Pith. sign in

REVIEW 3 major objections 5 minor 58 references

Bi-Axial Transformers: Addressing the Increasing Complexity of EHR Classification

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

Pith's one-line read Bi-Axial Transformer reaches state-of-the-art sepsis prediction

desk verdict A clear architecture paper for EHR classification whose headline sepsis result depends on an ambiguous hyperparameter-selection protocol that must be resolved before the SOTA claim can be trusted. read the letter →

arxiv 2508.12418 v1 pith:67PKAH3A submitted 2025-08-17 cs.LG

classification cs.LG
keywords ElectronicHealthRecordsPredictiveModelsTimeSeriesAnalysisAxialAttentionTransformerMissingDataSepsisPredictionTransferLearning
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

This paper introduces the Bi-Axial Transformer (BAT), a classifier that applies attention along both the time axis and the sensor axis of electronic health record data instead of attending along a single flattened axis. The paper claims BAT reaches state-of-the-art sepsis prediction on the P19 benchmark and is competitive with the best RNN methods for mortality prediction on two ICU datasets, while outperforming other transformers under increasing data sparsity. It argues that retaining informative missingness (the pattern of which readings are absent) and learning explicit per-sensor identity embeddings are what give BAT its edge. If true, the model is a practical route to combining multiple EHR datasets through shared sensor embeddings for large-scale pretraining, and a reusable benchmark suite for EHR classification.

What carries the argument

The central object is bi-axial attention: instead of flattening every (sensor, time) observation into one long sequence or attending along only one dimension, BAT clones the embedding matrix and runs self-attention along the time axis in one track and along the sensor axis in the other, switching the axis order between tracks. This keeps the sequence length at $T + D$ per track rather than $T \times D$, avoiding the quadratic blow-up of a dense tuple representation while still letting information flow across both axes. The design also preserves the binary observation mask inside every embedding and learns a separate embedding per sensor identity, which is what later enables sharing sensors across datasets.

What would settle it

Re-run the P19 comparison with a nested procedure: hold out one test split untouched until after hyperparameter selection, or select hyperparameters using only training and validation data; if BAT's reported 2.32-point AUPRC margin over the second-best model disappears or reverses, the state-of-the-art claim is not established.

Watch

Extended reading notes

Core claim

BAT's central claim is that EHR classification improves when a transformer is allowed to attend along both axes of the observation matrix at once. Each observation embedding concatenates the sensor reading with a binary observed/missing indicator, a learned sensor-identity embedding, and a sinusoidal time encoding; the data are then duplicated into two attention tracks that perform self-attention over time points and over sensors in opposite orders, with weights shared inside each track. On sepsis prediction for P19, BAT reports AUPRC 77.05 and AUROC 95.41, ahead of every baseline; on P12 and MIMIC-III mortality it is competitive with GRU-D, the best overall method. The paper also reports that BAT maintains higher AUPRC than two transformer baselines at every induced sparsity level on a human-activity dataset, and that sharing sensor embeddings between P12 and MIMIC-III reduces variance and mostly improves performance. The authors read this as evidence that cross-axis attention and explicit sensor identities address sparsity, informative missingness, and dataset integration.

Load-bearing premise

The reported state-of-the-art sepsis result rests on the assumption that the benchmark comparison is unbiased—in particular, that selecting hyperparameters by best AUROC on the same five splits used to report test performance, and the modified GRU-D and IP-Nets baselines, did not favor BAT.

Editorial extensions

If this is right

  • On the P19 sepsis benchmark, BAT reports the best AUPRC and AUROC among all compared models; if the comparison is sound, bi-axial attention is a stronger transformer design for sparse ICU time series than one-axis attention.
  • Attention visualizations show observed measurements attending strongly to absent time points, supporting the claim that BAT uses informative missingness rather than treating missingness as noise.
  • In HAR experiments with induced missingness, BAT outperforms both Transformer and iTransformer at every sparsity level up to 99%, where all models reach chance; this points to cross-axis attention as the source of robustness.
  • Shared sensor embeddings across P12 and MIMIC-III reduce variance and improve most metrics, suggesting that datasets with overlapping sensors can be trained jointly or used for transfer.
  • The re-implemented baselines and BAT are released in one open-source repository, making the benchmark reproducible.

