcement learning and deep learning in improving alzheimer’s disease classification

Hatami,M.,Yaghmaee,F.,Ebrahimpour,R.,2024. Investigatingthe potential of reinforcement learning and deep learning in improving alzheimer’s disease classification. Neurocomputing 597, 128119

2024
[62]

Deep recurrent q-learning for partially observable mdps., in: AAAI fall symposia, p

Hausknecht, M.J., Stone, P., 2015. Deep recurrent q-learning for partially observable mdps., in: AAAI fall symposia, p. 141

2015
[63]

IEEE Wireless Communications 28, 6–11

Hayyolalam,V.,Aloqaily,M.,Özkasap,O.,Guizani,M.,2021.Edge intelligence for empowering iot-based healthcare systems. IEEE Wireless Communications 28, 6–11. doi:10.1109/MWC.001.2000345

2021 doi
[64]

Understanding the patient perspective of epilepsy treatment through text mining of online patient support groups

He, K., Hong, N., Lapalme-Remis, S., Lan, Y., Huang, M., Li, C., Yao, L., 2019. Understanding the patient perspective of epilepsy treatment through text mining of online patient support groups. Epilepsy & Behavior 94, 65–71

2019
[65]

A survey of large language models for healthcare: fromdata,technology,andapplicationstoaccountabilityandethics

He, K., Mao, R., Lin, Q., Ruan, Y., Lan, X., Feng, M., Cambria, E., 2025. A survey of large language models for healthcare: fromdata,technology,andapplicationstoaccountabilityandethics. Information Fusion 118, 102963

2025
[66]

A blockchain-based scheme for secure data offloading in healthcare with deep reinforcement learning

He,Q.,Feng,Z.,Fang,H.,Wang,X.,Zhao,L.,Yao,Y.,Yu,K.,2023. A blockchain-based scheme for secure data offloading in healthcare with deep reinforcement learning. IEEE/ACM Transactions on Networking 32, 65–80

2023
[67]

Model based reinforcement learning for personalized heparin dosing

He, Q., Mintz, Y., 2023. Model based reinforcement learning for personalized heparin dosing. arXiv preprint arXiv:2304.10000 . D. Perera et al.:Preprint submitted to Elsevier Page 23 of 39 Reinforcement Learning as the Defining Leap in Healthcare AI

2023 arXiv
[68]

Heess, N., Wayne, G., Silver, D., Lillicrap, T., Erez, T., Tassa, Y.,
[69]

Long short-term memory

Hochreiter, S., Schmidhuber, J., 1997. Long short-term memory. Neural computation 9, 1735–1780

1997
[70]

Vime: Variational information maximizing ex- ploration

Houthooft, R., Chen, X., Duan, Y., Schulman, J., De Turck, F., Abbeel, P., 2016. Vime: Variational information maximizing ex- ploration. Advances in neural information processing systems 29

2016
[71]

Dynamic programming and markov processes

Howard, R., 1960. Dynamic programming and markov processes

1960
[72]

Reinforcement learning for sepsis treatment: A continuous action space solution

Huang, Y., Cao, R., Rahmani, A., 2022. Reinforcement learning for sepsis treatment: A continuous action space solution. Proceedings of Machine Learning Research , 1–17

2022
[73]

Deep variational reinforcement learning for pomdps, in: International conference on machine learning, PMLR

Igl, M., Zintgraf, L., Le, T.A., Wood, F., Whiteson, S., 2018. Deep variational reinforcement learning for pomdps, in: International conference on machine learning, PMLR. pp. 2117–2126

2018
[74]

Personal- ized insulin dosing using reinforcement learning for high-fat meals and aerobic exercises in type 1 diabetes: a proof-of-concept trial

Jafar, A., Kobayati, A., Tsoukas, M.A., Haidar, A., 2024. Personal- ized insulin dosing using reinforcement learning for high-fat meals and aerobic exercises in type 1 diabetes: a proof-of-concept trial. Nature Communications 15, 6585

2024
[75]

Irl for restless multi-armed bandits with applications in maternal and child health, in: PRICAI 2024: Trends in Artificial Intelligence, pp

Jain, G., Varakantham, P., Xu, H., Taneja, A., Doshi, P., Tambe, M., 2024. Irl for restless multi-armed bandits with applications in maternal and child health, in: PRICAI 2024: Trends in Artificial Intelligence, pp. 165–178

2024
[76]

Jaiteh, M., Phalane, E., Shiferaw, Y.A., Phaswana-Mafuya, R.N., et al., 2025. The application of machine learning algorithms to predict hiv testing in repeated adult population–based surveys in southafrica:Protocolforamultiwavecross-sectionalanalysis.JMIR Research Protocols 14, e59916

2025
[77]

Basal-bolus advisor for type 1 diabetes (t1d) patients using multi-agent reinforcement learning (rl) methodology

Jaloli, M., Cescon, M., 2024. Basal-bolus advisor for type 1 diabetes (t1d) patients using multi-agent reinforcement learning (rl) methodology. Control Engineering Practice 142, 105762

2024
[78]

A primer on reinforcement learning in medicine for clinicians

Jayaraman,P.,Desman,J.,Sabounchi,M.,Nadkarni,G.N.,Sakhuja, A., 2024. A primer on reinforcement learning in medicine for clinicians. npj Digital Medicine 7

2024
[79]

Pku-saferlhf: Towards multi-level safety alignment for llms with human preference

Ji, J., Hong, D., Zhang, B., Chen, B., Dai, J., Zheng, B., Qiu, T., Li, B., Yang, Y., 2024. Pku-saferlhf: Towards multi-level safety alignment for llms with human preference. arXiv preprint arXiv:2406.15513

2024 arXiv
[80]

Safe reinforcement learningforsepsistreatment,in:2020IEEEInternationalconference on healthcare informatics (ICHI), IEEE

Jia, Y., Burden, J., Lawton, T., Habli, I., 2020. Safe reinforcement learningforsepsistreatment,in:2020IEEEInternationalconference on healthcare informatics (ICHI), IEEE. pp. 1–7

2020
[81]

Doubly robust off-policy value evaluation for reinforcement learning, in: International conference on machine learning, PMLR

Jiang, N., Li, L., 2016. Doubly robust off-policy value evaluation for reinforcement learning, in: International conference on machine learning, PMLR. pp. 652–661

2016
[82]

Opti- maltreatmentstrategiesforcriticalpatientswithdeepreinforcement learning

Job,S.,Tao,X.,Li,L.,Xie,H.,Cai,T.,Yong,J.,Li,Q.,2024. Opti- maltreatmentstrategiesforcriticalpatientswithdeepreinforcement learning. ACMTransactionsonIntelligentSystemsandTechnology 15, Article 36

2024
[83]

The malmoplatformforartificialintelligenceexperimentation.,in:Ijcai, pp

Johnson, M., Hofmann, K., Hutton, T., Bignell, D., 2016. The malmoplatformforartificialintelligenceexperimentation.,in:Ijcai, pp. 4246–4247

2016
[84]

Reinforcement learn- ing: A survey

Kaelbling, L., Littman, M., Moore, A., 1996. Reinforcement learn- ing: A survey. Journal of artificial intelligence research 4, 237–285

1996
[85]

Context-aware symp- tomcheckingfordiseasediagnosisusinghierarchicalreinforcement learning

Kao, H.C., Tang, K.F., Chang, E., 2018. Context-aware symp- tomcheckingfordiseasediagnosisusinghierarchicalreinforcement learning. Proceedings of the AAAI Conference on Artificial In- telligence 32. URL:https://ojs.aaai.org/index.php/AAAI/article/ view/11902, doi:10.1609/aaai....

2018 doi
[86]

Elrl-md: a deep learning approach for myocarditis diagnosis using cardiac magnetic resonance images with ensemble and reinforcement learning inte- gration

Kasmaee, A.M.M., Ataei, A., Moravvej, S.V., Alizadehsani, R., Gorriz, J.M., Zhang, Y.D., Acharya, U.R., 2025. Elrl-md: a deep learning approach for myocarditis diagnosis using cardiac magnetic resonance images with ensemble and reinforcement learning inte- gration. Physiologic...

2025
[87]

Kendall, A., Hawke, J., Janz, D., Mazur, P., Reda, D., Allen, J.M., Lam,V.D.,Bewley,A.,Shah,A.,2019.Learningtodriveinaday,in: 2019 international conference on robotics and automation (ICRA), IEEE. pp. 8248–8254

2019
[88]

Reinforcement learning for clinical applica- tions

Khezeli, K., Siegel, S., Shickel, B., Ozrazgat-Baslanti, T., Bihorac, A., Rashidi, P., 2023. Reinforcement learning for clinical applica- tions. Clinical Journal of the American Society of Nephrology 18, 521–523

2023
[89]

An empirical study of representation learning for rein- forcement learning in healthcare

Killian, T.W., Zhang, H., Subramanian, J., Fatemi, M., Ghassemi, M., 2020. An empirical study of representation learning for rein- forcement learning in healthcare. arXiv preprint arXiv:2011.11235

2020 arXiv
[90]

Physiological Measurement 42, 025006

Kim,D.,Chae,J.,Oh,Y.,Lee,J.,Kim,I.Y.,2021.Automatedremote decision-makingalgorithmasaprimarytriagesystemusingmachine learning techniques. Physiological Measurement 42, 025006

2021
[91]

Building a personalized messaging system for health intervention inunderprivilegedregionsusingreinforcementlearning.,in:IJCAI, pp

Kinsey, S.E., Wolf, J., Saligram, N., Ramesan, V., Walavalkar, M., Jaswal, N., Ramalingam, S., Sinha, A., Nguyen, T.H., 2023. Building a personalized messaging system for health intervention inunderprivilegedregionsusingreinforcementlearning.,in:IJCAI, pp. 6022–6030

