Add Transfer Prediction Task

[Hoogashaga]

Authors: Sheng-Hao Chen([email protected])
NetID: sc161

This module implements the TransferPrediction task, which provides a baseline solution for predicting patient transfers between care units. The implementation uses the MIMIC-III Clinical Database (version 1.4) and focuses on the transfer prediction aspect of the paper.

The MIMIC-III database is available at https://physionet.org/content/mimiciii/1.4/

Key Features:

• Implements transfer prediction using structured EHR data
• Uses MIMIC-III TRANSFERS table for training and evaluation
• Provides baseline performance metrics for comparison
• Follows PyHealth task interface standards

Changed files:

• pyhealth/tasks/transfer_prediction.py
• pyhealth/tasks/init.py
• pyhealth/datasets/configs/mimic3.yaml