Reading between the lines

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

  • A direct extension the authors leave implicit: because sensor identities are stored as separate embeddings, a model pretrained on one EHR system could be adapted to a new hospital by only learning new sensor embeddings and freezing the rest—this is testable on any pair of datasets with overlapping measurements.
  • The sparsity result on HAR suggests the advantage may extend beyond EHR to other irregularly sampled multivariate signals such as wearables; the same induced-missingness protocol would settle that.
  • If the supplementary selection procedure is taken literally—choosing the candidate with highest AUROC on the same five test splits—reported margins over baselines could shrink under a properly nested validation scheme; the authors do not discuss this, so the robustness of the state-of-the-art claim remains an open question.
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 paper proposes the Bi-Axial Transformer (BAT), an architecture for EHR classification that applies axial attention over both the sensor and time axes of multivariate irregularly sampled time series, while retaining explicit missingness indicators and learning per-sensor embeddings. The authors evaluate BAT on mortality prediction (P12, MIMIC-III), sepsis prediction (P19), and a controlled sparsity experiment on HAR, comparing against SeFT, GRU-D, IP-Nets, a standard Transformer, and iTransformer. They report that BAT achieves state-of-the-art sepsis prediction on P19, is competitive on mortality tasks, is more robust to missing data than other transformers, and that shared sensor embeddings enable joint training across datasets. All baselines were re-implemented in PyTorch and released in a public repository. The core empirical claim is the P19 SOTA result and the associated architectural advantages.

Significance. If the empirical results are unbiased, BAT is a sensible and modular extension of axial attention to sparse, irregular EHR time series. The sensor-embedding transfer experiment is a useful idea for multi-dataset pretraining, and the reimplementation of several baselines in a modern framework is a practical community contribution. The ablations and sparsity experiments are thoughtful and give insight into where the architecture helps. However, the central claims rest entirely on the benchmark numbers. The hyperparameter selection protocol described in the supplement appears to use test-set labels for model selection, which would invalidate the reported metrics and the SOTA statement. Until that issue is resolved, the significance of the contribution cannot be assessed.

major comments (3)
  1. [Supplementary §7.1] The hyperparameter sweep protocol states that 'each candidate was trained and tested on 5 splits of train, validation, and test data' and that 'the model with the highest resulting AUROC was selected.' Taken literally, this selects hyperparameters based on test-set AUROC, which makes the reported AUROC and AUPRC values in Table 2 in-sample selection results rather than unbiased estimates of generalization. This directly affects the central claim in the Abstract and Section 5 that BAT achieves state-of-the-art performance on P19 sepsis prediction. Please clarify whether the selection AUROC was computed on the validation splits or the test splits. If test splits were used, the experiments must be re-run with hyperparameters selected on validation data (or via nested cross-validation) and all tables regenerated; the repository contains the sweep code, so this is feasible.
  2. [§4.3.1] The modification of GRU-D and IP-Nets for P19 — pooling hidden states across all time points instead of using the final hidden state — is a substantial deviation from the original model definitions. The authors state this was done to avoid missed diagnoses when sepsis resolves before the end of the sequence, but this changes the inductive bias of the baselines. To establish a fair comparison, please report results for the original final-state versions as well, or justify with a citation that this pooling modification is standard practice for P19. Without this, it is unclear whether the reported baseline numbers are the strongest reasonable comparators. This issue is compounded by the hyperparameter-selection problem in §7.1, since the modified baselines' hyperparameters were selected under the same protocol.
  3. [Abstract; §5] The claim of 'state-of-the-art performance on sepsis prediction' is supported only by comparisons against the authors' reimplemented baselines. No published state-of-the-art numbers for the PhysioNet 2019 sepsis challenge or other recent P19 sepsis-prediction results are cited or compared. If the term 'state-of-the-art' is meant globally, the paper needs to benchmark against published results; if it is meant only relative to the included baselines, the wording should be changed accordingly. This is a substantive point because the P19 result is the paper's main empirical contribution.
minor comments (5)
  1. [§3.2.1] In the paragraph after Eq. (3), 'concatentated' should be 'concatenated.' Also, the description of tau as 'the maximum possible time point' is ambiguous; since the encoding uses tau in the denominator as a scaling factor, please clarify whether tau is the largest timestamp in the dataset or a fixed hyperparameter analogous to max_timescale in SeFT.
  2. [Table 2] The entry for Transformer on P19 AUPRC appears to be missing from the table (only the AUROC is visible in the text). Please ensure all cells are reported consistently, as the claim of a 2.32-point AUPRC improvement over the second-best model depends on that value.
  3. [§4.4.2] In the HAR sparsity experiment, the statement that all models perform 'equivalently to random guessing' at 99% sparsity would be more informative if the random-guess performance level were stated explicitly and shown as a reference line in Fig. 6.
  4. [§4.3.1] The sentences describing the average improvements over iTransformer and Transformer would benefit from a small table or explicit arithmetic, since the numbers are not directly readable from Table 2 due to the formatting of the bold/italic markers.
  5. [Fig. 4] The attention visualization is described in the text, but the figure itself is hard to interpret without a more detailed legend explaining the circles, arrows, and density plot. Consider moving some of the supplementary description into the main caption.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: BAT's claims are empirical and evaluated against external baselines; no equation reduces to a fitted prediction.

