Pith. sign in

REVIEW 4 major objections 6 minor 56 references

PT: A Plain Transformer is Good Hospital Readmission Predictor

T0 review · 4 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read A plain Transformer with separate per-modality blocks and simple training add-ons predicts 30-day hospital readmissions with an area-under-the-ROC-curve of 0.896, beating LSTM and GRU at comparable parameter counts and runtime.

desk verdict A useful baseline idea whose reported AUC is probably optimistic—split unspecified, hyperparameters tuned on the reported metric; needs heavy revision before the numbers can be trusted. read the letter →

arxiv 2412.12909 v1 pith:PPS2RBBA submitted 2024-12-17 cs.LG

Zhenyi Fan , Jiaqi Li , Dongyu Luo , Yuqi Yuan This is my paper · ORCID
classification cs.LG
keywords hospitalreadmissionpredictiontransformermultimodallearningelectronichealthrecordsclinicalnoteschestradiographsAUCclassification
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

PT is a plain Transformer for 30-day hospital readmission prediction that combines electronic health records, chest radiographs, and clinical notes. The paper's central claim is that this simple per-modality Transformer architecture beats recurrent baselines like LSTM and GRU at the same parameter and runtime budget, reaching an area-under-the-ROC-curve (AUC) of 0.896 on EHR plus notes. The model is also claimed to stay accurate when only clinical notes or only EHR data are used, and when temporal information is missing or unclear. A strong, simple baseline matters because hospitals can deploy and maintain it more easily, and researchers can compare future models against a meaningful lower bound.

What carries the argument

The load-bearing mechanism is a per-modality Transformer encoder: each data type gets its own self-attention block with its own parameters, with positional encodings preserving sequence order, followed by attention pooling to collapse the sequence into one vector, concatenation of the modality vectors, and an MLP for the final probability. Feature preparation uses random-forest selection to keep the top-$k$ EHR variables, a self-supervised chest X-ray encoder for radiographs, and TF-IDF vectors for clinical notes. Training adds label-smoothing focal loss, a cosine learning-rate schedule, dynamic Gaussian noise, and a 10-fold test-time ensemble, and the paper reports these components individually raise the headline AUC.

What would settle it

Re-run the EHR+Notes experiment with the random-forest feature selector and the noise-scheduler search nested inside each training fold, then apply the final model once to a held-out set of admissions; if the AUC falls to the 0.878–0.880 range of the recurrent baselines, the claimed advantage is leakage rather than architecture.

Watch

Extended reading notes

Core claim

The paper's central claim is that a plain Transformer—one self-attention block per modality with its own parameters—can outperform recurrent sequence models for 30-day all-cause readmission without requiring explicit timestamps. Each modality is fed in as an ordered sequence: hospital-day rows for EHR, radiographs in recorded order, and notes in note order. After attention pooling and concatenation, an MLP produces the readmission probability. On 13,763 admissions the paper reports an area-under-the-ROC-curve (AUC) of 0.896 for EHR plus notes, against 0.878 for LSTM and 0.880 for GRU with comparable parameter counts and runtime, and notes-only and EHR-only configurations reach 0.832 and 0.761. The authors attribute the outcome to treating each modality as a sequence rather than requiring timestamped events, supported by random-forest feature selection, dynamic noise, and a 10-fold ensemble.

Load-bearing premise

The headline results depend on an evaluation in which every tuning choice—the random-forest feature count and the noise-scheduling parameters—is made inside the training folds and never sees the held-out admissions; if that separation is breached, the reported AUC would not reflect real-world generalization.

Editorial extensions

If this is right

  • Recurrent models LSTM and GRU are not required to reach strong readmission AUC; PT matches their runtime and parameter count and reports higher AUC.
  • Institutions without imaging data can still deploy the model: notes alone give AUC 0.832 and EHR alone gives 0.761.
  • The approach works when explicit timestamps are missing or unclear, because each modality is treated as an ordered sequence rather than requiring event times.
  • The reported gains come from a bundle of training techniques—feature selection, dynamic noise, and ensembling—not from the architecture alone; ablations show each component contributes.
  • Because the design is modular, adding or dropping a modality only requires adding or removing a Transformer block, supporting the scalability claim.