2023
[92]

The artificial intelligence clinician learns optimal treatment strategies for sepsis in intensive care

Komorowski, M., et al., 2018. The artificial intelligence clinician learns optimal treatment strategies for sepsis in intensive care. Na- ture Medicine 24, 1716–1720

2018
[93]

Towards safe mechanical ventilation treatment using deep offline reinforcement learning, in: Proceedings of the AAAI Conference on Artificial Intelligence, pp

Kondrup, F., Jiralerspong, T., Lau, E., de Lara, N., Shkrob, J., Tran, M.D., Precup, D., Basu, S., 2023. Towards safe mechanical ventilation treatment using deep offline reinforcement learning, in: Proceedings of the AAAI Conference on Artificial Intelligence, pp. 15696–15702

2023
[94]

Designanefficientvarmalstmgru model for identification of deep-fake images via dynamic window- based spatio-temporal analysis

Kosarkar,U.,Sakarkar,G.,2025. Designanefficientvarmalstmgru model for identification of deep-fake images via dynamic window- based spatio-temporal analysis. Multimedia Tools and Applications 84, 3841–3857

2025
[95]

Hierarchical deep reinforcement learning: Integrating temporal ab- straction and intrinsic motivation

Kulkarni, T.D., Narasimhan, K., Saeedi, A., Tenenbaum, J., 2016. Hierarchical deep reinforcement learning: Integrating temporal ab- straction and intrinsic motivation. Advances in neural information processing systems 29

2016
[96]

Conservative q-learning for offline reinforcement learning

Kumar, A., Zhou, A., Tucker, G., Levine, S., 2020. Conservative q-learning for offline reinforcement learning. Advances in neural information processing systems 33, 1179–1191

2020
[97]

Artificial intel- ligence in disease diagnosis: a systematic literature review, synthe- sizing framework and future research agenda

Kumar, Y., Koul, A., Singla, R., Ijaz, M.F., 2023. Artificial intel- ligence in disease diagnosis: a systematic literature review, synthe- sizing framework and future research agenda. Journal of ambient intelligence and humanized computing 14, 8459–8486

2023
[98]

Mobile-fog-cloud assisted deep reinforcement learning and blockchain-enable iomt system for healthcare workflows

Lakhan, A., Mohammed, M.A., Kozlov, S., Rodrigues, J.J., 2024a. Mobile-fog-cloud assisted deep reinforcement learning and blockchain-enable iomt system for healthcare workflows. Transac- tions on Emerging Telecommunications Technologies 35, e4363
[99]

Fiber-optics iot healthcare system based on deep reinforcement learning combi- natorial constraint scheduling for hybrid telemedicine applications

Lakhan, A., Nedoma, J., Mohammed, M.A., Deveci, M., Fajkus, M., Marhoon, H.A., Memon, S., Martinek, R., 2024b. Fiber-optics iot healthcare system based on deep reinforcement learning combi- natorial constraint scheduling for hybrid telemedicine applications. Computers in Biolo...
[100]

Data-drivenhospitalsstaffandresourcesalloca- tion using agent-based simulation and deep reinforcement learning

Lazebnik,T.,2023. Data-drivenhospitalsstaffandresourcesalloca- tion using agent-based simulation and deep reinforcement learning. Engineering Applications of Artificial Intelligence 126, 106783

2023
[101]

Artificialintelligence conversational agents in mental health: Patients see potential, but preferhumansintheloop.FrontiersinPsychiatrydoi: 10.3389/fpsyt

Lee,H.S.,Wright,C.,Ferranto,J.,etal.,2025. Artificialintelligence conversational agents in mental health: Patients see potential, but preferhumansintheloop.FrontiersinPsychiatrydoi: 10.3389/fpsyt. 2024.1505024

2025
[102]

Reinforcement learning model for optimizing dexmedetomidinedosingtopreventdeliriumincriticallyillpatients

Lee, H.Y., Chung, S., Hyeon, D., Yang, H.L., Lee, H.C., Ryu, H.G., Lee, H., 2024. Reinforcement learning model for optimizing dexmedetomidinedosingtopreventdeliriumincriticallyillpatients. npj Digital Medicine 7, 325

2024
[103]

End-to-end training of deep visuomotor policies

Levine, S., Finn, C., Darrell, T., Abbeel, P., 2016. End-to-end training of deep visuomotor policies. Journal of Machine Learning Research 17, 1–40

2016
[104]

The actor search tree critic (astc) for off-policy pomdp learning in medical decision making

Li, L., Komorowski, M., Faisal, A.A., 2018. The actor search tree critic (astc) for off-policy pomdp learning in medical decision making. arXiv preprint arXiv:1805.11548

2018 arXiv
[105]

Electronic health records-based reinforcement learning for treatment optimizing

Li, T., Wang, Z., Lu, W., Zhang, Q., Li, D., 2022. Electronic health records-based reinforcement learning for treatment optimizing. In- formation Systems 104, 101878. D. Perera et al.:Preprint submitted to Elsevier Page 24 of 39 Reinforcement Learning as the Defining Leap in H...

2022
[106]

The predictive accuracy of machine learning for the risk of death in hiv patients: a systematic review and meta-analysis

Li, Y., Feng, Y., He, Q., Ni, Z., Hu, X., Feng, X., Ni, M., 2024. The predictive accuracy of machine learning for the risk of death in hiv patients: a systematic review and meta-analysis. BMC Infectious Diseases 24, 474

2024
[107]

Safe reinforcement learning via episodic control

Li, Z., Zhu, D., Grossklags, J., 2025. Safe reinforcement learning via episodic control. IEEE Access

2025
[108]

Episodicreinforcementlearning withexpandedstate-rewardspace

Liang,D.,Zhang,Y.,Liu,Y.,2024. Episodicreinforcementlearning withexpandedstate-rewardspace. arXivpreprintarXiv:2401.10516

2024 arXiv
[109]

Reinforcement learning using neural net- worksinestimatinganoptimaldynamictreatmentregimeinpatients with sepsis

Liang, W., Jia, J., 2025. Reinforcement learning using neural net- worksinestimatinganoptimaldynamictreatmentregimeinpatients with sepsis. Computer Methods and Programs in Biomedicine , 108754

2025
[110]

Development and validation of heparin dosing policies using an offline reinforcement learning algorithm

Lim, Y., Park, I., Lee, S., 2024. Development and validation of heparin dosing policies using an offline reinforcement learning algorithm. arXiv preprint arXiv:2409.15753

2024
[111]

Has multimodal learning delivered universal intelligence in healthcare? a comprehensive survey

Lin, Q., Zhu, Y., Mei, X., Huang, L., Ma, J., He, K., Peng, Z., Cambria, E., Feng, M., 2025. Has multimodal learning delivered universal intelligence in healthcare? a comprehensive survey. Infor- mation Fusion 116, 102795

2025
[112]

Deepseek-v2: A strong, economical,andefficientmixture-of-expertslanguagemodel

Liu,A.,Feng,B.,Wang,B.,Wang,B.,Liu,B.,Zhao,C.,Dengr,C., Ruan, C., Dai, D., Guo, D., et al., 2024a. Deepseek-v2: A strong, economical,andefficientmixture-of-expertslanguagemodel. arXiv preprint arXiv:2405.04434

arXiv
[113]

Deep reinforcement learning for personalized treatment recommendation

Liu, M., Shen, X., Pan, W., 2022. Deep reinforcement learning for personalized treatment recommendation. Statistics in medicine 41, 4034–4056

2022
[114]

Reinforcementlearningforclinicaldecisionsupportincriticalcare: comprehensive review

Liu,S.,See,K.C.,Ngiam,K.Y.,Celi,L.A.,Sun,X.,Feng,M.,2020. Reinforcementlearningforclinicaldecisionsupportincriticalcare: comprehensive review. Journal of medical Internet research 22, e18477

2020
[116]

Lu, M., Shahn, Z., Sow, D., Doshi-Velez, F., Li-wei, H.L., 2021. Is deep reinforcement learning ready for practical applications in healthcare? a sensitivity analysis of duel-ddqn for hemodynamic management in sepsis patients, in: AMIA Annual Symposium Pro- ceedings, p. 773

2021
[117]

Efficient reinforcement learning with prior causal knowledge, in: Conference on Causal Learning and Reasoning, PMLR

Lu, Y., Meisami, A., Tewari, A., 2022. Efficient reinforcement learning with prior causal knowledge, in: Conference on Causal Learning and Reasoning, PMLR. pp. 526–541

2022
[118]

Serl: A software suite for sample-efficient robotic reinforcement learning, in: 2024 IEEE In- ternational Conference on Robotics and Automation (ICRA), IEEE

Luo, J., Hu, Z., Xu, C., Tan, Y.L., Berg, J., Sharma, A., Schaal, S., Finn, C., Gupta, A., Levine, S., 2024a. Serl: A software suite for sample-efficient robotic reinforcement learning, in: 2024 IEEE In- ternational Conference on Robotics and Automation (ICRA), IEEE. pp. 16961–16969

2024
[119]

Position: re- inforcement learning in dynamic treatment regimes needs critical reexamination

Luo, Z., Pan, Y., Watkinson, P., Zhu, T., 2024b. Position: re- inforcement learning in dynamic treatment regimes needs critical reexamination. Journal of Machine Learning Research
[122]

Määttä,J.,Lindell,R.,Hayward,N.,Martikainen,S.,Honkanen,K., Inkala, M., Hirvonen, P., Martikainen, T.J., et al., 2023. Diagnostic performance, triage safety, and usability of a clinical decision sup- port system within a university hospital emergency department: al- gorithmper...

