Pith. sign in

REVIEW 3 major objections 5 minor 26 references

OPTIC: Optimizing Patient-Provider Triaging & Improving Communications in Clinical Operations using GPT-4 Data Labeling and Model Distillation

T0 review · 3 major / 5 minor · reviewed 2026-08-09 · deepseek-v4-flash

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

Pith's one-line read A GPT-4-distilled BERT model sorts patient messages with about 89% accuracy.

desk verdict A real deployment story with a flawed headline metric: the reported 88.85% accuracy is BERT agreeing with GPT-4 labels, not with physician ground truth. read the letter →

arxiv 2503.05701 v1 pith:Q4IU2URF submitted 2025-02-05 cs.LG cs.CL

Alberto Santamaria-Pang , Frank Tuan , Ross Campbell , Cindy Zhang , Ankush Jindal , Roopa Surapur , Brad Holloman , Deanna Hanisch
show 6 more authors
Rae Buckley Carisa Cooney Ivan Tarapov Kimberly S. Peairs Brian Hasselfeld Peter Greene
This is my paper · ORCID
classification cs.LGcs.CL
keywords patientportalmessagesInBaskettriageGPT-4datalabelingmodeldistillationBERTclassifieradministrativeversusclinicaloperationsEHRdeployment
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

The paper claims that a small, deployable language model can take over a large share of patient-portal triage: a BERT classifier distilled from GPT-4 labels separates administrative from clinical patient messages with 88.85% accuracy on a 3,454-message test set. The motivation is concrete: physicians receive dozens to hundreds of inbox messages a day, and prior studies cited in the paper put the administrative share above 50%, meaning much of that volume could be handled by non-clinical staff. Using roughly 405,000 patient portal encounters, the authors used GPT-4 few-shot prompting with 200 physician-reviewed examples to label about 36,000 messages, then trained BERT on those labels. The resulting model was deployed through the EHR vendor's Nebula cloud platform, and topic-level analysis shows it stays above 80% accuracy on 58 of 81 message topics. If the result holds, the pipeline offers a practical route from a large language model to a cheap, CPU-runnable classifier already embedded in the EHR workflow.

What carries the argument

The load-bearing mechanism is model distillation from a large language model to a small one. GPT-4-32K is given a few-shot prompt containing 200 messages, evenly split between administrative and clinical examples that were selected by topic-based sampling and reviewed by a physician, and is asked to classify each new message and explain its reasoning. Among four prompt variants, this configuration scored 0.99 on a 2,000-message physician-evaluated validation set, so its labels were used to train a BERT classifier. BERT (Bidirectional Encoder Representations from Transformers) is a pretrained transformer that learns a compact, CPU-runnable representation of text; distilled on roughly 33,000 GPT-4-labeled messages, it reproduces most of the teacher's triage judgment at a fraction of the inference cost. BERTopic, an unsupervised topic model, is then applied to the test set to show that accuracy remains above 80% across most of the 81 identified topics.

What would settle it

Take a random sample of several hundred test-set messages, have multiple physicians independently label each one as administrative or clinical, and compare BERT's predictions, GPT-4's labels, and the physicians' majority vote. If BERT's agreement with the human labels is substantially below the reported 88.85%, or if GPT-4 and the physicians disagree on a large share of messages, then the reported accuracy measures agreement with an unvalidated teacher rather than real-world performance.

Watch

Extended reading notes

Core claim

The paper's central claim is that a production-ready triage classifier can be built by using GPT-4 as an annotator and BERT as the distilled student model. On the task of labeling a patient's first message in an encounter as either administrative (scheduling, refills, insurance forms) or clinical (symptoms, test results, treatment questions), the distilled BERT model reaches 88.85% accuracy, 88.29% sensitivity, 89.38% specificity, and an F1 score of 0.8842. The same model is deployed through the EHR vendor's Nebula SaaS platform, meaning the classifier operates inside the same system that hosts patient portal inboxes rather than as a standalone research prototype. The authors position this as an improvement over prior patient-message classifiers, which either used smaller datasets or required higher-cost models, and as a response to the post-pandemic surge in portal messages.