Reading between the lines

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

  • A fully nested validation protocol—fitting the random-forest selector and noise schedule only on training folds—would make the 0.896 figure directly comparable to models tuned the same way; if the gap persists, plain Transformers should become the default baseline in clinical readmission papers.
  • The large gap between notes-only and EHR-only performance suggests that free-text notes carry more predictive signal than the structured features used here; replacing TF-IDF with a modern clinical language model might push the single-modality ceiling higher.
  • The drop when chest radiographs are added (0.896 to 0.881) is worth probing: if the images are noisy or misaligned with the admission, simple late fusion may dilute the EHR+notes signal, and gating or weighted fusion could recover the loss.
  • A clean test of the missing-timestamps claim would be to strip all timestamps and compare PT against a recurrent model that requires event times; the paper's assumptions imply PT should lose little or nothing.
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

4 major / 6 minor

Summary. The paper presents PT, a Transformer-based model for predicting 30-day hospital readmissions from three modalities: EHR tabular data, chest radiographs, and clinical notes. The model extracts features per modality, processes each with a separate Transformer block, pools with attention, concatenates, and passes through an MLP. Additional components include Random Forest feature selection for EHR, dynamic noise scheduling, label smoothing focal loss, and a K-fold ensemble. On a MIMIC-IV-derived dataset of 13,763 admissions, PT reports an AUC of 0.896 for EHR+Notes, compared to 0.878 for LSTM and 0.880 for GRU; ablations explore noise schedulers, feature counts, loss functions, and ensemble sizes. The paper claims superior accuracy, scalability, and robustness, including when temporal information is missing.

Significance. If the reported results are reliable, PT would provide a simple, competitive Transformer baseline for multimodal readmission prediction, and the modular design could be useful for future benchmarks. The paper's strengths are its use of a public dataset (MIMIC-IV/MIMIC-CXR) and the transparency of the architecture description. However, the evaluation protocol is currently under-specified: no patient-level split is documented, hyperparameters appear to be selected on the reported AUC, strong cited baselines are not compared, and the robustness claim is not directly tested. These issues prevent the reader from assessing whether the 0.896 AUC reflects true generalization. No code or repeated-run statistics are provided, so the precision of the comparison is unknown.

