Predictive modeling of CD4 count levels in HIV/AIDS patients with TB co-infection: a machine learning framework for treatment optimization

The study presents a machine learning model for examining CD4 count levels in HIV/AIDS patients with TB co-infection. This aimed to build improved treatment plans while enhancing patient health outcomes. This work contributes to Sustainable Development Goal 3 (Good Health and Well-being), particularly targets 3.3 and 3.8, by providing tools for improved HIV/TB patient care and treatment optimization. A collection of six machine learning algorithms, including linear regression, random forest, support vector machine, decision tree, gradient boosting machine, and extreme gradient boosting, was used to generate predictive models from the clinical and demographic data. Among the review models, the Random Forest algorithm produced the minimum Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and Mean Absolute Error (MAE). The most impactful factors for CD4 count levels in HIV/AIDS patients with TB co-infection, according to feature importance analyses, were weight, age, clinical stage, marital status, and educational level. Through these findings, healthcare providers now have an advanced tool to plan specific treatments and allocate resources for HIV/TB co-infected patient care, which can enhance both short-term and long-term clinical outcomes and patient quality of life.