The Agent Skills Directory

[SAFE]: The skill implements a standard research-oriented pipeline for clinical predictive modeling. All provided code snippets use the legitimate PyHealth library for its intended purpose of medical data analysis.
[SAFE]: No suspicious network activity, data exfiltration, or obfuscated code was detected. The instructions focus on local file operations on standard clinical datasets (e.g., MIMIC-III, MIMIC-IV) and provide appropriate environmental checks.
[SAFE]: The dependencies listed (PyHealth, PyTorch, and Pandas) are standard tools in the medical AI and data science communities, suitable for the skill's stated goals.
[SAFE]: The skill includes explicit security and quality checks, such as recommending patient-level splits to prevent data leakage and requiring calibration for clinical decision support outputs.

ehr-analysis