full rationale

The paper's contribution is an empirical architecture, the Bi-Axial Transformer, whose performance is measured against re-implemented baselines on P12, MIMIC-III, P19, and HAR. Walking the derivation chain, I found no step where a predicted quantity is defined in terms of the model's own inputs or fitted parameters. The loss in Eq. (10) is standard cross-entropy; the architecture in Section 3.2 is a composition of linear embeddings, sinusoidal time encodings taken from Horn et al. [43], two axial attention tracks, pooling, demographic fusion, and a classification head. The learned sensor embeddings are an architectural component, not a renamed benchmark result, and the transfer-learning experiment in Section 4.5 compares shared versus separate embeddings rather than deriving a prediction from a fit. The paper's only self-citation, reference [20], is a preliminary report of the same model and is used merely as a pointer ('A preliminary version of this work has been reported [20]'); it is not invoked to justify the architecture, to rule out alternatives, or to supply a uniqueness theorem. The concern raised in Supplementary Section 7.1, that each hyperparameter candidate was 'trained and tested on 5 splits of train, validation, and test data' with the highest-AUROC candidate selected, is a real evaluation-validity problem because test labels appear to be used for model selection, but it is not circular reasoning as defined here: it does not make the reported AUROC equal by construction to a fitted parameter, nor does it rename an input as a prediction. That issue belongs to correctness/benchmark-fairness, not to circular derivation. Accordingly, there are no circular steps and the circularity score is 0.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

No invented entities. The central empirical claims rest on hyperparameters chosen per dataset and a set of domain assumptions about missingness, axial attention, and sensor-transferability. The most fragile assumption is the fair-comparison premise, because the reported selection protocol uses test AUROC.

free parameters (3)
  • BAT hyperparameters = P12 embed 128, heads 2; MIMIC embed 16, heads 1; P19 embed 32, heads 1; full list in Table 4
    Selected by random sweep of 20 candidates per dataset to maximize AUROC; Supp. §7.1 indicates test AUROC was used for selection.
  • Baseline hyperparameters = Table 4 (SeFT widths, GRU-D units, IP-Nets impute steps, Transformer and iTransformer settings)
    Also selected via random sweep on AUROC; the paper does not report a shared validation-only selection protocol, so a fair comparison depends on unstated details.
  • Time encoding scale tau (Eq. 3) = not reported for BAT
    Scaling factor for the sinusoidal temporal encoding; a design choice that changes how time differences are represented.
assumptions (5)
  • standard math Scaled dot-product attention and sinusoidal positional encodings are taken as given from Vaswani et al. (2017).
    Equations (2)-(4) in §3.2.1-3.2.2 rely on these as the attention core.
  • domain assumption Axial attention from Ho et al. (2019) can be adapted by transposing time and sensor dimensions onto the batch axis, with shared encoder parameters across axes.
    This is assumption #2 in §3.2.2, the design premise that cross-axis parameter sharing enables learning time-sensor relationships.
  • domain assumption Informative missingness is present in EHRs and the binary indicator mask conveys clinically meaningful signal.
    Motivates including the mask in the embedding (§3.2.1) and is supported by cited literature and Fig. 5, but remains a modeling assumption.
  • domain assumption Sensor names are comparable across datasets when they overlap, so shared sensor embeddings transfer.
    Required for the Table 3 transfer experiment; results are mixed (MIMIC-III AUPRC drops with shared embeddings), so the premise is only weakly supported.
  • domain assumption The reimplemented baselines and their modifications are faithful enough for a fair comparison.
    §4.2-4.3.1; GRU-D and IP-Nets were modified for P19 (pooling all hidden states), which may change their behavior relative to original implementations.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Bi-Axial Transformers: Addressing the Increasing Complexity of EHR Classification." pith.science (2026). https://pith.science/paper/67PKAH3A