Load-bearing premise

The load-bearing premise is that GPT-4's labels on the test messages are correct enough to count as ground truth; only the 2,000-message validation set used for prompt selection was checked by three physicians, and the test-set labels themselves were not human-validated.

Editorial extensions

If this is right

  • A lightweight BERT model can run on CPU inside the EHR, making automated triage affordable at the scale of a large health system that processes over a million inbox messages a year.
  • Routing administrative messages to non-clinical staff could remove more than half of the inbox volume physicians currently see, since prior studies cited by the paper put the administrative share above 50%.
  • The same GPT-4-labeling-then-distillation recipe can be reused for other message taxonomies or other specialties, provided new few-shot exemplars are curated and physician-validated.
  • Deployment through the EHR vendor's Nebula platform means the classifier is available where messages already live, so it can be tested by real outpatient providers rather than only in batch experiments.
  • Topic-level analysis suggests the model is not tuned to a few frequent message types; it maintains over 80% accuracy on 58 of 81 topics, which supports broader rollout within primary care.

Reading between the lines

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

  • The true accuracy against human ground truth is still unknown: the reported 88.85% is agreement with GPT-4, and the paper does not state that physicians validated the test labels. A small human-labeled benchmark would settle this.
  • The binary admin/clinical split may understate real-world triage difficulty, because many portal messages mix clinical content with administrative requests; a third 'mixed' category or an urgency tier would be a natural extension.
  • The retrospective metadata groups (employee/service-role flags for administrative, order/note linkage for clinical) are plausible but imperfect proxies; if those groups mislabel a substantial share of messages, the distilled BERT inherits the bias.
  • The deployment claim would be strengthened by reporting how the model behaves in live use, for example the rate at which providers override its routing, which the paper does not yet provide.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The manuscript describes OPTIC, a pipeline that uses GPT-4 few-shot prompting to label roughly 36K patient portal messages drawn from 405,487 MyChart encounters at Johns Hopkins Medicine, then distills those labels into a BERT classifier that separates "Admin" from "Clinical" messages. The BERT model is reported to achieve 88.85% accuracy, 88.29% sensitivity, 89.38% specificity, and F1 0.8842 on a 3,454-message test set. The paper also presents BERTopic topic modeling of the test messages and describes deployment through Epic's Nebula Cloud Platform. The central claimed contribution is a scalable, deployed triaging tool for reducing physician InBasket burden.

Significance. If the reported performance holds against physician-labeled ground truth, the work would be a useful practical contribution: the dataset is substantially larger than those used in prior portal-message classification studies, and the integration with Epic's Nebula Cloud Platform is an unusual demonstration of real-world deployment. The physician review of the GPT-4 prompt-selection set (Table 4, E2: 0.99 accuracy) is a genuine strength, as is the explicit reporting of sensitivity, specificity, and F1. However, the headline evaluation is against GPT-4-generated labels, not human ground truth, and the deployment claims lack outcome metrics; both need to be addressed before the practical-effectiveness conclusion is justified.