2023
[123]

Artificialintelligenceinpublichealth:thepotentialofepidemicearly warning systems

MacIntyre, C.R., Chen, X., Kunasekaran, M., Quigley, A., Lim, S., Stone, H., Paik, H.y., Yao, L., Heslop, D., Wei, W., et al., 2023. Artificialintelligenceinpublichealth:thepotentialofepidemicearly warning systems. Journal of International Medical Research 51, 03000605231159335

2023
[124]

Proceedings of NeurIPS URL:https://arxiv.org/abs/2309.08560

MasonHargrave,AlexSpaeth,L.G.,2024.Epicare:Areinforcement learning benchmark for dynamic treatment regimes. Proceedings of NeurIPS URL:https://arxiv.org/abs/2309.08560

2024 arXiv
[125]

Inform: Mitigating reward hacking in rlhf via information-theoretic reward modeling

Miao, Y., Zhang, S., Ding, L., Bao, R., Zhang, L., Tao, D., 2025. Inform: Mitigating reward hacking in rlhf via information-theoretic reward modeling. Advances in Neural Information Processing Systems 37, 134387–134429

2025
[126]

Asynchronousmethodsfordeep reinforcement learning, in: International conference on machine learning, PmLR

Mnih,V.,Badia,A.P.,Mirza,M.,Graves,A.,Lillicrap,T.,Harley,T., Silver,D.,Kavukcuoglu,K.,2016. Asynchronousmethodsfordeep reinforcement learning, in: International conference on machine learning, PmLR. pp. 1928–1937

2016
[127]

Variational information maximisation for intrinsically motivated reinforcement learning

Mohamed, S., Jimenez Rezende, D., 2015. Variational information maximisation for intrinsically motivated reinforcement learning. Advances in neural information processing systems 28

2015
[128]

Mohanty, S.D., Lekan, D., McCoy, T.P., Jenkins, M., Manda, P.,
[129]

Human-in- the-loop machine learning: A state of the art

Mosqueira-Rey, E., Hernández-Pereira, E., et al., 2023. Human-in- the-loop machine learning: A state of the art. Artificial Intelligence Review 56, 3005–3054. doi:10.1007/s10462-022-10246-w

2023 doi
[130]

Safe and efficient off-policy reinforcement learning

Munos, R., Stepleton, T., Harutyunyan, A., Bellemare, M., 2016. Safe and efficient off-policy reinforcement learning. Advances in neural information processing systems 29

2016
[131]

Efficacy of robot-assisted double-flap techniques with refinements to mini- mizeanastomosis-relatedcomplicationsafterproximalgastrectomy

Nakao, E., Igeta, M., Murakami, M., Kohno, S., Hojo, Y., Naka- mura, T., Kurahashi, Y., Ishida, Y., Shinohara, H., 2025. Efficacy of robot-assisted double-flap techniques with refinements to mini- mizeanastomosis-relatedcomplicationsafterproximalgastrectomy. Surgery Today doi:...

2025 doi
[132]

Reinforcement learning in ophthalmology: potentialapplicationsandchallengestoimplementation

Nath, S., Korot, E., Fu, D.J., Zhang, G., Mishra, K., Lee, A.Y., Keane, P.A., 2022. Reinforcement learning in ophthalmology: potentialapplicationsandchallengestoimplementation. TheLancet Digital Health 4, e692–e697. doi:10.1016/S2589-7500(22)00188-5

2022 doi
[134]

Nemati, S., Ghassemi, M.M., Clifford, G.D., 2016. Optimal med- ication dosing from suboptimal clinical examples: A deep rein- forcement learning approach, in: 2016 38th annual international conferenceoftheIEEEengineeringinmedicineandbiologysociety (EMBC), IEEE. pp. 2978–2981

2016
[135]

Algorithms for inverse reinforce- ment learning., in: Icml, p

Ng, A.Y., Russell, S., et al., 2000. Algorithms for inverse reinforce- ment learning., in: Icml, p. 2

2000
[136]

Reinforcement-learningoptimalcontrolfortype-1diabetes,in:2018 IEEE EMBS International Conference on Biomedical & Health Informatics (BHI), IEEE

Ngo, P.D., Wei, S., Holubová, A., Muzik, J., Godtliebsen, F., 2018. Reinforcement-learningoptimalcontrolfortype-1diabetes,in:2018 IEEE EMBS International Conference on Biomedical & Health Informatics (BHI), IEEE. pp. 333–336

2018
[137]

In value-based deep reinforcement learning, a pruned network is a good network

Obando-Ceron, J., Courville, A., Castro, P.S., 2024. In value-based deep reinforcement learning, a pruned network is a good network. arXiv preprint arXiv:2402.12479

2024 arXiv
[138]

Oroojeni Mohammad Javad, M., Agboola, S., Jethwani, K., Zeid, I., Kamarthi, S., 2015. Reinforcement learning algorithm for blood glucose control in diabetic patients, in: ASME International Me- chanical Engineering Congress and Exposition, American Society of Mechanical Engine...

2015
[139]

Sim-to-real surgical robot learning and autonomous planning for internal tissue points manipulation using reinforcement learning

Ou, Y., Tavakoli, M., 2023. Sim-to-real surgical robot learning and autonomous planning for internal tissue points manipulation using reinforcement learning. IEEE Robotics and Automation Letters 8, 2502–2509

2023
[140]

A survey on transfer learning

Pan, S., Yang, Q., 2009. A survey on transfer learning. IEEE Transactions on knowledge and data engineering 22, 1345–1359

2009
[141]

Ai in medical imaging informatics: current challenges and future directions

Panayides, A.S., Amini, A., Filipovic, N.D., Sharma, A., Tsaftaris, S.A., Young, A., Pattichis, C.S., 2020. Ai in medical imaging informatics: current challenges and future directions. IEEE journal of biomedical and health informatics 24, 1837–1857. D. Perera et al.:Preprint s...

2020
[142]

Parbhoo, S., Bogojeska, J., Zazzi, M., Roth, V., Doshi-Velez, F.,
[143]

Areviewoffederated learningmethodsinheterogeneousscenarios

Pei,J.,Liu,W.,Li,J.,Wang,L.,Liu,C.,2024. Areviewoffederated learningmethodsinheterogeneousscenarios. IEEETransactionson Consumer Electronics

2024
[144]

Development and validation of a reinforcement learning algorithm to dynamically optimize mechanical ventilation in critical care

Peine, A., Hallawa, A., Bickenbach, J., Dartmann, G., Fazlic, L.B., Schmeink, A., Ascheid, G., Thiemermann, C., Schuppert, A., Kin- dle, R., et al., 2021. Development and validation of a reinforcement learning algorithm to dynamically optimize mechanical ventilation in critica...

2021
[145]

Improving sepsis treatment strategies by combining deep and kernel-based reinforcementlearning,in:AMIAAnnualSymposiumProceedings, p

Peng, X., Ding, Y., Wihl, D., Gottesman, O., Komorowski, M., Li- wei, H.L., Ross, A., Faisal, A., Doshi-Velez, F., 2018. Improving sepsis treatment strategies by combining deep and kernel-based reinforcementlearning,in:AMIAAnnualSymposiumProceedings, p. 887

2018
[146]

Smart imitator: Learningfromimperfectclinicaldecisions

Perera, D., Liu, S., See, K.C., Feng, M., 2025. Smart imitator: Learningfromimperfectclinicaldecisions. JournaloftheAmerican Medical Informatics Association , ocae320

2025
[147]

Optimizingwarfarindosingforpatientswithatrialfibrillationusing machine learning

Petch, J., Nelson, W., Wu, M., Ghassemi, M., Benz, A., Fatemi, M., Di, S., Carnicelli, A., Granger, C., Giugliano, R., et al., 2024. Optimizingwarfarindosingforpatientswithatrialfibrillationusing machine learning. Scientific reports 14, 4516

2024
[148]

Predicting cardiac infarctions with reinforcement algorithms through wavelet transform applications in healthcare

Pradeep, M., Bhattacharyya, D., Anguraj, D.K., Kim, T.h., Ogudo, K.A., Mohammed, M., 2025. Predicting cardiac infarctions with reinforcement algorithms through wavelet transform applications in healthcare. Information Sciences 691, 121513

2025
[149]

A reinforcement learning approach to weaning of mechanical ventilation in intensive care units

Prasad, N., et al., 2017. A reinforcement learning approach to weaning of mechanical ventilation in intensive care units. arXiv preprint arXiv:1704.06300

2017 arXiv
[150]

Guaranteeing out-of-distribution detection in deep rl via transition estimation, in: Proceedings of the AAAI Conference on Artificial Intelligence, pp

Prashant,M.,Easwaran,A.,Das,S.,Yuhas,M.,2025. Guaranteeing out-of-distribution detection in deep rl via transition estimation, in: Proceedings of the AAAI Conference on Artificial Intelligence, pp. 12452–12460

2025
[151]

Purwati, I.I., Purwanto, H., Anwar, F., Arianto, F., et al., 2025. Analysis of dose distribution and radiation source effectiveness in brachytherapy for right lung cancer using ornl-mird 1996 phantom with mcnp 6.2 software, in: Journal of Physics: Conference Series, IOP Publis...