major comments (4)
  1. [Section 4.1 and all experiments] The paper reports 13,763 admissions from 11,041 unique patients but never states whether the train/test split is by admission or by patient. If the split is by admission, the same patient can appear in both training and test sets, and the reported AUCs in Tables 1 and 2 would be inflated by within-patient correlation. This is load-bearing because the central claim of superiority depends on the integrity of the evaluation. Please specify the split granularity and, if it is admission-level, re-run the experiments with a patient-level split.
  2. [Section 3.4 and Tables 3, 4, 6, 7] The Random Forest feature count k, the noise-scheduler parameters, and the K-fold value are selected by comparing AUC values, yet the paper does not describe a nested validation procedure or a held-out test set that is distinct from the selection process. If the AUCs in Tables 3, 4, 6, and 7 come from the same split as the headline result in Table 1, then the reported 0.896 is a selection artifact. Moreover, Table 3 reports a configuration with AUC 0.900 and Table 7 reports K=10 with AUC 0.9014, both higher than the headline 0.896, so the paper must clarify which configuration yields the headline and how the selection was shielded from test-set information.
  3. [Section 4.3 and Related Work] The abstract claims PT 'outperforms existing ones,' but the only baselines presented are LSTM and GRU. The cited state-of-the-art models MuST (Ref. [33]) and ClinicalBERT (Ref. [3]) are not included in any experiment, so the claim of superiority over existing multimodal or text-based methods is not supported. The paper should either add these or comparable baselines, or restrict the claim to the recurrent architectures actually evaluated. In addition, all AUCs are reported as point estimates from what appears to be a single run; without error bars or significance tests, the differences (0.896 vs. 0.878) may not be statistically meaningful.
  4. [Section 1 and Table 2] The robustness claim—that PT 'performs well even when temporal information is missing'—is never tested experimentally. The paper does not define which temporal information is missing, nor does it compare against a variant with complete or shuffled temporal order. Furthermore, Table 2 shows that adding images to EHR+Notes decreases AUC from 0.896 to 0.881, which contradicts the intuition that more modalities help; the paper should explain this non-monotonicity and directly test the robustness claim by ablating temporal information.
minor comments (6)
  1. [Section 3.1] The CXR representation is defined as Rq×1024, but the notes section later says the feature dimension is '1024 or 4096 depending on the extraction method'; please clarify which dimensions are used in the experiments.
  2. [Section 3.3] The focal loss equation defines pt = σ(z), but Algorithm 1 computes pt = exp(-BCE_loss); these are not equivalent, so the pseudo-code and the equation should be reconciled.
  3. [Section 4.2] Hyperparameters such as embedding dimension, feed-forward size, dropout, and batch size are not reported; please provide a complete configuration to enable reproducibility.
  4. [Section 4.3] The description 'identical design configurations' for LSTM and GRU is vague; please state exactly what attention and positional encoding were applied to the recurrent baselines.
  5. [Section 2] The phrase 'a innegligible gap' contains a typo and should read 'a non-negligible gap.'
  6. [Table 2] The running time for EHR+Note+Image (12.5 s/epoch) is much higher than others; if the image features are pre-extracted rather than trained end-to-end, this should be clarified in the experimental setup.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the reported AUCs are empirical comparisons, and the hyperparameter-selection concerns are evaluation-protocol risks rather than reductions of outputs to inputs.

full rationale

This paper makes an empirical claim: a plain Transformer with separate blocks for EHR, chest radiographs, and clinical notes achieves a certain AUC on a readmission benchmark. There is no derivation chain in which a predicted quantity is defined in terms of the very result it is supposed to explain. The Random Forest feature selector (Section 3.4) is described as fitting on Xtrain and ytrain and then selecting the top k features by importance; selecting features by importance is not equivalent, by construction, to predicting readmission on the evaluated admissions. Likewise, the dynamic-noise scheduler and K-fold ensemble are hyperparameters whose values are compared by AUC in Tables 3, 4, 6, and 7; tuning a hyperparameter by an evaluation metric is a form of model selection, not a circular derivation, and the reported headline AUC of 0.896 is not even the maximum value shown in those tables (Table 3 shows 0.900 and Table 7 shows 0.9014). This internal inconsistency is a legitimate correctness and evaluation-protocol concern, but it does not make any reported result equal to its inputs by definition. The paper also does not rely on load-bearing self-citations: the cited methods (Transformer, Random Forest, TF-IDF, MoCo-CXR, MIMIC datasets, prior readmission models) are external, and no cited uniqueness theorem or ansatz from the present authors is invoked to force a choice. The absence of a stated patient-level split and the possibility that hyperparameters were selected on the same split used for the headline result are threats to external validity, not evidence that the derivation is circular. Under the hard rule that circularity must be exhibited by a specific reduction, no such reduction appears in this manuscript.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central claim rests on empirical hyperparameter choices (RF feature count, noise schedule, K-fold value, loss coefficients) and on assumptions about the data and evaluation protocol. No new physical or conceptual entities are introduced. The most audit-critical items are the undisclosed loss coefficients and the lack of evidence that feature selection and noise tuning are confined to training folds.