major comments (3)
  1. [Results, first paragraph; Abstract] The reported headline metrics (accuracy 88.85%, sensitivity 88.29%, specificity 89.38%, F1 0.8842) are computed against a test set whose labels were generated by the same GPT-4 pipeline used to create the training labels: 'A prompt-based, fine-tuned version of GPT4-32K was used to label the dataset for model distillation.' These numbers are therefore agreement rates between BERT and GPT-4, not classification accuracy against an independent ground truth. The only physician validation reported (Table 4, E2) was performed on a separate 2,000-message set used for prompt selection, and the manuscript does not state that the 3,454 test labels were physician-reviewed. Because systematic GPT-4 labeling errors on ambiguous messages would be inherited by BERT and then scored as correct, the central '88.85% accuracy' claim is not yet supported. Please provide a physician-labeled test sample with reported agreement statistics, or explicitly reframe the claims as agreement with GPT-4 labels and temper the corresponding effectiveness statements.
  2. [Retrospective Group Analysis and Cohort Selection] The 'Possible Administrative' and 'Possible Clinical' groups are defined by metadata (EMP/SER flags and Order/Note linkage) rather than by message content. Because the GPT-4 labeling and BERT distillation training pools are sampled from these groups, any systematic disagreement between the metadata-based categories and the true content-based categories will be encoded in the model. Furthermore, the 292,291 'Uncategorized' messages are excluded from training and from the reported evaluation, yet a deployed triaging system will encounter exactly such messages. The paper should report the GPT-4 label distribution within each retrospective group, quantify agreement between metadata categories and GPT-4 labels on a physician-reviewed sample, and evaluate the BERT model on a sample from the Uncategorized group; without these, the generalization claim to the full patient-message population is unsupported.
  3. [Abstract; Conclusion] The claim that deployment through Epic's Nebula Cloud Platform 'demonstrat[es] its practical effectiveness in healthcare settings' is not supported by any deployment outcome data in the manuscript. No metrics are reported on message volume processed, provider time saved, physician satisfaction, or comparison with the pre-deployment workflow. The deployment is a valuable feasibility statement, but the effectiveness claim requires either quantitative deployment evaluation or a more modest formulation.
minor comments (5)
  1. [Model distillation; Results] The text says 'For the validation dataset, we used the same 2,000 messages which were used for GPT prompt validation,' but the Results split reports a 3,387-message validation set; additionally, '~36k messages' is inconsistent with 33,861 + 3,387 + 3,454 = 40,702. Please clarify the dataset composition and the role of the 2,000 prompt-selection messages in the BERT validation.
  2. [Results, BERTopic analysis] The claim that the model achieves over 80% accuracy for 58 of 81 topics lacks information on topic sizes or confidence intervals; please include the number of messages per topic and the accuracy distribution.
  3. [GPT Prompt Analysis and Model Distillation] The manuscript does not report the BERT model variant, fine-tuning hyperparameters, or checkpoint selection procedure; for reproducibility, please provide these details and a data/code availability statement.
  4. [Figure 4] The clustering figures for Administrative and Clinical groups are referenced without axis labels or a description of the embedding projection, making them difficult to interpret.
  5. [References] Reference [23] appears to have an incomplete or incorrect author list; please verify the entry against the original publication.

Circularity Check

2 steps flagged · score 6.0 of 10

The headline BERT accuracy is computed against GPT-4-generated test labels after training on GPT-4 labels, so the central claim reduces to teacher-student agreement rather than independent task accuracy.

  1. fitted input called prediction [Abstract (Results) and Results, 'Model distillation' paragraph]
    "Abstract: 'The BERT model achieved an accuracy of 88.85% on the test set derived from validated GPT-4 labeling.' Results: 'A prompt-based, fine-tuned version of GPT4-32K was used to label the dataset for model distillation.'"

    The test-set labels are generated by the same GPT-4 labeling process that produced the training labels for BERT distillation. BERT was trained to reproduce GPT-4's classifications, so the reported sensitivity, specificity, and F1 measure BERT's agreement with GPT-4, not accuracy against physician ground truth. The only physician evaluation reported (Table 4, E2 on 2,000 messages) validated prompt selection, not the 3,454 test-set labels. The headline metric therefore reduces, by construction, to a teacher-student agreement score.

  2. fitted input called prediction [Results, BERTopic analysis paragraph]
    "BERTopic was employed to extract topics from the test set. This unsupervised topic modelling technique helped identify 81 distinct topics within the test data... Out of the 81 identified topics, the model achieved over 80% accuracy for 58 topics."

