pyhealth

Automate healthcare AI development with EHR data and clinical models.

1|Updated Mar 4, 2026
One-click install
npx skills add https://github.com/Hung-3008/agusta --skill pyhealth-hung-3008
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: pyhealth
Source: https://github.com/Hung-3008/agusta/tree/main/.agents/skills/pyhealth
Command: npx skills add https://github.com/Hung-3008/agusta --skill pyhealth-hung-3008

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

PyHealth provides a unified toolkit to streamline the development, testing, and deployment of healthcare AI models using clinical data and standardized datasets.

Core Features & Use Cases

  • Data loading from major healthcare datasets (MIMIC, eICU, OMOP) with a consistent interface.
  • Task definition and dataset transformation to task-specific samples.
  • Model selection and training pipeline with evaluation and deployment readiness.
  • Use cases include mortality/readmission/LOS prediction, medical coding translation, EEG/ECG processing, imaging and clinical text analysis.
  • Real-world workflow: load data, set a task, split data, train a model (e.g., Transformer), evaluate, and interpret.

Quick Start

Load the MIMIC-IV dataset, set mortality_prediction_mimic4_fn as the task, train a Transformer model, and evaluate on a held-out test set.

Frequently Asked Questions about pyhealth

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I train clinical prediction models using MIMIC-IV EHR data?

To train clinical prediction models using MIMIC-IV EHR data, you can load the dataset, define a specific task like mortality prediction, and train a model such as a Transformer through a modular pipeline that includes evaluation and deployment.

Can I use a unified pipeline for medical coding translation and readmission prediction?

Yes, you can use a unified pipeline for medical coding translation and readmission prediction. The toolkit supports task definition and dataset transformation to create task-specific samples for both clinical prediction and coding tasks.

What is the best way to process EEG and ECG signals for healthcare AI?

The best way to process EEG and ECG signals for healthcare AI is using a modular pipeline that includes data loading, task definition, and model selection, applicable to signal processing alongside imaging and clinical text analysis.

Does EHR model training require separate interfaces for OMOP and eICU datasets?

EHR model training does not require separate interfaces for OMOP and eICU datasets. The toolkit provides a consistent data loading interface across major healthcare datasets, streamlining the development and testing of clinical models.

How do I interpret mortality prediction results after training a Transformer model?

To interpret mortality prediction results after training a Transformer model, the modular pipeline includes an evaluation step that assesses the model on a held-out test set, providing deployment readiness and interpretability for clinical use.