free parameters (5)
  • Random Forest feature count k = 100
    Selected by AUC on validation in Fig. 2a; changes AUC from 0.874 at k=10 to 0.896 at k=100.
  • Linear noise scheduler initial/final ratios = 0.01 / 0.1
    Chosen by comparing AUC in Table 3; best configuration reported with AUC 0.900.
  • Sinusoidal noise scheduler amplitude and intercept = 0.05 / 0
    Best of four configurations in Table 4, reported with AUC 0.897.
  • Label smoothing factor and focal loss alpha/gamma = not reported
    Loss hyperparameters in Section 3.3; values are not disclosed but affect training and results.
  • K-fold ensemble K = 10
    Chosen by AUC comparison in Fig. 2b and Table 7; K=10 gives AUC 0.9014.
assumptions (4)
  • domain assumption The dataset of 13,763 admissions from MIMIC-IV, MIMIC-CXR-JPG, and MIMIC-IV-Note is a valid cohort for 30-day all-cause readmission, including deaths as readmissions.
    Section 4.1 states 2,379 readmissions include deaths during stay; this label definition may differ from common readmission definitions and affects the target.
  • domain assumption Random Forest feature selection and dynamic noise use only training-fold statistics.
    Section 3.4 describes these steps without specifying fold-local application; leakage would inflate AUC.
  • standard math Standard supervised learning assumptions hold: IID train/test splits and no distribution shift across modalities.
    Sections 3.2 to 3.4 rely on standard training and evaluation assumptions.
  • domain assumption MoCo-CXR and TF-IDF features are adequate fixed representations of images and clinical notes for this task.
    Section 3.1 assumes these frozen extractors transfer to the readmission prediction setting.

how reviews work

0 comments
Cite this review

Pith. "Pith review of PT: A Plain Transformer is Good Hospital Readmission Predictor." pith.science (2026). https://pith.science/paper/PPS2RBBA

@misc{pith2026241212909,
  author       = {Pith},
  title        = {Pith review of: PT: A Plain Transformer is Good Hospital Readmission Predictor},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PPS2RBBA}},
  note         = {Machine review of arXiv:2412.12909}
}
read the original abstract

Hospital readmission prediction is critical for clinical decision support, aiming to identify patients at risk of returning within 30 days post-discharge. High readmission rates often indicate inadequate treatment or post-discharge care, making effective prediction models essential for optimizing resources and improving patient outcomes. We propose PT, a Transformer-based model that integrates Electronic Health Records (EHR), medical images, and clinical notes to predict 30-day all-cause hospital readmissions. PT extracts features from raw data and uses specialized Transformer blocks tailored to the data's complexity. Enhanced with Random Forest for EHR feature selection and test-time ensemble techniques, PT achieves superior accuracy, scalability, and robustness. It performs well even when temporal information is missing. Our main contributions are: (1)Simplicity: A powerful and efficient baseline model outperforming existing ones in prediction accuracy; (2)Scalability: Flexible handling of various features from different modalities, achieving high performance with just clinical notes or EHR data; (3)Robustness: Strong predictive performance even with missing or unclear temporal data.

Figures

Figures reproduced from arXiv: 2412.12909 by the authors.

Figure 1
Figure 1. Overview of the Plain Transformer structure. Features from EHR modality follow previous preprocessing [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Left: AUC performance of the random forest model with different numbers of selected EHR features; Right: AUC performance of the model with different k values in K-fold cross-validation. 4.4 Ablation study To further evaluate the performance of our model, we conduct a series of ablation studies. First, we analyze the impact of using different numbers of EHR features on model performance with Random Forest. As shown i… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

56 extracted references · 43 canonical work pages

  1. [33]

    MuST: Multimodal Spatiotemporal Graph-Transformer for Hospital Readmission Prediction

    Yan Miao and Lequan Yu. “MuST: Multimodal Spatiotemporal Graph-Transformer for Hospital Readmission Prediction”. In: International Conference on Medical Image Computing and Computer-Assisted Intervention. Springer. 2023, pp. 276–285

    work page 2023

  2. [3]