    This topic-wise 'accuracy' is computed on the same GPT-4-labeled test set, so it inherits the same reduction: the numbers quantify how often BERT matches GPT-4's topic-level labels, not how often it matches an independently verified human classification. No additional human validation of these test labels is reported.

full rationale

The paper's central quantitative claim, 88.85% accuracy, is an agreement rate between BERT and GPT-4 labels: BERT is trained on GPT-4-generated labels and then evaluated on a test set labeled by the same GPT-4 pipeline. The only human validation is the 2,000-message set used for prompt selection, where three physicians judged GPT-4 predictions at 0.99 accuracy (Table 4, E2); that is genuine independent evidence that the teacher is accurate on that balanced selection set, but it does not validate the 3,454 test-set labels. Consequently, the reported sensitivity, specificity, and F1 are distillation-fidelity measures rather than task accuracy against physician ground truth. The BERTopic topic-wise accuracy suffers from the same issue. No self-citation chain is load-bearing; the circularity is entirely in the evaluation methodology. Because the teacher was partially human-validated but the headline metric remains self-referential, the circularity is partial, not total.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The central result rests on two unevidenced pillars. First, the metadata-based retrospective grouping (sender flagged 'EMP' without a clinician service role, or message linked to an Order/Note) is treated as a valid weak label for Administrative and Clinical categories; all downstream sampling inherits this assumption. Second, GPT-4 labels are treated as ground truth for the full labeled pool, with the 2,000-message physician validation (0.99 agreement) extrapolated to the 3,454-message test set without direct human review. The Admin/Clinical dichotomy itself is acknowledged in the Introduction as complicated by messages mixing both elements, yet no labeling hierarchy is defined. Hand-chosen design choices shape every reported number: 200 few-shot examples (selected via E1-E4 prompt comparison), a 50/50 class balance, unstated BERT fine-tuning hyperparameters, and unstated BERTopic parameters. No invented entities are introduced; the contribution is a trained classifier plus a deployment package.

free parameters (4)
  • Few-shot example count for GPT-4 labeling prompt = 200 (prompt experiment E2)
    Comparison of E1 (10 examples, 0.89 accuracy), E2 (200 examples, 0.99), E3 (120 examples on GPT-3.5 Turbo, 0.85), and E4 (zero-shot, 0.65) on 2,000 physician-validated messages selected the deployed prompt configuration; the BERT distillation inherits this choice.
  • Class balance in labeled pool and evaluation splits = 50/50 Admin/Clinical
    The text states the labeled dataset had 'equal proportions' and the validation set was 'evenly distributed'; sensitivity and specificity are therefore conditional on an artificial prevalence, not the real-world mix.
  • BERT fine-tuning hyperparameters and checkpoint = not reported
    The paper never states which BERT variant (BERT vs BioBERT vs ClinicalBERT), learning rate, epochs, batch size, or tokenizer was used; these choices directly determine the reported metrics.
  • BERTopic hyperparameters = not reported
    min_topic_size, UMAP/embedding settings, and the resulting 81-topic granularity are unstated; the per-topic accuracy analysis depends on these choices.
assumptions (4)
  • domain assumption The retrospective metadata grouping (sender flagged EMP without clinician SER, or message linked to an Order or Note activity) is a valid weak proxy for Administrative and Clinical categories.
    Constructs the 'Possible Administrative' (44,926) and 'Possible Clinical' (68,270) pools from which all training, validation, and test messages were sampled (Data and Cohort Selection).
  • domain assumption GPT-4 labels are valid ground truth for the full labeled pool, including the 3,454-message test set, based on 0.99 agreement with three physicians on a separate 2,000-message set.
    The physician validation was used for prompt selection; the paper does not show direct human review of training or test labels (GPT prompt validation and RESULTS).
  • domain assumption Admin versus Clinical is a well-defined dichotomy.
    The Introduction acknowledges 'many messages contain elements of both clinical needs and administrative requests,' yet no labeling hierarchy or mixed-message rule is specified; boundary cases are resolved implicitly by GPT-4 examples.
  • domain assumption The January to June 2020 primary care message distribution matches the current production deployment distribution.
    No distribution shift analysis or external validation is presented; the deployment claim assumes 2020-trained performance transfers to live Epic Nebula traffic (CONCLUSION).

how reviews work

0 comments
Cite this review

Pith. "Pith review of OPTIC: Optimizing Patient-Provider Triaging & Improving Communications in Clinical Operations using GPT-4 Data Labeling and Model Distillation." pith.science (2026). https://pith.science/paper/Q4IU2URF

