Research map / Time series / Foundation models for time series
Contrast Everything: A Hierarchical Contrastive Framework for Medical Time-Series
Yihe Wang, Yu Han, Haishuai Wang, Xiang Zhang
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 |
|---|---|---|---|
| Use hierarchical contrastive objectives across observation, sample, trial, and patient levels | COMET · AD, F-FT, 10% labeled data | F1 score: 91.14\std{3.31} vs 75.38\std{11.47} | Improved |
| Use hierarchical contrastive objectives across observation, sample, trial, and patient levels | COMET · AD, F-FT, 1% labeled data | F1 score: 88.14\std{3.37} vs 74.33\std{14.45} | Improved |
| Use hierarchical contrastive objectives across observation, sample, trial, and patient levels | COMET · PTB, F-FT, 10% labeled data | F1 score: 84.01\std{5.61} vs 83.84\std{4.03} | Improved |
| Use hierarchical contrastive objectives across observation, sample, trial, and patient levels | COMET · PTB, F-FT, 1% labeled data | F1 score: 88.23\std{2.10} vs 85.57\std{3.41} | Improved |
| Use hierarchical contrastive objectives across observation, sample, trial, and patient levels | COMET · TDBrain, F-FT, 10% labeled data | F1 score: 79.19\std{4.62} vs 76.99\std{2.28} | Improved |
| Use hierarchical contrastive objectives across observation, sample, trial, and patient levels | COMET · TDBrain, F-FT, 1% labeled data | F1 score: 72.57\std{7.37} vs 63.70\std{2.30} | Improved |
| Remove the observation and sample blocks, retaining only the trial-level block (ablation) | COMET · AD, PTB, and TDBrain | F1 score | Mixed |
| Remove the observation, sample, and trial blocks, retaining only the patient-level block (ablation) | COMET · AD, PTB, and TDBrain | F1 score | Mixed |
| Add further contrastive blocks to a single-level objective (ablation) | COMET · AD, PTB, and TDBrain | F1 score | Mixed |
| Change the relative weights of the four hierarchical losses (ablation) | COMET · AD, PTB, and TDBrain | F1 score | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- TS2vec — TS2Vec: Towards Universal Representation of Time Series (AAAI 2022) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- JoLT: Jointly Learned Representations of Language and Time-Series for Clinical Time-Series Interpretation (Student Abstract) (AAAI 2024 · similarity 0.86)
- TRACE: Contrastive learning for multi-trial time series data in neuroscience (NeurIPS 2025 · similarity 0.86)
- Integrating Sequence and Image Modeling in Irregular Medical Time Series Through Self-Supervised Learning (AAAI 2025 · similarity 0.85)
- Parametric Augmentation for Time Series Contrastive Learning (ICLR 2024 · similarity 0.84)
- Frequency-Masked Embedding Inference: A Non-Contrastive Approach for Time Series Representation Learning (AAAI 2025 · similarity 0.84)
- Separating common from salient patterns with Contrastive Representation Learning (ICLR 2024 · similarity 0.84)
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.