    Clinicalbert: Modeling clinical notes and predicting hospital readmission

    Kexin Huang, Jaan Altosaar, and Rajesh Ranganath. “Clinicalbert: Modeling clinical notes and predicting hospital readmission”. In: arXiv preprint arXiv:1904.05342 (2019)

    arXiv 2019

  3. [1]

    Predicting hospital readmission via cost-sensitive deep learning

    Haishuai Wang et al. “Predicting hospital readmission via cost-sensitive deep learning”. In: IEEE/ACM transac- tions on computational biology and bioinformatics 15.6 (2018), pp. 1968–1978

    work page 2018

  4. [2]

    Risk prediction models for hospital readmission: a systematic review

    Devan Kansagara et al. “Risk prediction models for hospital readmission: a systematic review”. In: Jama 306.15 (2011), pp. 1688–1698

    work page 2011

  5. [4]

    Reducing hospital readmission rates: current strategies and future directions

    Sunil Kripalani et al. “Reducing hospital readmission rates: current strategies and future directions”. In: Annual review of medicine 65.1 (2014), pp. 471–485

    work page 2014

  6. [5]

    Underlying reasons associated with hospital readmission following surgery in the United States

    Ryan P Merkow et al. “Underlying reasons associated with hospital readmission following surgery in the United States”. In: Jama 313.5 (2015), pp. 483–495

    work page 2015

  7. [6]

    Clinical assistant diagnosis for electronic medical record based on convolutional neural network

    Zhongliang Yang et al. “Clinical assistant diagnosis for electronic medical record based on convolutional neural network”. In: Scientific reports 8.1 (2018), p. 6329

    work page 2018

  8. [7]

    Based on hybrid CNN-SVM diabetes patients’ predictive hospital readmis- sion model

    GG Rajput and Ashvini Alashetty. “Based on hybrid CNN-SVM diabetes patients’ predictive hospital readmis- sion model”. In: Acta Biomed 94.1 (2023)

    work page 2023

Show all 56 references
  1. [8]

    Analysis and prediction of unplanned intensive care unit readmission using recurrent neural networks with long short-term memory

    Yu-Wei Lin et al. “Analysis and prediction of unplanned intensive care unit readmission using recurrent neural networks with long short-term memory”. In: PloS one 14.7 (2019), e0218942

    2019

  2. [9]

    Predicting hospital readmission for lupus patients: an RNN-LSTM-based deep-learning methodology

    Bhargava K Reddy and Dursun Delen. “Predicting hospital readmission for lupus patients: an RNN-LSTM-based deep-learning methodology”. In: Computers in biology and medicine 101 (2018), pp. 199–209

    2018

  3. [10]

    Heterogeneous patient graph embedding in readmission prediction

    Hirad Daneshva and Reza Samavi. “Heterogeneous patient graph embedding in readmission prediction”. In: (2022)

    2022

  4. [11]

    DeepNote-GNN: predicting hospital readmission using clinical notes and patient network

    Sara Nouri Golmaei and Xiao Luo. “DeepNote-GNN: predicting hospital readmission using clinical notes and patient network”. In: Proceedings of the 12th ACM Conference on Bioinformatics, Computational Biology, and Health Informatics. 2021, pp. 1–9

    2021

  5. [12]

    Predicting 30-day hospital readmission in medicare patients: Insights from an lstm deep learning model

    Xintao Li and Sibei Liu. “Predicting 30-day hospital readmission in medicare patients: Insights from an lstm deep learning model”. In: medRxiv (2024), pp. 2024–09

    2024

  6. [13]

    Predicting heart failure readmission from clinical notes using deep learning

    Xiong Liu et al. “Predicting heart failure readmission from clinical notes using deep learning”. In: 2019 IEEE International Conference on Bioinformatics and Biomedicine (BIBM). IEEE. 2019, pp. 2642–2648

    2019

  7. [14]

    Jeremiah R Brown et al. “Information extraction from electronic health records to predict readmission following acute myocardial infarction: does natural language processing using clinical notes improve prediction of readmission?” In: Journal of the American Heart Association ...