@misc{pith2026250305701,
  author       = {Pith},
  title        = {Pith review of: OPTIC: Optimizing Patient-Provider Triaging & Improving Communications in Clinical Operations using GPT-4 Data Labeling and Model Distillation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Q4IU2URF}},
  note         = {Machine review of arXiv:2503.05701}
}
read the original abstract

The COVID-19 pandemic has accelerated the adoption of telemedicine and patient messaging through electronic medical portals (patient medical advice requests, or PMARs). While these platforms enhance patient access to healthcare, they have also increased the burden on healthcare providers due to the surge in PMARs. This study seeks to develop an efficient tool for message triaging to reduce physician workload and improve patient-provider communication. We developed OPTIC (Optimizing Patient-Provider Triaging & Improving Communications in Clinical Operations), a powerful message triaging tool that utilizes GPT-4 for data labeling and BERT for model distillation. The study used a dataset of 405,487 patient messaging encounters from Johns Hopkins Medicine between January and June 2020. High-quality labeled data was generated through GPT-4-based prompt engineering, which was then used to train a BERT model to classify messages as "Admin" or "Clinical." The BERT model achieved 88.85% accuracy on the test set validated by GPT-4 labeling, with a sensitivity of 88.29%, specificity of 89.38%, and an F1 score of 0.8842. BERTopic analysis identified 81 distinct topics within the test data, with over 80% accuracy in classifying 58 topics. The system was successfully deployed through Epic's Nebula Cloud Platform, demonstrating its practical effectiveness in healthcare settings.

Figures

Figures reproduced from arXiv: 2503.05701 by the authors.

Figure 1
Figure 1. Overview of our Optimizing Patient-Provider Triaging & Improving Communications in Clinical Operations (OPTIC) system. Telemedicine and asynchronous electronic care have become and will likely continue to be essential tools for providing healthcare service delivery. On average, physicians receive 40 InBasket messages per day through MyChart, with some receiving up to 150, leading to considerable time spent on [PITH… view at source ↗
Figure 2
Figure 2. Overview of the proposed triaging system. Data Source An overview of data cleaning and processing is presented in [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 4
Figure 4. Clustering for Administrative(above) and Clinical(below) retrospective groups [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figures from the paper (1 more)
Figure 6
Figure 6. Figure 6: Hierarchical clustering for Admin retrospective group [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

26 extracted references · 22 canonical work pages

  1. [1]

    https://mychart.org/

    work page

  2. [2]

    Assessing the impact of the COVID -19 pandemic on clinician ambulatory electronic health record use

    Holmgren A, Downing N, Tang M, et al. Assessing the impact of the COVID -19 pandemic on clinician ambulatory electronic health record use. J Am Med Inform Assoc. 2022;29(3):453-460. doi: 10.1093/jamia/ocab268

    work page doi:10.1093/jamia/ocab268 2022

  3. [3]

    How We’re Improving Physicians’ Messaging Experience Through Digital Tools,

    N. L. H. T. Hub, “How We’re Improving Physicians’ Messaging Experience Through Digital Tools,” NYU Langone Health Tech Hub. Accessed: Nov. 15, 2023. [Online]. Available: https://medium.com/nyu-langones-health-tech-hub/how-were-improving-physicians- messaging-experience-through-digital-tools-1c0abd8e711b

    work page 2023

  4. [4]

    The Rules of Engagement: Perspectives on Secure Messaging From Experienced Ambulatory Patient Portal Users

    Sieck C, Hefner J, Schnierle J, et al. The Rules of Engagement: Perspectives on Secure Messaging From Experienced Ambulatory Patient Portal Users. JMIR Med Inform , vol. 5, no. 3, p. e13, Jul. 2017, doi: 10.2196/medinform.7516

    work page doi:10.2196/medinform.7516 2017

  5. [5]

    Increasing Patient Portal Usage: Preliminary Outcomes from the MyChart Genius Project