@misc{pith2026250812418,
  author       = {Pith},
  title        = {Pith review of: Bi-Axial Transformers: Addressing the Increasing Complexity of EHR Classification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/67PKAH3A}},
  note         = {Machine review of arXiv:2508.12418}
}
read the original abstract

Electronic Health Records (EHRs), the digital representation of a patient's medical history, are a valuable resource for epidemiological and clinical research. They are also becoming increasingly complex, with recent trends indicating larger datasets, longer time series, and multi-modal integrations. Transformers, which have rapidly gained popularity due to their success in natural language processing and other domains, are well-suited to address these challenges due to their ability to model long-range dependencies and process data in parallel. But their application to EHR classification remains limited by data representations, which can reduce performance or fail to capture informative missingness. In this paper, we present the Bi-Axial Transformer (BAT), which attends to both the clinical variable and time point axes of EHR data to learn richer data relationships and address the difficulties of data sparsity. BAT achieves state-of-the-art performance on sepsis prediction and is competitive to top methods for mortality classification. In comparison to other transformers, BAT demonstrates increased robustness to data missingness, and learns unique sensor embeddings which can be used in transfer learning. Baseline models, which were previously located across multiple repositories or utilized deprecated libraries, were re-implemented with PyTorch and made available for reproduction and future benchmarking.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

58 extracted references · 35 canonical work pages

  1. [1]

    Birkhead, Michael Klompas, and Nirav R

    Guthrie S. Birkhead, Michael Klompas, and Nirav R. Shah. Uses of electronic health records for public health surveillance to advance public health. ������ ������ �� ������ ������, 36:345–359, 2015. ISSN 1545-2093. doi:10.1146/annurev-publhealth-031914-122747

  2. [2]

    National electronic health records survey

    CDC. National electronic health records survey. 2021, 2021. URL ������������������������������� ������������������

  3. [3]

    Progress on implementing and using electronic health record systems

    Luke Slawomirski, Luca Lindner, Katherine de Bienassis, Philip Haywood, Tiago Cravo Oliveira Hashiguchi, Melanie Steentjes, and Jillian Oderkirk. Progress on implementing and using electronic health record systems. Technical report, OECD, 2023

  4. [4]

    Data-driven identification of predictive risk biomarkers for subgroups of osteoarthritis using interpretable machine learning

    Rikke Nielsen, Thomas Monfeuga, Robert Kitchen, Line Egerod, Luis Leal, August Schreyer, Frederik Gade, Carol Sun, Marianne Helenius, Lotte Simonsen, Marianne Willert, Abd Tahrani, Zahra McVey, and Ramneek Gupta. Data-driven identification of predictive risk biomarkers for subgroups of osteoarthritis using interpretable machine learning. ������ ����������...

  5. [5]

    Identification of risk factors of long COVID and predictive modeling in the RECOVER EHR cohorts

    Chengxi Zang, Yu Hou, Edward Schenck, Zhenxing Xu, Yongkang Zhang, Jie Xu, Jiang Bian, Dmitry Morozyuk, Dhruv Khullar, Anna Nordvig, Elizabeth Shenkman, Russell Rothman, Jason Block, Kristin Lyman, Yiye Zhang, Jay Varma, Mark Weiner, Thomas Carton, Fei Wang, and Rainu Kaushal. Identification of risk factors of long COVID and predictive modeling in the REC...

  6. [6]

    Leveraging electronic health records and knowledge networks for alzheimer’s disease prediction and sex-specific biological insights

    Alice Tang, Katherine Rankin, Gabriel Cerono, Silvia Miramontes, Hunter Mills, Jacquelyn Roger, Billy Zeng, Charlotte Nelson, Karthik Soman, Sarah Woldemariam, Yaqiao Li, Albert Lee, Riley Bove, Maria Glymour, Nima Aghaeepour, Tomiko Oskotsky, Zachary Miller, Isabel Allen, Stephan Sanders, and Marina Sirota. Leveraging electronic health records and knowle...

