Applying Machine Learning to Tackle the Double Burden of HIV and Diabetes in Rwanda

Background: HIV and Type 2 diabetes (T2DM) are global health challenges, and the burden of HIV is pronounced in sub Saharan Africa, with the rising prevalence of non-communicable diseases (NCDs) such as T2DM. Objective: The study was to apply machine learning techniques to explore: (i) the proportion of T2DM among people living with HIV (PLWH); and (ii) the association between HIV and diabetes. Methods: The analysis utilized a dataset of 774,189 electronic medical records (EMR) obtained from 10 healthcare facilities in Rwanda between 2019 and 2023. Machine learning models, including logistic regression, random forests, and gradient boosting machines (GBM), were applied to predict the onset of diabetes and evaluate the impact of coexistence of HIV and Diabetes. Statistical analysis was conducted to assess the performance of these models based on accuracy, precision, recall, and F1-score, alongside identifying key risk factors like age, BMI, and blood sugar levels. Results: The prevalence of T2DM in the general population was 4.71%, while among PLWH, the prevalence was significantly higher at 10.22%. Logistic regression and random forest models indicated key predictors of diabetes among HIV-positive individuals as BMI, age, and blood sugar levels. Conclusion and implication: This study underscores the complex interplay between HIV and diabetes in Rwanda. Machine learning models demonstrated high accuracy in predicting outcomes, offering valuable insights for clinical management. Integrated care models that focus on managing Diabetes Mellitus in PLWH to mitigate the risks of complications are important.