    Ramsey A, Lanzo E, Huston -Paterson H, et al. Increasing Patient Portal Usage: Preliminary Outcomes from the MyChart Genius Project. J Adolesc Health, vol. 62, no. 1, pp. 29–35, Jan. 2018, doi: 10.1016/j.jadohealth.2017.08.029

    work page doi:10.1016/j.jadohealth.2017.08.029 2018

  6. [6]

    Exploring how virtual primary care visits affect patient burden of treatment

    Kelley L, Phung M, Stamenova V, et al. Exploring how virtual primary care visits affect patient burden of treatment. International Journal of Medical Informatics, vol. 141, p. 104228, Sep. 2020, doi: 10.1016/j.ijmedinf.2020.104228

    work page arXiv 2020

  7. [7]

    Physician burnout: contributors, consequences and solutions

    West C, Dyrbye L, and Shanafelt T. Physician burnout: contributors, consequences and solutions. Journal of Internal Medicine , vol. 283, no. 6, pp. 516 –529, 2018, doi: 10.1111/joim.12752

    work page doi:10.1111/joim.12752 2018

  8. [8]

    Are specific elements of electronic health record use associated with clinician burnout more than others? J Am Med Inform Assoc

    Hilliard RW, Haskell J, Gardner RL. Are specific elements of electronic health record use associated with clinician burnout more than others? J Am Med Inform Assoc . 2020 Jul 1;27(9):1401-1410. doi: 10.1093/jamia/ocaa092. PMID: 32719859; PMCID: PMC7647296

    work page doi:10.1093/jamia/ocaa092 2020

Show all 26 references
  1. [9]

    MyChart Messages the Wild West of Patient Communication,

    “MyChart Messages the Wild West of Patient Communication,” 33 Charts. Accessed: Mar. 04,

  2. [10]

    Impact of MyChart Communication on Provider Burden

    Schuetz S. Impact of MyChart Communication on Provider Burden. Family Medicine Clerkship Student Projects, Jan. 2021, [Online]. Available: https://scholarworks.uvm.edu/fmclerk/693

    2021

  3. [11]

    Medical emergency triage and patient prioritisation in a telemedicine environment: a systematic review

    Napi NM, Zaidan AA, Zaidan BB, et al . Medical emergency triage and patient prioritisation in a telemedicine environment: a systematic review. Health Technol., vol. 9, no. 5, pp. 679–700, Nov. 2019, doi: 10.1007/s12553-019-00357-w

    2019 doi

  4. [12]

    Devlin J, Chang MW, Lee K. et al. Bert: Pre-training of deep bidirectional transformers for language understanding. arXiv preprint arXiv:1810.04805 (2018)

    2018 arXiv

  5. [13]

    A comparison of rule -based and machine learning approaches for classifying patient portal messages

    Cronin RM, Fabbri D, Denny JC, et al. A comparison of rule -based and machine learning approaches for classifying patient portal messages. International Journal of Medical Informatics, vol. 105, pp. 110–120, Sep. 2017, doi: 10.1016/j.ijmedinf.2017.06.004

    2017 doi

  6. [14]

    Medical Dialogue Summarization for Automated Reporting in Healthcare

    Molenaar S, Maas L, Burriel V, et al. Medical Dialogue Summarization for Automated Reporting in Healthcare. Advanced Information Systems Engineering Workshops, S. Dupuy-Chessa and H. A. Proper, Eds., in Lecture Notes in Business Information Processing. Cham: Springer Internati...

    2020 doi

  7. [15]

    Joshi A, Katariya N, Amatriain X, et al. Dr. Summarize: Global Summarization of Medical Dialogue by Exploiting Local Structures. arXiv:2009.08666 [cs] , Sep. 2020, Accessed: Mar. 05, 2022. [Online]. Available: http://arxiv.org/abs/2009.08666

    2009 arXiv

  8. [16]

    Medically Aware GPT -3 as a Data Generator for Medical Dialogue Summarization

    Chintagunta B, Katariya N, Amatriain X, et al . Medically Aware GPT -3 as a Data Generator for Medical Dialogue Summarization. Proceedings of the 6th Machine Learning for Healthcare Conference, PMLR, Oct. 2021, pp. 354 –372. Accessed: Mar. 05, 2022. [Online]. Available: https:...