2025
[152]

A latent batch-constrained deep reinforcement learning approach for preci- sion dosing clinical decision support

Qiu, X., Tan, X., Li, Q., Chen, S., Ru, Y., Jin, Y., 2022. A latent batch-constrained deep reinforcement learning approach for preci- sion dosing clinical decision support. Knowledge-based systems 237, 107689

2022
[153]

Continuous state-space models for optimal sepsis treatment-a deep reinforcement learning approach

Raghu, A., et al., 2017. Continuous state-space models for optimal sepsis treatment-a deep reinforcement learning approach. arXiv preprint arXiv:1705.08422

2017 arXiv
[154]

Scalable and accurate deep learning with electronic health records

Rajkomar, A., Oren, E., Chen, K., Dai, A.M., Hajaj, N., Hardt, M., Liu, P.J., Liu, X., Marcus, J., Sun, M., et al., 2018. Scalable and accurate deep learning with electronic health records. NPJ digital medicine 1, 18

2018
[155]

Bayesian inverse reinforcement learning., in: IJCAI, pp

Ramachandran, D., Amir, E., 2007. Bayesian inverse reinforcement learning., in: IJCAI, pp. 2586–2591

2007
[156]

Introduction to artifi- cial intelligence and machine learning in pathology and medicine: Generative and nongenerative artificial intelligence basics

Rashidi, H.H., Pantanowitz, J., Hanna, M.G., Tafti, A.P., Sanghani, P., Buchinsky, A., Pantanowitz, L., 2025. Introduction to artifi- cial intelligence and machine learning in pathology and medicine: Generative and nongenerative artificial intelligence basics. Modern Pathology...

2025 doi
[157]

Human-in-the-loop reinforcement learning: A survey and position on requirements, challenges, and opportunities

Retzlaff, C.O., et al., 2024. Human-in-the-loop reinforcement learning: A survey and position on requirements, challenges, and opportunities. ArtificialIntelligenceReview56,3005–3054. doi: 10. 1007/s10462-022-10246-w

2024
[158]

Chal- lenges for reinforcement learning in healthcare

Riachi, E., Mamdani, M., Fralick, M., Rudzicz, F., 2021. Chal- lenges for reinforcement learning in healthcare. arXiv preprint arXiv:2103.05612

2021 arXiv
[159]

Rl-based federated learningframeworkoverblockchain(rl-fl-bc),in:IEEETransactions onNetworkandServiceManagement,pp.1587–1597

Riahi, A., Mohamed, A., Erbad, A., 2023. Rl-based federated learningframeworkoverblockchain(rl-fl-bc),in:IEEETransactions onNetworkandServiceManagement,pp.1587–1597. doi: 10.1109/ TNSM.2023.3241437

2023
[160]

Rieke, N., Hancox, J., Li, W., Milletari, F., Roth, H.R., Albarqouni, S., Bakas, S., Galtier, M.N., Landman, B.A., Maier-Hein, K., et al., 2020.Thefutureofdigitalhealthwithfederatedlearning.NPJdigital medicine 3, 119

2020
[161]

Reinforcement learning for intensive care medicine: actionable clinical insights from novel approaches to reward shaping and off-policy model evaluation

Roggeveen, L.F., Hassouni, A.E., de Grooth, H.J., Girbes, A.R., Hoogendoorn, M., Elbers, P.W., 2024. Reinforcement learning for intensive care medicine: actionable clinical insights from novel approaches to reward shaping and off-policy model evaluation. In- tensive Care Medic...

2024
[162]

Multi-objective model- basedreinforcementlearningforinfectiousdiseasecontrol.Proceed- ings of KDD URL:https://doi.org/10.1145/3447548.3467303

Runzhe Wan, Xinyu Zhang, R.S., 2024. Multi-objective model- basedreinforcementlearningforinfectiousdiseasecontrol.Proceed- ings of KDD URL:https://doi.org/10.1145/3447548.3467303

2024
[163]

Saboo, K., Choudhary, A., Cao, Y., Worrell, G., Jones, D., Iyer, R.,
[164]

arXiv:2407.05800 URL:https://arxiv.org/ abs/2407.05800

Sahoo,P.,Tripathi,A.,Saha,S.,Mondal,S.,2024.Fedmrl:Datahet- erogeneity aware federated multi-agent deep reinforcement learning for medical imaging. arXiv:2407.05800 URL:https://arxiv.org/ abs/2407.05800

2024 arXiv
[165]

Enhanced decision-making in healthcare cloud-edgenetworksusingdeepreinforcementandlionoptimization algorithm

Saranya, S., Anusha, P., Chandragandhi, S., Kishore, O.K., Kumar, N.P., Srihari, K., 2024. Enhanced decision-making in healthcare cloud-edgenetworksusingdeepreinforcementandlionoptimization algorithm. Biomedical Signal Processing and Control 92, 105963. doi:10.1016/j.bspc.2024.105963

2024
[166]

Enhancing arrhythmia prediction through an adaptive deep rein- forcement learning framework for ecg signal analysis

Serhani, M.A., Ismail, H., El-Kassabi, H.T., Al Breiki, H., 2025. Enhancing arrhythmia prediction through an adaptive deep rein- forcement learning framework for ecg signal analysis. Biomedical Signal Processing and Control 101, 107155. doi:10.1016/j.bspc. 2024.107155

2025
[167]

Deep EHR: A Survey of Recent Advances in Deep Learning Techniques for Electronic Health Record (EHR) Analysis

Shickel, B., Tighe, P.J., Bihorac, A., Rashidi, P., 2018. Deep EHR: A Survey of Recent Advances in Deep Learning Techniques for Electronic Health Record (EHR) Analysis. IEEE Journal of Biomedical and Health Informatics 22, 1589–1604. doi:10.1109/ JBHI.2017.2767063

2018
[168]

Informing sequential clinical decision-making through reinforcement learning: an empirical study

Shortreed, S.M., Laber, E., Lizotte, D.J., Stroup, T.S., Pineau, J., Murphy, S.A., 2011. Informing sequential clinical decision-making through reinforcement learning: an empirical study. Machine learn- ing 84, 109–136

2011
[169]

Mastering the game of go with deep neural networks and tree search

Silver, D., Huang, A., Maddison, C.J., Guez, A., Sifre, L., Van Den Driessche, G., Schrittwieser, J., Antonoglou, I., Panneershel- vam, V., Lanctot, M., et al., 2016. Mastering the game of go with deep neural networks and tree search. nature 529, 484–489

2016
[170]

Masteringthegameofgowithout human knowledge

Silver, D., Schrittwieser, J., Simonyan, K., Antonoglou, I., Huang, A.,Guez,A.,Hassabis,D.,2017. Masteringthegameofgowithout human knowledge. nature 550, 354–359

2017
[171]

A reinforcementlearningagentforheadandneckintensity-modulated radiation therapy

Stephens, H., Li, X., Sheng, Y., Wu, Q., Ge, Y., Wu, Q.J., 2024. A reinforcementlearningagentforheadandneckintensity-modulated radiation therapy. Frontiers in Physics 12, 1331849

2024
[172]

Introduction to reinforcement learning

Sutton, R., Barto, A., 1998. Introduction to reinforcement learning. volume 135. MIT press Cambridge

1998
[173]

Reinforcement learning

Sutton, R.S., Barto, A.G., et al., 1999. Reinforcement learning. Journal of Cognitive Neuroscience 11, 126–134

1999
[174]

Role of artificial intelligence in haematolymphoid diagnostics

Syrykh, C., van den Brand, M., Kather, J.N., Laurent, C., 2025. Role of artificial intelligence in haematolymphoid diagnostics. Histopathology 86, 58–68. doi:10.1111/his.15327

2025 doi
[175]

Arti- ficialintelligenceanddeeplearning:Newtoolsforhistopathological diagnosis of nonalcoholic fatty liver disease/nonalcoholic steato- hepatitis

Takahashi,Y.,Dungubat,E.,Kusano,H.,Fukusato,T.,2023a. Arti- ficialintelligenceanddeeplearning:Newtoolsforhistopathological diagnosis of nonalcoholic fatty liver disease/nonalcoholic steato- hepatitis. Computational and Structural Biotechnology Journal 21, 2495–2501. doi:10.101...

2023 doi
[176]

Arti- ficialintelligenceanddeeplearning:Newtoolsforhistopathological diagnosis of nonalcoholic fatty liver disease/nonalcoholic steato- hepatitis

Takahashi,Y.,Dungubat,E.,Kusano,H.,Fukusato,T.,2023b. Arti- ficialintelligenceanddeeplearning:Newtoolsforhistopathological diagnosis of nonalcoholic fatty liver disease/nonalcoholic steato- hepatitis. Computational and Structural Biotechnology Journal 21, 2495–2501
[177]

Advancing multi-organ disease care: A hierarchical multi-agent reinforcement D

Tan,D.J.,Xu,Q.,See,K.C.,Perera,D.,Feng,M.,2024. Advancing multi-organ disease care: A hierarchical multi-agent reinforcement D. Perera et al.:Preprint submitted to Elsevier Page 26 of 39 Reinforcement Learning as the Defining Leap in Healthcare AI learning framework. arXiv pre...

2024 arXiv
[178]

Equilibrate rlhf: Towards balancing helpfulness-safetytrade-offinlargelanguagemodels

Tan, Y., Jiang, Y., Li, Y., Liu, J., Bu, X., Su, W., Yue, X., Zhu, X., Zheng, B., 2025. Equilibrate rlhf: Towards balancing helpfulness-safetytrade-offinlargelanguagemodels. arXivpreprint arXiv:2502.11555

2025 arXiv
[179]

Tang, K.F., Kao, H.C., Chou, C.N., Chang, E.Y., 2016. Inquire and diagnose:Neuralsymptomcheckingensembleusingdeepreinforce- ment learning, in: Advances in Neural Information Processing Sys- tems 29 (NIPS 2016) Workshop on Deep Reinforcement Learning, Neural Information Process...

