Early stage HIV diagnosis using optimized ensemble learning technique

Abstract

Human immunodeficiency Virus most commonly known as HIV, is a retrovirus that attacks and weakens the immune system, leaving the body vulnerable to opportunistic infections and cancers and can be transmitted through unprotected intercourse or shared syringes. HIV can become AIDS, which is a more deadly form of HIV if left untreated. Individuals and NGOs are working hard to encourage open discussions, decrease the negative feelings around HIV, and make sure more people know and understand it better. Early prevention and proper medical care can help reduce the mortality rate and harmful effects of HIV. Machine learning applications aid in early prediction of the disease. In this work, we have proposed an optimized ensemble machine learning technique using grid search hyper-parameter optimization, to diagnose disease in early stages on the basis of learning received from the HIV dataset from Kaggle used in our experiment. The proposed approach outperformed other individual and existing models with the highest classification accuracy of 97.86%.