Harmonizing Health Data: A Machine Learning-Based Detection of Tuberculosis (TB) Co-Infection in HIV Patient’s Data in Pakistan

Abstract

Tuberculosis (TB) as a co-infection in People Living with HIV (PLHIV) is a serious public health concern, especially in underdeveloped nations like Pakistan. Robust data harmonization and sophisticated machine learning approaches are essential for the efficient management and treatment of chronic diseases. This work uses national health data from the Ministry of Health Pakistan to provide a machine learning-based method for detecting TB as co-infection in PLHIVs. To train different machine learning models, we used an extensive dataset that included patient demographics, clinical history, behaviors, lab results. The machine learning models trained on the extensive data set we used accuracy, recall, precision, and F1-score and Area Under Curve (AUC) parameters to assess the efficiency of models. According to our findings, machine learning methods can greatly improve the identification of TB co-infection in HIV patients, giving public health professionals a useful tool for tracking and containing the spread of these illnesses. Real time dashboard, data analysis and decision-making are made easier and disease detection accuracy is increased when machine learning algorithms are integrated with national health databases. This study emphasizes how machine learning can revolutionize disease management and public health surveillance in environments with limited resources.