    2022

  8. [15]

    MIMIC-III, a freely accessible critical care database

    Alistair EW Johnson et al. “MIMIC-III, a freely accessible critical care database”. In: Scientific data 3.1 (2016), pp. 1–9

    2016

  9. [16]

    Multimodal spatiotemporal graph neural networks for improved prediction of 30-day all-cause hospital readmission

    Siyi Tang et al. “Multimodal spatiotemporal graph neural networks for improved prediction of 30-day all-cause hospital readmission”. In: arXiv preprint arXiv:2204.06766 (2022)

    2022 arXiv

  10. [17]

    Moco pretraining improves representation and transferability of chest x-ray models

    Hari Sowrirajan et al. “Moco pretraining improves representation and transferability of chest x-ray models”. In: Medical Imaging with Deep Learning. PMLR. 2021, pp. 728–744

    2021

  11. [18]

    Text mining: use of TF-IDF to examine the relevance of words to documents

    Shahzad Qaiser and Ramsha Ali. “Text mining: use of TF-IDF to examine the relevance of words to documents”. In: International Journal of Computer Applications 181.1 (2018), pp. 25–29

    2018

  12. [19]

    Using tf-idf to determine word relevance in document queries

    Juan Ramos et al. “Using tf-idf to determine word relevance in document queries”. In: Proceedings of the first instructional conference on machine learning. V ol. 242. 1. Citeseer. 2003, pp. 29–48

    2003

  13. [20]

    Attention is all you need

    Ashish Vaswani et al. “Attention is all you need”. In:Advances in neural information processing systems. V ol. 30. 2017

    2017

  14. [21]

    Overview of the Transformer-based Models for NLP Tasks

    Anthony Gillioz et al. “Overview of the Transformer-based Models for NLP Tasks”. In: 2020 15th Conference on computer science and information systems (FedCSIS). IEEE. 2020, pp. 179–183

    2020

  15. [22]

    Transformers: State-of-the-art natural language processing

    Thomas Wolf et al. “Transformers: State-of-the-art natural language processing”. In:Proceedings of the 2020 conference on empirical methods in natural language processing: system demonstrations. 2020, pp. 38–45

    2020

  16. [23]

    Gpt-4 technical report

    Josh Achiam et al. “Gpt-4 technical report”. In: arXiv preprint arXiv:2303.08774 (2023)

    2023 arXiv

  17. [24]

    Qwen technical report

    Jinze Bai et al. “Qwen technical report”. In: arXiv preprint arXiv:2309.16609 (2023)

    2023 arXiv

  18. [25]

    An image is worth 16x16 words: Transformers for image recognition at scale

    Alexey Dosovitskiy. “An image is worth 16x16 words: Transformers for image recognition at scale”. In: arXiv preprint arXiv:2010.11929 (2020)

    2020 arXiv

  19. [26]

    Scalable diffusion models with transformers

    William Peebles and Saining Xie. “Scalable diffusion models with transformers”. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. 2023, pp. 4195–4205. 9

    2023

  20. [27]

    Medical transformer: Gated axial-attention for medical image segmentation

    Jeya Maria Jose Valanarasu et al. “Medical transformer: Gated axial-attention for medical image segmentation”. In: Medical image computing and computer assisted intervention–MICCAI 2021: 24th international conference, Strasbourg, France, September 27–October 1, 2021, proceedin...