  7. [7]

    Multi-layer representation learning for medical concepts

    Edward Choi, Taha Bahadori, Elizabeth Searles, Catherine Coffey, Michael Thompson, James Bost, Javier Tejedor-Sojo, and Jimeng Sun. Multi-layer representation learning for medical concepts. In ����������� �� ��� ���� ��� ������ ������������� ���������� �� ��������� ��������� ��� ���� ������, Aug 2016. doi:10.1145/2939672.2939823

  8. [8]

    Modelling 30-day hospital readmission after discharge for COPD patients based on electronic health records

    Meng Li, Kun Cheng, Keisun Ku, Junlei Li, Hao Hu, and Carolina Ung. Modelling 30-day hospital readmission after discharge for COPD patients based on electronic health records. ��� ������� ���� ����������� ��������, 33, Apr 2023. doi:10.1038/s41533-023-00339-6

Show all 58 references
  1. [9]

    Deep learning prediction models based on EHR trajectories: A systematic review

    Ali Amirahmadi, Mattias Ohlsson, and Kobra Etminani. Deep learning prediction models based on EHR trajectories: A systematic review. ������� �� ���������� �����������, 144:104430, 2023. ISSN 1532-0464. doi:10.1016/j.jbi.2023.104430

  2. [10]

    ������������ ���������� ��� ��������� ��� �������

    Robert Hyndman and George Athanasopoulos. ������������ ���������� ��� ��������� ��� �������. OTexts, Melbourne, Australia, 2018

  3. [11]

    Unsupervised pattern discovery in electronic health care data using probabilistic clustering models

    Benjamin Marlin, David Kale, Robinder Khemani, and Randall Wetzel. Unsupervised pattern discovery in electronic health care data using probabilistic clustering models. In ����������� �� ��� ��� ��� ������ ������������� ������ ����������� ��������� ��������, 2012. doi:10.1145/2...

  4. [12]

    Machine learning and decision support in critical care

    Alistair Johnson, Mohammad Ghassemi, Shamim Nemati, Katherine Niehaus, David Clifton, and Gari Clifford. Machine learning and decision support in critical care. ����������� �� ��� ����, 104:444–466, Feb 2016. doi:10.1109/JPROC.2015.2501978

  5. [13]

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

    Feng Xie, Han Yuan, Yilin Ning, Marcus Eng Hock Ong, Mengling Feng, Wynne Hsu, Bibhas Chakraborty, and Nan Liu. Deep learning for temporal data representation in electronic health records: A systematic review of challenges and methodologies. ������� �� ���������� �����������, ...

  6. [14]

    Goldberger, Luis A

    Ary L. Goldberger, Luis A. N. Amaral, Leon Glass, Jeffrey M. Hausdorff, Plamen Ch. Ivanov, Roger G. Mark, Joseph E. Mietus, George B. Moody, Chung-Kang Peng, and H. Eugene Stanley. Physiobank, physiotoolkit, and physionet: Components of a new research resource for complex phys...

  7. [15]

    The Danish National Patient Registry: a review of content, data quality, and research potential

    Morten Schmidt, Sigrun Johannesdottir, Jakob Sandegaard, Vera Ehrenstein, Lars Pedersen, and Henrik Sørensen. The Danish National Patient Registry: a review of content, data quality, and research potential. �������� ������������, 7:449, Nov 2015. doi:10.2147/CLEP.S91125

  8. [16]

    MIMIC-IV (version 3.1)

    Alistair Johnson, Lucas Bulgarelli, Tom Pollard, Brian Gow, Benjamin Moody, Steven Horng, Leo Anthony Celi, and Roger Mark. MIMIC-IV (version 3.1). Technical report, PhysioNet, 2024. 13 This work has been submitted to the IEEE for possible publication. Copyright may be transfe...

  9. [17]

    Data snapshots, 2025

    NIH. Data snapshots, 2025. URL ����������������������������������������������������������

  10. [18]

    Data resource profile: Clinical practice research datalink (CPRD) aurum

    Achim Wolf, Daniel Dedman, Jennifer Campbell, Helen Booth, Darren Lunn, Jennifer Chapman, and Puja Myles. Data resource profile: Clinical practice research datalink (CPRD) aurum. ������������� ������� �� ������������, 48(6):1740–1740g, Mar 2019. ISSN 0300-5771. doi:10.1093/ije...

  11. [19]

    UK biobank: An open access resource for identifying the causes of a wide range of complex diseases of middle and old age

    Cathie Sudlow, John Gallacher, Naomi Allen, Valerie Beral, Paul Burton, John Danesh, Paul Downey, Paul Elliott, Jane Green, Martin Landray, Bette Liu, Paul Matthews, Giok Ong, Jill Pell, Alan Silman, Alan Young, Tim Sprosen, Tim Peakman, and Rory Collins. UK biobank: An open a...