2016
[180]

Leveraging factored action spaces for efficient offline reinforcement learning in healthcare, in: Advances in Neural Information Process- ing Systems, pp

Tang, S., Makar, M., Sjoding, M., Doshi-Velez, F., Wiens, J., 2022. Leveraging factored action spaces for efficient offline reinforcement learning in healthcare, in: Advances in Neural Information Process- ing Systems, pp. 34272–34286

2022
[181]

Model selection for offline reinforce- ment learning: Practical considerations for healthcare settings, in: Machine Learning for Healthcare Conference, PMLR

Tang, S., Wiens, J., 2021. Model selection for offline reinforce- ment learning: Practical considerations for healthcare settings, in: Machine Learning for Healthcare Conference, PMLR. pp. 2–35

2021
[182]

Challenges and opportunities beyond structured data in analysis of electronic health records

Tayefi,M.,Ngo,P.,Chomutare,T.,Dalianis,H.,Salvi,E.,Budrionis, A., Godtliebsen, F., 2021. Challenges and opportunities beyond structured data in analysis of electronic health records. Wiley Interdisciplinary Reviews: Computational Statistics 13, e1549

2021
[183]

Tchango,A.F.,Goel,R.,Martel,J.,Wen,Z.,Caron,G.M.,Ghosn,J.,
[184]

Rule- bottleneck reinforcement learning: Joint explanation and decision optimization for resource allocation with language agents

Tec,M.,Xiong,G.,Wang,H.,Dominici,F.,Tambe,M.,2025. Rule- bottleneck reinforcement learning: Joint explanation and decision optimization for resource allocation with language agents. arXiv preprint arXiv:2502.10732

2025 arXiv
[185]

Decentralized federated learning forhealthcarenetworks:Acasestudyontumorsegmentation

Tedeschini, B.C., Savazzi, S., Stoklasa, R., Barbieri, L., Stathopou- los, I., Nicoli, M., Serio, L., 2022. Decentralized federated learning forhealthcarenetworks:Acasestudyontumorsegmentation. IEEE Access 10, 8693–8705. doi:10.1109/ACCESS.2022.3141913

2022
[186]

Aknowledgeinfused context driven dialogue agent for disease diagnosis using hierarchi- calreinforcementlearning.Knowledge-BasedSystems242,108292

Tiwari,A.,Saha,S.,Bhattacharyya,P.,2022a. Aknowledgeinfused context driven dialogue agent for disease diagnosis using hierarchi- calreinforcementlearning.Knowledge-BasedSystems242,108292. doi:10.1016/j.knosys.2021.108292

2021
[187]

Aknowledgeinfused context driven dialogue agent for disease diagnosis using hierarchi- calreinforcementlearning.Knowledge-BasedSystems242,108292

Tiwari,A.,Saha,S.,Bhattacharyya,P.,2022b. Aknowledgeinfused context driven dialogue agent for disease diagnosis using hierarchi- calreinforcementlearning.Knowledge-BasedSystems242,108292
[188]

Voicegrpo: Modern moe transformers with group relative policy optimization grpo for ai voice health care applications on voice pathology detec- tion

Togootogtokh, E., Klasen, C., 2025. Voicegrpo: Modern moe transformers with group relative policy optimization grpo for ai voice health care applications on voice pathology detec- tion. arXiv:2503.03797 URL: https://github.com/enkhtogtokh/ voicegrpo

2025 arXiv
[189]

Whatclinicianswant:contextualizingexplainablemachinelearning for clinical end use, in: Machine learning for healthcare conference, PMLR

Tonekaboni,S.,Joshi,S.,McCradden,M.D.,Goldenberg,A.,2019. Whatclinicianswant:contextualizingexplainablemachinelearning for clinical end use, in: Machine learning for healthcare conference, PMLR. pp. 359–380

2019
[190]

Deepreinforcementlearn- ing for fractionated radiotherapy in non-small cell lung carcinoma

Tortora, M., Cordelli, E., Sicilia, R., Miele, M., Matteucci, P., Iannello,G.,Ramella,S.,Soda,P.,2021. Deepreinforcementlearn- ing for fractionated radiotherapy in non-small cell lung carcinoma. Artificial Intelligence in Medicine 119, 102137

2021
[191]

URL: https://www.sciencedirect.com/science/article/pii/ S1566253523000891, doi:https://doi.org/10.1016/j.inffus.2023.03. 008

2023 doi
[192]

NeurIPS 2022 Conference Paper URL:https://arxiv.org/abs/2207.07288

Towards trustworthy automatic diagnosis systems by emulat- ing doctors’ reasoning with deep reinforcement learning. NeurIPS 2022 Conference Paper URL:https://arxiv.org/abs/2207.07288

2022 arXiv
[193]

Ctd4–a deep continuous distributional actor- criticagentwithakalmanfusionofmultiplecritics,in:Proceedings of the AAAI Conference on Artificial Intelligence, pp

Valencia, D., Williams, H., Xing, Y., Gee, T., MacDonald, B.A., Liarokapis, M., 2025. Ctd4–a deep continuous distributional actor- criticagentwithakalmanfusionofmultiplecritics,in:Proceedings of the AAAI Conference on Artificial Intelligence, pp. 20956– 20963

2025
[194]

Optimized glycemic control of type 2 diabetes with reinforcement learning: a proof-of-concept trial

Wang,G.,Liu,X.,Ying,Z.,Yang,G.,Chen,Z.,Liu,Z.,Zhang,M., Yan, H., Lu, Y., Gao, Y., et al., 2023. Optimized glycemic control of type 2 diabetes with reinforcement learning: a proof-of-concept trial. Nature Medicine 29, 2633–2642

2023
[195]

Wang, L., Zhang, W., He, X., Zha, H., 2018. Supervised reinforce- ment learning with recurrent neural network for dynamic treatment recommendation, in: Proceedings of the 24th ACM SIGKDD inter- national conference on knowledge discovery & data mining, ACM. pp. 2447–2456

2018
[196]

Scoping review and proposed curriculum for robotic hepatopancreatobiliarysurgerytraining.SurgicalEndoscopydoi: 10

Wang,Y.,Kirkpatrick,J.,Chao,P.,Koea,J.,Srinivasa,K.,Srinivasa, S., 2025. Scoping review and proposed curriculum for robotic hepatopancreatobiliarysurgerytraining.SurgicalEndoscopydoi: 10. 1007/s00464-025-11546-2

2025
[197]

Clinical knowledge-guided deep reinforcement learning for sepsis antibiotic dosing recommendations

Wang,Y.,Liu,A.,Yang,J.,Wang,L.,Xiong,N.,Cheng,Y.,Wu,Q., 2024a. Clinical knowledge-guided deep reinforcement learning for sepsis antibiotic dosing recommendations. Artificial Intelligence in Medicine 150, 102811
[198]

Deepreinforcement learning for early diagnosis of lung cancer, in: Proceedings of the AAAI Conference on Artificial Intelligence, pp

Wang,Y.,Zhang,Q.,Ying,L.,Zhou,C.,2024b. Deepreinforcement learning for early diagnosis of lung cancer, in: Proceedings of the AAAI Conference on Artificial Intelligence, pp. 22410–22419
[199]

Deepreinforcement learning for early diagnosis of lung cancer, in: Proceedings of the AAAI Conference on Artificial Intelligence, pp

Wang,Y.,Zhang,Q.,Ying,L.,Zhou,C.,2024c. Deepreinforcement learning for early diagnosis of lung cancer, in: Proceedings of the AAAI Conference on Artificial Intelligence, pp. 22410–22419
[200]

MedicalPhysics44,6690–6705

Tseng,H.H.,etal.,2017.Deepreinforcementlearningforautomated radiationadaptationinlungcancer. MedicalPhysics44,6690–6705

2017
[201]

Fromdata to optimal decision making: a data-driven, probabilistic machine learningapproachtodecisionsupportforpatientswithsepsis

Tsoukalas,A.,Albertson,T.,Tagkopoulos,I.,etal.,2015. Fromdata to optimal decision making: a data-driven, probabilistic machine learningapproachtodecisionsupportforpatientswithsepsis. JMIR medical informatics 3, e3445

2015
[202]

MEGACare: Knowledge-guidedmulti-viewhypergraphpredictiveframeworkfor healthcare

Wu, J., He, K., Mao, R., Li, C., Cambria, E., 2023a. MEGACare: Knowledge-guidedmulti-viewhypergraphpredictiveframeworkfor healthcare. Information Fusion 100, 101939
[203]

Harnessing the potential of multimodal EHR data: A comprehensive survey of clinical predictive modeling for intelligent healthcare

Wu,J.,He,K.,Mao,R.,Shang,X.,Cambria,E.,2025a. Harnessing the potential of multimodal EHR data: A comprehensive survey of clinical predictive modeling for intelligent healthcare. Information Fusion , 103283
[204]

Promise: A pre-trained knowledge-infused multimodal representation learning frameworkformedicationrecommendation

Wu, J., Yu, X., He, K., Gao, Z., Gong, T., 2024. Promise: A pre-trained knowledge-infused multimodal representation learning frameworkformedicationrecommendation. InformationProcessing & Management 61, 103758

2024
[205]

Reinforce- ment learning for healthcare operations management: methodologi- cal framework, recent developments, and future research directions

Wu,Q.,Han,J.,Yan,Y.,Kuo,Y.H.,Shen,Z.J.M.,2025b. Reinforce- ment learning for healthcare operations management: methodologi- cal framework, recent developments, and future research directions. Health Care Management Science , 1–36
[206]

A value-based deepreinforcementlearningmodelwithhumanexpertiseinoptimal treatmentofsepsis

Wu, X., Li, R., He, Z., Yu, T., Cheng, C., 2023b. A value-based deepreinforcementlearningmodelwithhumanexpertiseinoptimal treatmentofsepsis. npjDigitalMedicine6. URL: https://doi.org/ 10.1038/s41746-023-00755-5, doi:10.1038/s41746-023-00755-5

doi
[207]

