PyHealth Roadmap
What We're Building Next
A transparent look at the features the PyHealth community is actively developing — from new model modalities to agentic workflows and rigorous evaluation frameworks.
What the PyHealth Community is Currently Building
These extensions are actively in development. Want to help? See how to contribute below.
| Extension | Description | Status |
|---|---|---|
| Standardized Interpretability API | Build recipes to interpret custom models with state-of-the-art interpretability approaches. | In Progress |
| PyHealth Agent / MCP | Directly enable agents to write and run PyHealth code for you. | In Progress |
| Multimodal Models | Feed any combination of features and modalities and train a model — even in cases of data missingness. | In Progress |
| EHR Foundational Models | Enable fine-tuning of existing EHR foundational models within the PyHealth pipeline. | In Progress |
| EHR Generative Models | Train and deploy your own synthetic patient generators. | In Progress |
| Conformal Prediction Benchmarks | Test whether conformal methods can provide robust guarantees of predictions in high-stakes healthcare settings. | In Progress |
Want to help build PyHealth?
Join 400+ community members and 10+ active researchers — contributions of all kinds are welcome.