  12. [20]

    Parallel time-sensor attention for electronic health record classification

    Rachael DeVries, Marie Lisandra Zepeda Mendoza, and Ole Winther. Parallel time-sensor attention for electronic health record classification. ����������, 2024. URL ������������������������������������������ . Preprint

  13. [21]

    Risk prediction with electronic health records: A deep learning approach

    Yu Cheng, Fei Wang, Ping Zhang, and Jianying Hu. Risk prediction with electronic health records: A deep learning approach. In ���� ���� ������������� ���������� �� ���� ������, pages 432–440, Jun 2016. doi:10.1137/1.9781611974348.49

  14. [22]

    Stochastic imputation and uncertainty-aware attention to EHR for mortality prediction

    Eunji Jun, Ahmad Wisnu Mulyadi, and Heung-Il Suk. Stochastic imputation and uncertainty-aware attention to EHR for mortality prediction. In ���� ������������� ����� ���������� �� ������ �������� �������, pages 1–7. IEEE, 2019

  15. [23]

    Feature rearrangement based deep learning system for predicting heart failure mortality

    Zhe Wang, Yiwen Zhu, Dongdong Li, Yichao Yin, and Jing Zhang. Feature rearrangement based deep learning system for predicting heart failure mortality. �������� ������� ��� �������� �� �����������, 191:105383, 2020

  16. [24]

    Graph-guided network for irregularly sampled multivariate time series

    Xiang Zhang, Marko Zeman, Theodoros Tsiligkaridis, and Marinka Zitnik. Graph-guided network for irregularly sampled multivariate time series. In ������������� ���������� �� �������� ���������������� ����, 2022

  17. [25]

    Phased LSTM: Accelerating recurrent network training for long or event-based sequences

    Daniel Neil, Michael Pfeiffer, and Shih-Chii Liu. Phased LSTM: Accelerating recurrent network training for long or event-based sequences. ���� , abs/1610.09513, 2016

  18. [26]

    A multi-layered GRU model for COVID-19 patient representation and phenotyping from large-scale EHR data

    Arpita Saha, Maggie Samaan, Bo Peng, and Xia Ning. A multi-layered GRU model for COVID-19 patient representation and phenotyping from large-scale EHR data. In ����������� �� ��� ���� ��� ������������� ���������� �� ��������������� ������������� �������� ��� ������ �����������,...

  19. [27]

    Recurrent neural networks for multivariate time series with missing values

    Zhengping Che, Sanjay Purushotham, Kyunghyun Cho, David Sontag, and Yan Liu. Recurrent neural networks for multivariate time series with missing values. ��������� �������, 8(1):6085, 2018. doi:10.1038/s41598-018- 24271-9

  20. [28]

    Payberah, Mariagrazia Zottoli, Milad Nazarzadeh, Nathalie Conrad, Kazem Rahimi, and Gholamreza Salimi-Khorshidi

    Jose Roberto Ayala Solares, Francesca Elisa Diletta Raimondi, Yajie Zhu, Fatemeh Rahimian, Dexter Canoy, Jenny Tran, Ana Catarina Pinho Gomes, Amir H. Payberah, Mariagrazia Zottoli, Milad Nazarzadeh, Nathalie Conrad, Kazem Rahimi, and Gholamreza Salimi-Khorshidi. Deep learning...

  21. [29]

    A review of deep learning models and online healthcare databases for electronic health records and their use for health prediction

    Nurul Nasarudin, Fatma Al Jasmi, Richard Sinnott, Nazar Zaki, Hany Alashwal, Elfadil Mohamed, and Mohd Mohamad. A review of deep learning models and online healthcare databases for electronic health records and their use for health prediction. ��������� ������������ ������, 57...

  22. [30]

    EHRXQA: A multi-modal question answering dataset for electronic health records with chest x-ray images

    Seongsu Bae, Daeun Kyung, Jaehee Ryu, Eunbyeol Cho, Gyubok Lee, Sunjun Kweon, Jungwoo Oh, Lei Ji, Eric Chang, Tackeun Kim, and Edward Choi. EHRXQA: A multi-modal question answering dataset for electronic health records with chest x-ray images. In �������� �� ������ �����������...

