Clinical decision support systems for patient safety, risk prediction, and medical interaction analysis.
Machine learning and decision intelligence models targeting clinical risk prediction and medical interactions.
This repository provides clinical decision-support and patient-safety systems designed to handle high-consequence medical domains. It suits clinical researchers and engineers building risk prediction models or analyzing medical interactions.
The framework integrates operations research and applied AI across four core projects, including ICU early-warning models for hypoxemia and hypotension, a PyHealth extension, a physiological waveform pipeline, and a clinical empathy analysis tool for consultation transcripts. It enforces strict ground rules requiring run logs and real data scales to back up performance claims.
It suits data scientists and clinical researchers developing evidence-based safety models for ICU environments or medical text analysis. It does not suit clinicians looking for a ready-to-deploy commercial electronic medical record application.
Predicts hypoxemia and hypotension on MIMIC-IV or eICU data with built-in calibration and uncertainty estimates.
Extends PyHealth and RHealth as a installable package to track adoption metrics via registry download statistics.
Supplies preprocessing and a derived dataset for high-frequency physiological waveforms feeding the ICU models.
Quantifies doctor empathy in Chinese clinical consultation transcripts using weighted lexicons and machine learning features.
The repository provides open-source project structures and pipelines. The only cost is running it yourself and acquiring access to required clinical databases.
885 open remote AI-training roles, updated November 17, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
These reviewed AI tools do the same job instantly — from our scored directory.
The models target MIMIC-IV and eICU datasets at a scale of 12,000 plus patients.
The repository is structured as a framework; results must be generated locally using run logs as mandated by the ground rules.
See Zekai first in Google