    2021

  21. [28]

    Unetr: Transformers for 3d medical image segmentation

    Ali Hatamizadeh et al. “Unetr: Transformers for 3d medical image segmentation”. In: Proceedings of the IEEE/CVF winter conference on applications of computer vision. 2022, pp. 574–584

    2022

  22. [29]

    Multi-modal transformer for video retrieval

    Valentin Gabeur et al. “Multi-modal transformer for video retrieval”. In:Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part IV 16. Springer. 2020, pp. 214–229

    2020

  23. [30]

    Attention bottlenecks for multimodal fusion

    Arsha Nagrani et al. “Attention bottlenecks for multimodal fusion”. In: Advances in neural information process- ing systems 34 (2021), pp. 14200–14213

    2021

  24. [31]

    Multi-modal fusion transformer for end-to-end au- tonomous driving

    Aditya Prakash, Kashyap Chitta, and Andreas Geiger. “Multi-modal fusion transformer for end-to-end au- tonomous driving”. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2021, pp. 7077–7087

    2021

  25. [32]

    Mm-vit: Multi-modal video transformer for compressed video action recogni- tion

    Jiawei Chen and Chiu Man Ho. “Mm-vit: Multi-modal video transformer for compressed video action recogni- tion”. In: Proceedings of the IEEE/CVF winter conference on applications of computer vision. 2022, pp. 1910– 1921

    2022

  26. [34]

    Deep learning approaches for electronic health records: A comprehensive review

    Jian Wang, Xiaofeng Chen, and Yong Zhang. “Deep learning approaches for electronic health records: A comprehensive review”. In: Journal of Biomedical Informatics 107 (2020), p. 103436

    2020

  27. [35]

    Deep Learning for Electronic Health Record Analytics

    Gaspard Harerimana, Jong Kim, and Beakchol Jang. “Deep Learning for Electronic Health Record Analytics”. In: IEEE Access PP (July 2019), pp. 1–1. DOI: 10.1109/ACCESS.2019.2928363

    2019

  28. [36]

    Natural Language Processing of Clinical Notes on Chronic Diseases: Systematic Review

    Seyedmostafa Sheikhalishahi et al. “Natural Language Processing of Clinical Notes on Chronic Diseases: Systematic Review”. In: JMIR Med Inform 7.2 (Apr. 2019), e12239. ISSN : 2291-9694. DOI: 10.2196/12239. URL: http://www.ncbi.nlm.nih.gov/pubmed/31066697

    2019

  29. [37]

    Automated Machine Learning for Healthcare and Clinical Notes Analysis

    Akram Mustafa and Mostafa Rahimi Azghadi. “Automated Machine Learning for Healthcare and Clinical Notes Analysis”. In: Computers 10.2 (2021). ISSN : 2073-431X. DOI: 10.3390/computers10020024. URL: https://www.mdpi.com/2073-431X/10/2/24

    2021 doi

  30. [38]

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

    Feng Xie et al. “Deep learning for temporal data representation in electronic health records: A systematic review of challenges and methodologies”. In: Journal of Biomedical Informatics 126 (2022), p. 103980. ISSN : 1532-0464. DOI: https://doi.org/10.1016/j.jbi.2021.103980. UR...

    2022

  31. [39]

    Recent Advances in Large Language Models for Health- care

    Khalid Nassiri and Moulay A. Akhloufi. “Recent Advances in Large Language Models for Health- care”. In: BioMedInformatics 4.2 (2024), pp. 1097–1143. ISSN : 2673-7426. DOI: 10 . 3390 / biomedinformatics4020062. URL: https://www.mdpi.com/2673-7426/4/2/62

    2024

  32. [40]

    Multimodal biomedical AI

    Julián N Acosta et al. “Multimodal biomedical AI”. In: Nature Medicine 28.9 (2022), pp. 1773–1784

    2022

  33. [41]

    Research progress on electronic health records multimodal data fusion based on deep learning

    Y Fan, Z Zhang, and J Wang. “Research progress on electronic health records multimodal data fusion based on deep learning”. zh. In: Sheng Wu Yi Xue Gong Cheng Xue Za Zhi 41.5 (Oct. 2024). Chinese, pp. 1062–1071. DOI: 10.7507/1001-5515.202310011

    2024

  34. [42]

    Late fusion of multimodal deep neural networks for weeds classification

    V o Hoang Trong et al. “Late fusion of multimodal deep neural networks for weeds classification”. In:Computers and Electronics in Agriculture 175 (2020), p. 105506. ISSN : 0168-1699. DOI: https : / / doi . org / 10 . 1016/j.compag.2020.105506 . URL: https://www.sciencedirect.c...