  23. [31]

    Fusion of medical imaging and electronic health records using deep learning: A systematic review and implementation guidelines

    Shih-Cheng Huang, Anuj Pareek, Saeed Seyyedi, Imon Banerjee, and Matthew Lungren. Fusion of medical imaging and electronic health records using deep learning: A systematic review and implementation guidelines. ��� ������� ��������, 3, Dec 2020. doi:10.1038/s41746-020-00341-z

  24. [32]

    Integrating multi-omics data with EHR for precision medicine using advanced artificial intelligence

    Li Tong, Wenqi Shi, Monica Isgut, Yishan Zhong, Peter Lais, Logan Gloster, Jimin Sun, Aniketh Swain, Felipe Giuste, and May Wang. Integrating multi-omics data with EHR for precision medicine using advanced artificial intelligence. ���� ������� �� ���������� �����������, PP:1–1...

  25. [33]

    Mohr, Carmen P

    Alex E. Mohr, Carmen P. Ortega-Santos, Corrie M. Whisner, Judith Klein-Seetharaman, and Paniz Jasbi. Navi- gating challenges and opportunities in multi-omics integration for personalized healthcare. ������������, 12(7),

  26. [34]

    Larry Hill, Gretchen Sanders, Judith C

    Keith Marsolo, Lesley Curtis, Laura Qualls, Jennifer Xu, Yinghong Zhang, Thomas Phillips, C. Larry Hill, Gretchen Sanders, Judith C. Maro, Daniel Kiernan, et al. Assessing the harmonization of structured electronic health record data to reference terminologies and data complet...

  27. [35]

    Learning long-term dependencies with gradient descent is difficult

    Yoshua Bengio, Patrice Simard, and Paolo Frasconi. Learning long-term dependencies with gradient descent is difficult. ���� ������������ �� ������ ��������, 5:157–166, Feb 1994. doi:10.1109/72.279181

  28. [36]

    BEHRT: Transformer for electronic health records

    Yikuan Li, Shishir Rao, Jose Roberto Ayala Solares, Abdelâali Hassaïne, Rema Ramakrishnan, Dexter Canoy, Yajie Zhu, Kazem Rahimi, and Gholamreza Salimi-Khorshidi. BEHRT: Transformer for electronic health records. ��������� �������, 10, 2020. doi:10.1038/s41598-020-62922-y

  29. [37]

    TransEHR: Self-supervised transformer for clinical time series data

    Yanbo Xu, Shangqing Xu, Manav Ramprassad, Alexey Tumanov, and Chao Zhang. TransEHR: Self-supervised transformer for clinical time series data. In ����������� �� ��� ��� ������� �������� ��� ������ ���������, volume 225 of ����������� �� ������� �������� ��������, pages 623–635, 2023

  30. [38]

    Sindhu Tipirneni and Chandan K. Reddy. Self-supervised transformer for sparse and irregularly sampled multivariate clinical time-series. ��� ������������ �� ��������� ��������� ���� ����, 16(6), Jul 2022. ISSN 1556-4681. doi:10.1145/3516367

  31. [39]

    Gomez, Łukasz Kaiser, and Illia Polosukhin

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need. �������� �� ������ ����������� ���������� �������, 30, 2017

  32. [40]

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

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, and Neil Houlsby. An image is worth 16x16 words: Transformers for image recognition at...

  33. [41]

    Med-BERT: Pre-trained contextualized embed- dings on large-scale structured electronic health records for disease prediction

    Laila Rasmy, Yang Xiang, Ziqian Xie, Cui Tao, and Degui Zhi. Med-BERT: Pre-trained contextualized embed- dings on large-scale structured electronic health records for disease prediction. ��� ������� ��������, 4, 2020. doi:10.1038/s41746-021-00455-y

  34. [42]

    ClinicalBERT: Modeling clinical notes and predicting hospital readmission

    Kexin Huang, Jaan Altosaar, and Rajesh Ranganath. ClinicalBERT: Modeling clinical notes and predicting hospital readmission. ����� �������� ����������������, 2019

  35. [43]

    Borgwardt

    Max Horn, Michael Moor, Christian Bock, Bastian Rieck, and Karsten M. Borgwardt. Set functions for time series. In ����������� �� ��� ���� ������������� ���������� �� ������� ��������, volume 119 of ����������� �� ������� �������� ��������, pages 4353–4363. PMLR, Jul 2020

