Diabetes is a disease which kills an estimated 90000 South Africans annually, and cost the public health sector around R15bn in 2021. There is no cure for diabetes, but there are
many successful strategies that can be employed in order to reduce the harm the disease causes. This makes diagnosis especially important as early diagnosis can lead to lifestyle
changes and treatment which dramatically improve patient outcomes. In South Africa, it is estimated that 2 million people with diabetes are undiagnosed. One strategy to mitigate this is to
build a risk prediction model, in which undiagnosed diabetes can be predicted from various recorded factors. To be maximally useful, the research should suggest which features are the most important in predicting diabetes, and identify subsets of risk factors which can accurately predict the presence of diabetes. Note that this project can be taken by many students as there
are a number of variations based around different diseases, including diabetes, heart disease, stroke and cirrhosis, and can be tested using a number of different classification algorithms and
feature selection strategies.