A value-based deepreinforcementlearningmodelwithhumanexpertiseinoptimal treatment of sepsis

Wu, X., Li, R., He, Z., Yu, T., Cheng, C., 2023c. A value-based deepreinforcementlearningmodelwithhumanexpertiseinoptimal treatment of sepsis. npj Digital Medicine 6
[208]

Adaptive wavelet base selection for deep learning-based ecg diagnosis: A reinforcement learning approach

Xiao, Q., Wang, C., 2025. Adaptive wavelet base selection for deep learning-based ecg diagnosis: A reinforcement learning approach. PloS One 20, e0318070. doi:10.1371/journal.pone.0318070

2025 doi
[209]

Reinforcement learning for adaptive scheduling and optimization of healthcare staff and resources in multi-departmental hospitals

Wattanapanit, S., 2025. Reinforcement learning for adaptive scheduling and optimization of healthcare staff and resources in multi-departmental hospitals. Open Journal of Robotics, Au- tonomousDecision-Making,andHuman-MachineInteraction10,1– 15

2025
[210]

Do noharm:aroadmapforresponsiblemachinelearningforhealthcare

Wiens, J., Saria, S., Sendak, M., Ghassemi, M., Liu, V.X., Doshi- Velez, F., Jung, K., Heller, K., Kale, D., Saeed, M., et al., 2019. Do noharm:aroadmapforresponsiblemachinelearningforhealthcare. Nature medicine 25, 1337–1340

2019
[211]

Deep generative adversarial reinforcement learning for semi-supervised segmentation of low-contrast and small objects in medical images

Xu, C., Zhang, T., Zhang, D., Zhang, D., Han, J., 2024a. Deep generative adversarial reinforcement learning for semi-supervised segmentation of low-contrast and small objects in medical images. IEEE Transactions on Medical Imaging
[212]

Meddreamer:Model- based reinforcement learning with latent imagination on complex D

Xu,Q.,Habib,G.,Perera,D.,Feng,M.,2025. Meddreamer:Model- based reinforcement learning with latent imagination on complex D. Perera et al.:Preprint submitted to Elsevier Page 27 of 39 Reinforcement Learning as the Defining Leap in Healthcare AI ehrs for clinical decision suppor...

2025
[213]

Xrl-shap-cache: an explainable reinforcement learning approach for intelligent edge service caching in content delivery networks

Xu, X., Wu, F., Bilal, M., Xia, X., Dou, W., Yao, L., Zhong, W., 2024b. Xrl-shap-cache: an explainable reinforcement learning approach for intelligent edge service caching in content delivery networks. Science China Information Sciences 67, 170303
[214]

Low-costdataoffloadingstrategywithdeepreinforcement learningforsmarthealthcaresystem

Xue,Z.,Feng,Z.,Chen,Z.,Nan,T.,Li,K.,Shen,H.,Yu,K.,Wang, X.,2024. Low-costdataoffloadingstrategywithdeepreinforcement learningforsmarthealthcaresystem. IEEETransactionsonServices Computing doi:10.1109/TSC.2024.3404347

2024
[215]

A resource-constrained and privacy-preserving edge- computing-enabled clinical decision system: A federated reinforce- ment learning approach

Xue, Z., Zhou, P., Xu, Z., Wang, X., Xie, Y., Ding, X., Wen, S., 2021a. A resource-constrained and privacy-preserving edge- computing-enabled clinical decision system: A federated reinforce- ment learning approach. IEEE Internet of Things Journal 8, 9122–
[216]

A resource-constrained and privacy-preserving edge- computing-enabled clinical decision system: A federated reinforce- ment learning approach

Xue, Z., Zhou, P., Xu, Z., Wang, X., Xie, Y., Ding, X., Wen, S., 2021b. A resource-constrained and privacy-preserving edge- computing-enabled clinical decision system: A federated reinforce- ment learning approach. IEEE Internet of Things Journal 8, 9122–
[217]

Smart healthcare: Rl-based task offloading scheme for edge-enable sensor networks

Yadav, R., Zhang, W., Elgendy, I.A., Dong, G., Shafiq, M., Laghari, A.A., Prakash, S., 2021. Smart healthcare: Rl-based task offloading scheme for edge-enable sensor networks. IEEE Sensors Journal 21, 24910–24921. doi:10.1109/JSEN.2021.3096245

2021
[218]

Deep learning for temporal data representation in electronic health records: A systematic review of challenges and methodologies

Xie, F., Yuan, H., Ning, Y., Ong, M.E.H., Feng, M., Hsu, W., Chakraborty, B., Liu, N., 2022. Deep learning for temporal data representation in electronic health records: A systematic review of challenges and methodologies. Journal of biomedical informatics 126, 103980

2022
[219]

Epicare:Areinforcementlearning benchmarkfordynamictreatmentregimes

XinqiDu,YikuanLi,Y.L.,2023. Epicare:Areinforcementlearning benchmarkfordynamictreatmentregimes. ProceedingsofNeurIPS URL: https://arxiv.org/abs/2309.02210

2023 arXiv
[220]

The Rise, Fall, and Resurrection of IBM Watson Health

Yang, J., Chesbrough, H., Hurmelinna-Laukkanen, P., 2020. The Rise, Fall, and Resurrection of IBM Watson Health. Research Project Report. UC Berkeley and University of Oulu. Research Project, Fall 2019

2020
[221]

Deep reinforcement learn- ing for multi-class imbalanced training: applications in healthcare

Yang, J., El-Bouri, R., O’Donoghue, O., Lachapelle, A.S., Soltan, A.A., Eyre, D.W., Clifton, D.A., 2024a. Deep reinforcement learn- ing for multi-class imbalanced training: applications in healthcare. Machine Learning 113, 2655–2674
[222]

A optimization framework for herbal prescription planning based on deep reinforcement learning

Yang, K., Yu, Z., Su, X., He, X., Wang, N., Zheng, Q., Yu, F., Liu, Z., Wen, T., Zhou, X., 2023b. A optimization framework for herbal prescription planning based on deep reinforcement learning. arXiv preprint arXiv:2304.12828

arXiv
[223]

Aligning large language models with radiologists by reinforcement learning from ai feedback for chest ct reports

Yang, L., Zhou, Y., Qi, J., Zhen, X., Sun, L., Shi, S., Su, Q., Yang, X., 2025a. Aligning large language models with radiologists by reinforcement learning from ai feedback for chest ct reports. European Journal of Radiology 184, 111984
[224]

Aligning large language models with radiologists by reinforcement learning from ai feedback for chest ct reports

Yang,L.,Zhou,Y.,Qi,J.,Zhen,X.,Sun,L.,Shi,S.,Yang,X.,2025b. Aligning large language models with radiologists by reinforcement learning from ai feedback for chest ct reports. European Journal of Radiology 111984. doi:10.1016/j.ejrad.2025.111984

2025
[225]

Federated natural policy gradient and actor critic methods for multi-task re- inforcement learning

Yang, T., Cen, S., Wei, Y., Chen, Y., Chi, Y., 2024b. Federated natural policy gradient and actor critic methods for multi-task re- inforcement learning. Advances in Neural Information Processing Systems 37, 121304–121375
[226]

Journal of Theoretical Biology 594, 111914

Yoon,C.,Choi,J.,Kwon,H.D.,Kang,M.,2024.Optimalsticontrols forhivpatientsbasedonanefficientdeepqlearningmethod. Journal of Theoretical Biology 594, 111914

2024
[227]

Reinforcementlearningin healthcare: A survey

Yu,C.,Liu,J.,Nemati,S.,Yin,G.,2021. Reinforcementlearningin healthcare: A survey. ACM Computing Surveys

2021
[228]

Inverse reinforcement learning for intelligent mechanical ventilation and sedative dosing in intensive care units

Yu, C., Liu, J., Zhao, H., 2019. Inverse reinforcement learning for intelligent mechanical ventilation and sedative dosing in intensive care units. BMC Medical Informatics and Decision Making 19, 57

2019
[229]

Optimiz- ing risk-based breast cancer screening policies with reinforcement learning

Yala, A., Mikhael, P.G., Lehman, C., Lin, G., Strand, F., Wan, Y.L., Hughes,K.,Satuluru,S.,Kim,T.,Banerjee,I.,etal.,2022. Optimiz- ing risk-based breast cancer screening policies with reinforcement learning. Nature medicine 28, 136–143

2022
[230]

Reinforcement learning strategies in cancer chemother- apy treatments: A review

Yang, C.Y., Shiranthika, C., Wang, C.Y., Chen, K.W., Sumathipala, S., 2023a. Reinforcement learning strategies in cancer chemother- apy treatments: A review. Computer Methods and Programs in Biomedicine 229, 107280
[231]

Integratingreinforcementlearningandaiagentsfor adaptive robotic interaction and assistance in dementia care

Yuan, F., Hasnaeen, N., Zhang, R., Bible, B., Taylor, J.R., Qi, H., Zhao,X.,2025. Integratingreinforcementlearningandaiagentsfor adaptive robotic interaction and assistance in dementia care. arXiv preprint arXiv:2501.17206

2025 arXiv
[232]

Journal of biomedical informatics 137, 104267

Zadeh,S.A.,Street,W.N.,Thomas,B.W.,2023.Optimizingwarfarin dosing using deep reinforcement learning. Journal of biomedical informatics 137, 104267

2023
[233]

From decisiontoactioninsurgicalautonomy:Multi-modallargelanguage models for robot-assisted blood suction

Zargarzadeh, S., Mirzaei, M., Ou, Y., Tavakoli, M., 2025. From decisiontoactioninsurgicalautonomy:Multi-modallargelanguage models for robot-assisted blood suction. IEEE Robotics and Au- tomation Letters 10, 2598–2607. doi:10.1109/LRA.2025.3535184

