Seminar: Isaac Sekitoleko, 'Machine learning reveals predictors of diabetes in rural Uganda'

Isaac Sekitoleko discusses his work on machine learning models for diabetes prediction in rural Uganda, where diagnostic resources are limited and machine learning is underutilised. The team have evaluated multiple algorithms using data on socio-demographics, anthropometry, diet, physical activity, and clinical assessments. Sugar consumption was the strongest diabetes predictor, followed by age, dietary patterns, and number of children. The findings suggest that incorporating machine learning into community-based screening would be effective at identifying high-risk individuals in rural Uganda, particularly those with high sugary drink consumption.