    2020

  35. [43]

    Random forests

    Leo Breiman. “Random forests”. In: Machine learning 45.1 (2001), pp. 5–32. DOI: 10 . 1023 / A : 1010933404324

    2001

  36. [44]

    Rethinking the inception architecture for computer vision

    Christian Szegedy et al. “Rethinking the inception architecture for computer vision”. In: Proceedings of the IEEE conference on computer vision and pattern recognition. 2016, pp. 2818–2826

    2016

  37. [45]

    Focal Loss for Dense Object Detection

    T Lin. “Focal Loss for Dense Object Detection”. In: arXiv preprint arXiv:1708.02002 (2017)

    2017 arXiv

  38. [46]

    Stabilizing Machine Learning Prediction of Dynamics: Noise and Noise-inspired Regularization

    Alexander Wikner et al. “Stabilizing Machine Learning Prediction of Dynamics: Noise and Noise-inspired Regularization”. In: arXiv preprint arXiv:2211.05262 (2022)

    2022 arXiv

  39. [47]

    The’K’in K-fold Cross Validation

    Davide Anguita et al. “The’K’in K-fold Cross Validation.” In: ESANN. V ol. 102. 2012, pp. 441–446

    2012

  40. [48]

    Estimation of prediction error by using K-fold cross-validation

    Tadayoshi Fushiki. “Estimation of prediction error by using K-fold cross-validation”. In:Statistics and Computing 21 (2011), pp. 137–146

    2011

  41. [49]

    [2024 FALL] STAT3612 group project

    Fuying Wang. [2024 FALL] STAT3612 group project. https://www.kaggle.com/competitions/stat- 3612-group-project-2024-fall/overview . Accessed: 2024-12-06. 2024. 10

    2024

  42. [50]

    MIMIC-IV , a freely accessible electronic health record dataset

    Alistair EW Johnson et al. “MIMIC-IV , a freely accessible electronic health record dataset”. In: Scientific data 10.1 (2023), p. 1

    2023

  43. [51]

    PhysioBank, PhysioToolkit, and PhysioNet: components of a new research resource for complex physiologic signals

    Ary L Goldberger et al. “PhysioBank, PhysioToolkit, and PhysioNet: components of a new research resource for complex physiologic signals”. In: circulation 101.23 (2000), e215–e220

    2000

  44. [52]

    MIMIC-CXR-JPG, a large publicly available database of labeled chest radiographs

    Alistair EW Johnson et al. “MIMIC-CXR-JPG, a large publicly available database of labeled chest radiographs”. In: arXiv preprint arXiv:1901.07042 (2019)

    2019 arXiv

  45. [53]

    Decoupled weight decay regularization

    I Loshchilov. “Decoupled weight decay regularization”. In: arXiv preprint arXiv:1711.05101 (2017)

    2017 arXiv

  46. [54]

    Sgdr: Stochastic gradient descent with warm restarts

    Ilya Loshchilov and Frank Hutter. “Sgdr: Stochastic gradient descent with warm restarts”. In: arXiv preprint arXiv:1608.03983 (2016)

    2016 arXiv

  47. [55]

    Long short-term memory

    Alex Graves and Alex Graves. “Long short-term memory”. In: Supervised sequence labelling with recurrent neural networks (2012), pp. 37–45

    2012

  48. [56]

    Empirical evaluation of gated recurrent neural networks on sequence modeling

    Junyoung Chung et al. “Empirical evaluation of gated recurrent neural networks on sequence modeling”. In: arXiv preprint arXiv:1412.3555 (2014). 11 A Metric In this paper, we use AUC as the metric. The AUC is the area under the ROC curve, and can be calculated as: AU C= Z ∞ −∞...

    2014 arXiv

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.