2025
[234]

Meta-aad: Active anomalydetectionwithdeepreinforcementlearning,in:2020IEEE International Conference on Data Mining (ICDM), IEEE

Zha, D., Lai, K.H., Wan, M., Hu, X., 2020. Meta-aad: Active anomalydetectionwithdeepreinforcementlearning,in:2020IEEE International Conference on Data Mining (ICDM), IEEE. pp. 771– 780

2020
[235]

Antibiotics versus non-antibiotic in the treatment of aspiration pneumonia: analysis of the mimic-iv database

Zhang, D., Yang, G., Hu, X., Liu, X., Zhang, J., Jia, D., Zhang, A., 2024a. Antibiotics versus non-antibiotic in the treatment of aspiration pneumonia: analysis of the mimic-iv database. BMC Pulmonary Medicine 24, 621
[236]

European Journal of Medical Research 29, 156

Zhang,G.,Shao,F.,Yuan,W.,Wu,J.,Qi,X.,Gao,J.,Shao,R.,Tang, Z.,Wang,T.,2024b.Predictingsepsisin-hospitalmortalitywithma- chinelearning:amulti-centerstudyusingclinicalandinflammatory biomarkers. European Journal of Medical Research 29, 156
[237]

An explainable machine learning model for predicting response to targeted therapy and immunother- apy in advanced hepatocellular carcinoma: a multicentric study

Zhang,J.X.,Yan,H.T.,Fan,W.L.,Wei,J.,Liu,S.,Shi,H.B.,Cheng, Y., Feng, Z.l., Zu, Q.Q., 2024c. An explainable machine learning model for predicting response to targeted therapy and immunother- apy in advanced hepatocellular carcinoma: a multicentric study
[238]

IEEE Transactions on Network Science and Engineering 10, 2864–2875

Zhang, L., Xu, J., Vijayakumar, P., Sharma, P.K., Ghosh, U., 2023.Homomorphicencryption-basedprivacy-preservingfederated learning in iot-enabled healthcare system. IEEE Transactions on Network Science and Engineering 10, 2864–2875. doi: 10.1109/ TNSE.2022.3185327

2023
[239]

Optimizing sepsis treatment strategies via a reinforcement learning model

Zhang, T., Qu, Y., Wang, D., Zhong, M., Cheng, Y., Zhang, M., 2024d. Optimizing sepsis treatment strategies via a reinforcement learning model. Biomedical Engineering Letters 14, 279–289
[240]

Mopo: Model-based offline policy optimization

Yu, T., Thomas, G., Yu, L., Ermon, S., Zou, J.Y., Levine, S., Finn, C., Ma, T., 2020. Mopo: Model-based offline policy optimization. Advances in Neural Information Processing Systems 33, 14129– 14142

2020
[241]

Deep reinforcement learning for cost-effective medical diagnosis

Yu, Z., Li, Y., Kim, J.C., Huang, K., Luo, Y., Wang, M., 2023. Deep reinforcement learning for cost-effective medical diagnosis. Proceedings of the International Conference on Learning Represen- tations (ICLR 2023)

2023
[242]

Reinforcement learning-based adaptive deep brain stimulation computational model for the treat- ment of tremor in parkinson’s disease

Zhao, T., Faustino, B.L., Jagatheesaperumal, S.K., Rolim, F.D.P.S., de Albuquerque, V.H.C., 2025a. Reinforcement learning-based adaptive deep brain stimulation computational model for the treat- ment of tremor in parkinson’s disease. Expert Systems with Appli- cations 267, 126154
[243]

A safe-enhanced fully closed-loop artificial pancreascontrollerbasedondeepreinforcementlearning

Zhao, Y.F., Chaw, J.K., Ang, M.C., Tew, Y., Shi, X.Y., Liu, L., Cheng, X., 2025b. A safe-enhanced fully closed-loop artificial pancreascontrollerbasedondeepreinforcementlearning. PloSone 20, e0317662
[244]

Deep learning enables rapid identification of potent ddr1 kinase inhibitors

Zhavoronkov, A., Ivanenkov, Y.A., Aliper, A., Veselov, M.S., Al- adinskiy, V.A., Aladinskaya, A.V., Terentiev, V.A., Polykovskiy, D.A., Kuznetsov, M.D., Asadulaev, A., et al., 2019. Deep learning enables rapid identification of potent ddr1 kinase inhibitors. Nature biotechnolo...

2019
[245]

Learning how to detect: A deep reinforce- ment learning method for whole-slide melanoma histopathology images

Zheng,T.,Chen,W.,Li,S.,Quan,H.,Zou,M.,Zheng,S.,Zhao,Y., Gao, X., Cui, X., 2023. Learning how to detect: A deep reinforce- ment learning method for whole-slide melanoma histopathology images. ComputerizedMedicalImagingandGraphics108,102275. doi:10.1016/j.compmedimag.2023.102275

2023
[246]

Federated offline reinforcement learning

Zhou,D.,Zhang,Y.,Sonabend-W,A.,Wang,Z.,Lu,J.,Cai,T.,2024. Federated offline reinforcement learning. Journal of the American D. Perera et al.:Preprint submitted to Elsevier Page 28 of 39 Reinforcement Learning as the Defining Leap in Healthcare AI Statistical Association 119, 3152–3163

2024
[247]

Towardsphysiologically sensible predictionsvia the rule-basedreinforcement learning layer

Zhu,L.,Chen,Z.,Nagai,Y.,Sun,J.,2025. Towardsphysiologically sensible predictionsvia the rule-basedreinforcement learning layer. arXiv preprint arXiv:2501.19055

2025 arXiv
[248]

AmericanJournalofClinicalExperimentalUrology 12, 200–215

Zhu, M., Sali, R., Baba, F., Khasawneh, H., Ryndin, M., Leveillee, R.J.,Hurwitz,M.D.,Lui,K.,Dixon,C.,Zhang,D.Y.,2022.Artificial intelligence in pathologic diagnosis, prognosis, and prediction of prostatecancer. AmericanJournalofClinicalExperimentalUrology 12, 200–215. doi:10.6...

2022 doi
[249]

Offline deep reinforcement learning and off-policy evaluation for personalized basal insulin control in type 1 diabetes

Zhu, T., Li, K., Georgiou, P., 2023. Offline deep reinforcement learning and off-policy evaluation for personalized basal insulin control in type 1 diabetes. IEEE Journal of Biomedical and Health Informatics 27, 5087–5098

2023
[250]

Improving anticoagulant treat- mentstrategiesofatrialfibrillationusingreinforcementlearning,in: AMIA Annual Symposium Proceedings, p

Zuo, L., Du, X., Zhao, W., Jiang, C., Xia, S., He, L., Liu, R., Tang, R., Bai, R., Dong, J., et al., 2021. Improving anticoagulant treat- mentstrategiesofatrialfibrillationusingreinforcementlearning,in: AMIA Annual Symposium Proceedings, p. 1431. D. Perera et al.:Preprint subm...

2021
[251]

Zhang, Z., 2024. Advancing sample efficiency and explainability in multi-agent reinforcement learning, in: Proceedings of the 23rd International Conference on Autonomous Agents and Multiagent Systems, pp. 2791–2793

2024
[252]

An implicit trust region approach to behaviorregularizedofflinereinforcementlearning,in:Proceedings of the AAAI Conference on Artificial Intelligence, pp

Zhang, Z., Tan, X., 2024. An implicit trust region approach to behaviorregularizedofflinereinforcementlearning,in:Proceedings of the AAAI Conference on Artificial Intelligence, pp. 16944– 16952

2024
[262]

SDQN, SDDQN, SBCQ Offline, MFRL Optimized warfarin dosing by enhancingTimeinTherapeutic Range(TTR)forstrokepreven- tion

Appliedsemi-MarkovPolicygra- dient methods to optimize war- farin dosing. SDQN, SDDQN, SBCQ Offline, MFRL Optimized warfarin dosing by enhancingTimeinTherapeutic Range(TTR)forstrokepreven- tion. Generalizability, computational complexity, overfitting, offline limitations
[263]

FQI, FQE

Framework for offline RL model selection in sepsis treatment. FQI, FQE. Offline, MFRL Effective framework to select models for sepsis treatment Simulated data, limited action space, ethical concerns
[264]

DQN, D- DDQN Offline, MFRL Effectivemulti-classclassifica- tion under imbalance

Multi-class classification for COVID-19 screening and patient diagnosis in ICU. DQN, D- DDQN Offline, MFRL Effectivemulti-classclassifica- tion under imbalance. Interpretability, trust issues, ethical and privacy concerns, computational complexity
[265]

Optimizing individualized medicine intake schedules for Parkinson’s patients. A3C (Actor Critic) Online, Offline, MFRL 7% reduction in a cost function balancingmotorsymptomcon- trol and dosage regularity, Simulated data limits general- izability, RL algorithm not ex- plicitly stated
[266]

Q- Learning Offline, MFRL Improved sample efficiency and policy value in decision- making

Used factored action spaces in offline RL for improving sample efficiency and policy value. Q- Learning Offline, MFRL Improved sample efficiency and policy value in decision- making. Limited exploration in offline settings, lack of systematic studyoffactoredactionspaces
[267]

Cooperative multi-agent learning via linear value decomposition optimizing treatment strategies for diabetic ketoacidosis (DKA). DQN. Offline, MBRL Improved BG control with a rangeof70-150mg/dlforDKA patients Generalization, limited real- time applications, incomplete EHR data...
[268]

A3C Offline, MBRL Optimized ventilator settings with high data efficiency in ICU Generalization and clinical adoption to real-time patient treatments D

