Research map / Time series / Foundation models for time series
Improving Medical Predictions by Irregular Multimodal Electronic Health Records Modeling
Xinlu Zhang, Shiyang Li, Zhiyu Chen, Xifeng Yan, Linda Ruth Petzold
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Combine imputation and mTAND embeddings with a learned gate | MISTS predictions with UTDE · 48-IHM | F1: 43.87 -> 45.26 | Improved |
| Combine imputation and mTAND embeddings with a learned gate | MISTS predictions with UTDE · 24-PHE | F1: 23.36 -> 24.89 | Improved |
| Model irregular note times with mTAND time attention | Clinical-note predictions · 48-IHM | F1: 51.01 -> 52.57 | Improved |
| Model irregular note times with mTAND time attention | Clinical-note predictions · 24-PHE | F1: 51.09 -> 52.95 | Improved |
| Interleave self-attention and cross-attention across temporal steps | Multimodal fusion · 48-IHM | F1: 54.13 -> 56.45 | Improved |
| Interleave self-attention and cross-attention across temporal steps | Multimodal fusion · 24-PHE | F1: 54.20 -> 54.84 | Improved |
| Replace mTAND with IP-Net in the gated TDE module (ablation) | UTDE · 48-IHM | F1: 45.26 -> 44.88 | Hurt |
| Replace mTAND with IP-Net in the gated TDE module (ablation) | UTDE · 24-PHE | F1: 24.89 -> 24.06 | Hurt |
| Use UTDE rather than imputation-only time-series representations (ablation) | Multimodal fusion · 48-IHM | F1: 54.59 -> 56.45 | Improved |
| Use UTDE rather than mTAND-only time-series representations (ablation) | Multimodal fusion · 48-IHM | F1: 54.89 -> 56.45 | Improved |
| Use UTDE rather than imputation-only time-series representations (ablation) | Multimodal fusion · 24-PHE | F1: 54.46 -> 54.84 | Improved |
| Use UTDE rather than mTAND-only time-series representations (ablation) | Multimodal fusion · 24-PHE | F1: 54.07 -> 54.84 | Improved |
Closest papers by meaning
- Integrating Sequence and Image Modeling in Irregular Medical Time Series Through Self-Supervised Learning (AAAI 2025 · similarity 0.88)
- Advancing Medical Multimodal Learning and Data Generation with Diffusion Model and LLM (AAAI 2025 · similarity 0.86)
- Context Clues: Evaluating Long Context Models for Clinical Prediction Tasks on EHR Data (ICLR 2025 · similarity 0.85)
- One Loss to Rule Them All: Marked Time-to-Event for Structured EHR Foundation Models (NeurIPS 2026 · similarity 0.85)
- XTSFormer: Cross-Temporal-Scale Transformer for Irregular-Time Event Prediction in Clinical Applications (AAAI 2025 · similarity 0.85)
- Revitalizing Medical Time Series with Vision-Informed Retrieval: A Vision-Language Perspective (NeurIPS 2026 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