    2021

  9. [17]

    More Data Key to Understand AI for Sorting Patient Portal Messages,

    Patient Engagement HIT, “More Data Key to Understand AI for Sorting Patient Portal Messages,”. Accessed: Mar. 07, 2023. [Online]. Available: https://patientengagementhit.com/news/more - data-key-to-understand-ai-for-sorting-patient-portal-messages

    2023

  10. [18]

    Classifying patient portal messages using Convolutional Neural Networks

    Sulieman L, Glimore D, French C, et al. Classifying patient portal messages using Convolutional Neural Networks. Journal of Biomedical Informatics , vol. 74, pp. 59 –70, Oct. 2017, doi: 10.1016/j.jbi.2017.08.014

    2017 doi

  11. [19]

    Artificial Intelligence to Identify Commonly Asked Questions via an Electronic Patient Portal - Lessons From a Cardiology Department Within a Large Health System [abstract]

    Wosik J, Si S, Henao R, et al. Artificial Intelligence to Identify Commonly Asked Questions via an Electronic Patient Portal - Lessons From a Cardiology Department Within a Large Health System [abstract]. Circulation, vol. 140, no. Suppl_1, pp. A14805–A14805, Nov. 2019, doi: 1...

    2019 doi

  12. [20]

    Rapid design and implementation of an integrated patient self -triage and self -scheduling tool for COVID -19

    Judson TJ, Odisho AY, Neinstein AB, et al. Rapid design and implementation of an integrated patient self -triage and self -scheduling tool for COVID -19. Journal of the American Medical Informatics Association, vol. 27, no. 6, pp. 860–866, Jun. 2020, doi: 10.1093/jamia/ocaa051

    2020 doi

  13. [21]

    Analysis of Patient Contacts with the Radiation Oncology Triage Nurse: The Experience of a Single Center

    Weber BW, Blitzer GC, Anderson BM, et al. Analysis of Patient Contacts with the Radiation Oncology Triage Nurse: The Experience of a Single Center. International Journal of Radiation Oncology, Biology, Physics, vol. 114, no. 1, p. e25, Sep. 2022, doi: 10.1016/j.ijrobp.2022.06.087

    2022 doi

  14. [22]

    Analyzing Patient Secure Messages Using a Fast Health Care Interoperability Resources (FIHR) –Based Data Model: Development and Topic Modeling Study

    De A, Huang M, Feng T, et al. Analyzing Patient Secure Messages Using a Fast Health Care Interoperability Resources (FIHR) –Based Data Model: Development and Topic Modeling Study. Journal of Medical Internet Research, vol. 23, no. 7, p. e26770, Jul. 2021, doi: 10.2196/26770

    2021 doi

  15. [23]

    Artificial intelligence to organize patient portal messages: a journey from an ensemble deep learning text classification to rule-based named entity recognition

    Ahmed PT, Sunyang F, Aditya K, et al. Artificial intelligence to organize patient portal messages: a journey from an ensemble deep learning text classification to rule-based named entity recognition. IEEE International Conference on Bioinformatics and Biomedicine (BIBM), pp. 1...

    2021

  16. [24]

    Medical Call Center Software, Nurse Triage on Call,

    “Medical Call Center Software, Nurse Triage on Call,” TriageLogic: Remote Nurse Triage Software and Services. Accessed: Mar. 07, 2023. [Online]. Available: https://triagelogic.com/

    2023

  17. [25]

    BERTopic: Neural topic modeling with a class -based TF-IDF procedure

    Grootendorst M. BERTopic: Neural topic modeling with a class -based TF-IDF procedure. arXiv, Mar. 11, 2022. doi: 10.48550/arXiv.2203.05794

    2022 doi arXiv

  18. [2022]

    Available: https://33charts.com/mychart -messages-the-wild-west-of-patient- communication/

    [Online]. Available: https://33charts.com/mychart -messages-the-wild-west-of-patient- communication/

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.