ED-based epidemiology & interventions

This groundbreaking study develops culturally-adapted machine learning algorithms for early NCD detection using basic health measurements accessible in rural settings. Trained on data from 15,000 patients across Kenya and Uganda, our AI model achieves 89% accuracy in predicting diabetes risk and 92% accuracy for hypertension using only BMI, age, family history, and simple questionnaire responses. The low-cost screening tool requires no specialized equipment and can be deployed via smartphones by community health workers.
View Full Study
Topics
AI in Healthcare
NCD Prevention
Community Health
Countries
Kenya
Uganda
Collaborators
Dr. Christine Ngaruiya
Dr. James Chen
Dr. Linda Jepkoech Kimaru, PhD, MPH
Output
April 2024 - August 2024
Nature Digital Medicine