What types of models are available in this skill?

The skill provides access to over 33 models, including general-purpose architectures like Transformers and GNNs, and healthcare-specific ones like RETAIN, SafeDrug, and StageNet.

Can PyHealth help with medical code translation?

Yes, it includes specialized tools for mapping and translating between systems like ICD-9/10, NDC, RxNorm, and ATC.

What healthcare datasets does PyHealth support?

PyHealth supports major datasets including MIMIC-III, MIMIC-IV, eICU, OMOP, and various physiological signal datasets like SleepEDF.

How does PyHealth handle patient privacy and data splitting?

PyHealth provides built-in functions to split data by patient ID, preventing data leakage and ensuring that a patient's data does not appear in both training and testing sets.

PyHealth

Name: PyHealth
Author: jimmc414

byjimmc414

•

324

Data Science & ML

Builds and deploys specialized machine learning models for clinical healthcare data and electronic health records.

About

PyHealth is a comprehensive healthcare AI toolkit designed to streamline the development of clinical prediction models using electronic health records (EHR), medical imaging, and physiological signals. By providing a modular 5-stage pipeline—covering data loading, task definition, model selection, training, and deployment—it enables researchers and developers to implement state-of-the-art architectures like RETAIN and SafeDrug while handling complex medical coding systems such as ICD-10 and ATC. It is optimized for scientific discovery and high-performance healthcare data processing.

Key Features

Standardized data loaders for MIMIC-III/IV, eICU, and OMOP datasets
Automated medical coding translation across ICD, NDC, RxNorm, and ATC systems
Library of 33+ deep learning models optimized for clinical prediction tasks
Predefined workflows for mortality, readmission, and drug recommendation
Advanced evaluation tools for clinical fairness, calibration, and interpretability
324 GitHub stars

Use Cases

Predicting ICU patient mortality and hospital readmission risk
Analyzing physiological signals like EEG and ECG for sleep disorder diagnosis
Generating safe medication recommendations with drug-drug interaction constraints

About

Key Features

Standardized data loaders for MIMIC-III/IV, eICU, and OMOP datasets
Automated medical coding translation across ICD, NDC, RxNorm, and ATC systems
Library of 33+ deep learning models optimized for clinical prediction tasks
Predefined workflows for mortality, readmission, and drug recommendation
Advanced evaluation tools for clinical fairness, calibration, and interpretability
324 GitHub stars

Use Cases

Predicting ICU patient mortality and hospital readmission risk
Analyzing physiological signals like EEG and ECG for sleep disorder diagnosis
Generating safe medication recommendations with drug-drug interaction constraints