  36. [44]

    Characterizing and managing missing structured data in electronic health records: Data analysis

    Brett Beaulieu-Jones, Daniel Lavage, John Snyder, Jason Moore, Sarah Pendergrass, and Christopher Bauer. Characterizing and managing missing structured data in electronic health records: Data analysis. ���� ������� �����������, 6, Feb 2018. doi:10.2196/medinform.8960

  37. [45]

    Deep time series forecasting models: A comprehensive survey

    Xinhe Liu and Wenmin Wang. Deep time series forecasting models: A comprehensive survey. �����������, 12, Jul 2024. doi:10.3390/math12101504

  38. [46]

    itransformer: Inverted transformers are effective for time series forecasting

    Yong Liu, Tengge Hu, Haoran Zhang, Haixu Wu, Shiyu Wang, Lintao Ma, and Mingsheng Long. itransformer: Inverted transformers are effective for time series forecasting. ����� �������� ����������������, 2023

  39. [47]

    Are transformers effective for time series forecasting? In ����������� �� ��� ���� ���������� �� ��������� ������������, 2023

    Ailing Zeng, Muxi Chen, Lei Zhang, and Qiang Xu. Are transformers effective for time series forecasting? In ����������� �� ��� ���� ���������� �� ��������� ������������, 2023

  40. [48]

    Long-term forecasting with TiDE: Time-series dense encoder

    Abhimanyu Das, Weihao Kong, Andrew Leach, Rajat Sen, and Rose Yu. Long-term forecasting with TiDE: Time-series dense encoder. ����� �������� ����������������, 2023

  41. [49]

    Crossformer: Transformer utilizing cross-dimension dependency for multivariate time series forecasting

    Yunhao Zhang and Junchi Yan. Crossformer: Transformer utilizing cross-dimension dependency for multivariate time series forecasting. In ��� �������� ������������� ���������� �� �������� ���������������, 2022

  42. [50]

    TSMixer: Lightweight MLP-mixer model for multivariate time series forecasting

    Vijay Ekambaram, Arindam Jati, Nam Nguyen, Phanwadee Sinthong, and Jayant Kalagnanam. TSMixer: Lightweight MLP-mixer model for multivariate time series forecasting. ����� �������� ����������������, 2023

  43. [51]

    Interpolation-prediction networks for irregularly sampled time series

    Satya Narayan Shukla and Benjamin Marlin. Interpolation-prediction networks for irregularly sampled time series. In ������������� ���������� �� �������� ���������������, 2019

  44. [52]

    Alistair E. W. Johnson, Tom J. Pollard, Lu Shen, Li-Wei H. Lehman, Mengling Feng, Mohammad Ghassemi, Benjamin Moody, Peter Szolovits, Leo Anthony Celi, and Roger G. Mark. MIMIC-III, a freely accessible critical care database. ��������� ����, 3(1):1–9, 2016

  45. [53]

    Axial attention in multidimensional transformers

    Jonathan Ho, Nal Kalchbrenner, Dirk Weissenborn, and Tim Salimans. Axial attention in multidimensional transformers. ���� , abs/1912.12180, 2019

  46. [54]

    Early prediction of sepsis from clinical data: The Phys- ioNet/computing in cardiology challenge 2019, 2019

    Alistair Johnson, Tom Pollard, and Roger Mark. Early prediction of sepsis from clinical data: The Phys- ioNet/computing in cardiology challenge 2019, 2019. 15 This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after ...

  47. [55]

    Daniel Garcia-Gonzalez, Daniel Rivero, Enrique Fernandez-Blanco, and Miguel R. Luaces. A public domain dataset for real-life human activity recognition using smartphone sensors. �������, 20(8), 2020. ISSN 1424-8220. doi:10.3390/s20082200

  48. [56]

    Human activity recognition using smartphones

    Jorge Reyes-Ortiz, Davide Anguita, Alessandro Ghio, Luca Oneto, and Xavier Parra. Human activity recognition using smartphones. UCI Machine Learning Repository, 2013. doi:10.24432/C54S4K. 16 This work has been submitted to the IEEE for possible publication. Copyright may be tr...

  49. [2023]

    doi:10.1145/3584371.3612986

    Association for Computing Machinery. doi:10.1145/3584371.3612986

  50. [2024]

    doi:10.3390/biomedicines12071496

    ISSN 2227-9059. doi:10.3390/biomedicines12071496. 14 This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.