Short Paper
Up to 4 pages, excluding references and appendices, following ICML'26 formatting guidelines and anonymized for double-blind review.
Machine Learning/CORE Unranked
SD4H ICML 2026
ICML 2026 Workshop on Structured Data for Health
The Structured Data for Health (SD4H) workshop at ICML 2026 brings together researchers to advance machine learning methods for modeling structured healthcare data, including electronic health records, physiological time-series, and irregular clinical measurements. The workshop focuses on unifying fragmented data modalities, improving interpretability and clinical deployment, and fostering cross-domain collaboration through non-archival presentations.
Key deadlines
Verified September 9, 2026Full paper
May 1, 2026
AoE
Workshop timeline
Submission and decisions
Full paperKey deadline
May 1, 2026 · AoE
Paper fit
Contribution paths
A strong submission should clearly identify its contribution and evaluate it appropriately.
ML4H-to-TS4H Cross-submission
Full-length 9-page papers submitted to ML4H with a cross-submission option to SD4H; subject to stricter ML4H review and area fit assessment.
Research areas in scope
01
Models & Methods
Novel Architectures: State-space models, diffusion models, etc.Foundation Models: Pre-training, scaling, and alignment.Deep Learning: Supervised, self-supervised, and unsupervised methods.Probabilistic Modeling: Uncertainty quantification and Bayesian methods.Sequential Decision-Making: Reinforcement learning and optimal control.
02
Data-Specific Challenges
Multimodal Learning: Fusing structured data (e.g. time-series, tables, graphics) with images or text.Structured Adaptation: Using LLMs or other methods to introduce structure to unorganized data (i.e. free-text).Irregular & Missing Data: Handling sparse or irregularly-sampled series.Complex Signals: Modeling high-dimensional or multi-resolution data.Causal Inference: Inferring cause-and-effect from observational data.Representation Learning & Adaptation: Pre-training, adaptation strategies that transfer across patients, devices, test-time adaptation.
03
Applications & Trustworthy AI
Clinical Applications: Forecasting, risk stratification, digital biomarkers.Trust & Reliability: Explainability, fairness, robustness, and privacy.Deployment & Implementation: Real-world case studies and MLOps, federated evaluation, online learning during deployment.New Resources: Public datasets, benchmarks, and software.
Policies worth checking twice
- Submissions must be original and unpublished, though arXiv preprints are permitted.
- All submissions must be fully anonymized for double-blind review.
- Papers not conforming to formatting or anonymization guidelines will be desk-rejected.
- Accepted papers must be presented in person at the workshop.
- The workshop is non-archival: no formal proceedings, and accepted papers may be submitted elsewhere.
- Concurrent submissions to other venues (including other ICML workshops) are allowed, provided the other venue permits it.
- Papers previously accepted elsewhere may be submitted, if the original venue allows presentation at a non-archival workshop.
- Papers withdrawn from a main conference track may be submitted.
Official sources
Compiled from the official call for papers. The organizers’ pages remain authoritative.
Last verified September 9, 2026