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.