Integrates discrete and continu- ous action handling for individu- alized ventilator control in ICU. A3C Offline, MBRL Optimized ventilator settings with high data efficiency in ICU Generalization and clinical adoption to real-time patient treatments D. Perera et al.:Preprint ...
[269]

Optimizing vasopressor policies forARDSsepsiswithDRLusing LSTM-GRUstaterepresentation. DDQN, DDDQN Offline, MFRL Optimizedvasopressorpolicies for ARDS sepsis using DRL with LSTM-GRU Limited patient data, synthetic data generation using Time GAN, computational complex- ity,andg...
[270]

Optimizing sepsis treatment with DDQN, reducing in-hospital mortality DDQN Offline, MFRL Reduced in-hospital mortality by 3% in sepsis treatment Limited applicability to other databases, state heterogeneity, ethical privacy, and interpretability concerns
[271]

DDPG and Twin Delayed DDPG (TD3) Offline, MFRL Optimized sepsis treatment, aligning with clinician decisions safely

Introduced continuous action RL for sepsis treatment. DDPG and Twin Delayed DDPG (TD3) Offline, MFRL Optimized sepsis treatment, aligning with clinician decisions safely. Extrapolation errors, overesti- mation in action performance, clinical safety, data general- izability, an...
[272]

WD3QNE Offline, Rule based, MFRL Achieved97.81%survival,out- performing other methods in sepsis treatment

Applies expert-informed value- based RL (WD3QNE) to person- alize sepsis treatments. WD3QNE Offline, Rule based, MFRL Achieved97.81%survival,out- performing other methods in sepsis treatment. Generalization, clinical safety, and ethical concerns
[273]

DQN Offline RL Outperformed baseline clinical models in early lung cancer diagnosis rate (60.88%) with lower error rates Explainability and clinical scalability

Combining reinforcement learn- ing with clinical and image data to support equitable and person- alized treatment decisions. DQN Offline RL Outperformed baseline clinical models in early lung cancer diagnosis rate (60.88%) with lower error rates Explainability and clinical scalability
[274]

IRL Offline RL, MFRL Upto260%improvementinin- tervention allocation efficiency and11%increaseinlisteneren- gagement,outperformingbase- lines

IRL to optimize intervention scheduling in maternal and child health programs, using expert-derived preferences over restless bandit environments. IRL Offline RL, MFRL Upto260%improvementinin- tervention allocation efficiency and11%increaseinlisteneren- gagement,outperformingb...
[275]

SM- DDPO Online, Offline, RL Achieved state-of-the-art diag- nosticaccuracywithupto80% reductionintestcost,adaptable to new diseases without prior knowledge

Formulating cost-sensitive medi- cal diagnosis as a multi-objective reinforcement learning problem to learn policies along the cost- accuracy Pareto frontier using clinical data. SM- DDPO Online, Offline, RL Achieved state-of-the-art diag- nosticaccuracywithupto80% reductionin...
[276]

Highlight enhanced DDDQN MoE (Offline & Rule Based), MFRL Up to 10.03% and 9.81% higher estimated survival rates on MIMIC-III and eICU datasets, vs actual treatments

Extracting optimal sepsis treat- ment policies using highlight- enhanced DDDQN trained on historical ICU records. Highlight enhanced DDDQN MoE (Offline & Rule Based), MFRL Up to 10.03% and 9.81% higher estimated survival rates on MIMIC-III and eICU datasets, vs actual treatmen...
[277]

Fo- cused on safe offline learning from physician behavior

Recommends personalized venti- lator settings (PEEP, FiO2, tidal volume) using ICU data. Fo- cused on safe offline learning from physician behavior. BCQ Offline, MFRL Reduction in estimated mor- tality (eICU:12.1%, MIMIC- IV:29.1%) and improved oxy- genation and BP vs. clinica...
[278]

Lack of features for certain diagnoses (e.g., renal events), need for more data for multi- classtasks,andsite-specificbi- ases in datasets

Imbalanced classification for COVID-19 prediction and ICU discharge prediction DDDQN Offline, MFRL Improved sensitivity (0.806) with a 0.077 improvement for minorityclassprediction,espe- cially for COVID-19 and ICU discharge tasks. Lack of features for certain diagnoses (e.g.,...
[279]

Semi-supervisedsegmentationof low-contrast and small objects in medical images using Deep gen- erative Adversarial RL. Soft Actor- Critic (SAC) Offline, MFRL Improved Dice: 85.02% (brain tumor, +1.91%), 73.18% (liver tumor, +4.28%), 70.85% (pan- creas, +2.73%) High computation...
[280]

Applied to sepsis and diabetes

Learning from imperfect clinical decisions via a 2-phase frame- work using adversarial cooper- ative imitation learning and in- verse RL. Applied to sepsis and diabetes. ACIL, IRL, DQN Offline, MFRL Reduced sepsis mortality by 19.6% compared to best base- line and HbA1c-High r...
[281]

Policy Iteration (AI Clin- ician) Offline, MFRL < 1% had complete data by hour 6; 49% received no treat- ment

Assessed transportability of the ICU-trained AI Clinician RL model to emergency department (ED) sepsis care; focused on 4 domains: missing data, measurement timing, diagnostic uncertainty, and treatment variability. Policy Iteration (AI Clin- ician) Offline, MFRL < 1% had comp...
[282]

text-to-plan

Clinical knowledge-guided an- tibiotic combination recommen- dation for sepsis using RL. DQN Offline, MFRL Achieved 79.09% survival rate (vs. 60.2% clinical baseline), 54.77% of clinically deceased patients predicted to benefit fromrecommendedtreatments. Reduced average antibi...

1996
[2015]

Advances in neural information processing systems 28

Learning continuous control policies by stochastic value gradients. Advances in neural information processing systems 28
[2017]

Journal of Machine Learning Research 18, 1–31

Combining kernel and model-based learning for hiv therapy selection. Journal of Machine Learning Research 18, 1–31
[2021]

20903–20915

Reinforcement learning based disease progression model for alzheimer’s disease, in: Advances in Neural Information Processing Systems, pp. 20903–20915
[2022]

Patterns 3

Machine learning for predicting readmission risk among the frail: Explainable ai for healthcare. Patterns 3
[2023]

IEEE Reviews in Biomedical Engineering 17, 42–62

Beyond supervised learning for pervasive healthcare. IEEE Reviews in Biomedical Engineering 17, 42–62
[2024]

BMC Medical Informatics and Decision Making 24, 228

Prediction of sepsis mortality in icu patients using machine learning methods. BMC Medical Informatics and Decision Making 24, 228
[2025]

medRxiv , 2025–01

Glucose: A distributional reinforcement learning model for optimal glucose control after cardiac surgery. medRxiv , 2025–01

2025
[9132]

doi:10.1109/JIOT.2021.3057653

2021

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [13]

Optimizing substance use treatment selection using reinforcement learning

Baucum, M., Doe, J., Smith, J., 2023. Optimizing substance use treatment selection using reinforcement learning. ACM Transac- tionsonIntelligentSystemsandTechnology14,45–67.URL: https: //doi.org/10.1145/1234567.1234568, doi:10.1145/1234567.1234568

work page arXiv 2023

[2] [115]

Reinforcement learning to optimize ventilator settings for patients on invasive mechanical ventilation: Retrospective study

Liu, S., Xu, Q., Xu, Z., Liu, Z., Sun, X., Xie, G., See, K.C., 2024b. Reinforcement learning to optimize ventilator settings for patients on invasive mechanical ventilation: Retrospective study. Journal of Medical Internet Research 26

[3] [133]

Challenges with reinforcement learning model transportability for sepsis treatment in emergency care

Nauka, P.C., Kennedy, J.N., Brant, E.B., Komorowski, M., Pirrac- chio, R., Angus, D.C., Seymour, C.W., 2025. Challenges with reinforcement learning model transportability for sepsis treatment in emergency care. npj Digital Medicine 8, 1–5

2025

[4] [120]

Probabalistic rein- forcement learning impairments predict negative symptom severity and risk for conversion in youth at clinical high-risk for psychosis

Luther, L., Raugh, I.M., Strauss, G.P., 2025a. Probabalistic rein- forcement learning impairments predict negative symptom severity and risk for conversion in youth at clinical high-risk for psychosis. Psychological Medicine 55, e28

[5] [121]

Probabilistic rein- forcement learning impairments predict negative symptom severity and risk for conversion in youth at clinical high-risk for psychosis

Luther, L., Raugh, I.M., Strauss, G.P., 2025b. Probabilistic rein- forcement learning impairments predict negative symptom severity and risk for conversion in youth at clinical high-risk for psychosis. Psychological Medicine 55, e28

[6] [1]

Apprenticeshiplearningviainverserein- forcement learning, in: Proceedings of the twenty-first international conference on Machine learning, p

Abbeel,P.,Ng,A.Y.,2004. Apprenticeshiplearningviainverserein- forcement learning, in: Proceedings of the twenty-first international conference on Machine learning, p. 1

2004

[7] [2]

A prehospital triage system to detect traumatic intracranial hemorrhage using machine learning algorithms

Abe, D., Inaji, M., Hase, T., Takahashi, S., Sakai, R., Ayabe, F., Tanaka, Y., Otomo, Y., Maehara, T., 2022. A prehospital triage system to detect traumatic intracranial hemorrhage using machine learning algorithms. JAMA Network Open 5, e2216393. URL: https://jamanetworkopen.jamanetwork.com/article.aspx?doi= 10.1001/jamanetworkopen.2022.16393, doi:10.1001...

work page arXiv 2022

[8] [3]

Machine learning in clinical decision making

Adlung, L., Cohen, Y., Mor, U., Elinav, E., 2021. Machine learning in clinical decision making. Med 2, 642–665

2021