Metaheuristic Optimization Review MOR Pending 10.54216/MOR https://www.americaspg.com/journals/show/3365 2024 2024 Improving Tuberculosis Diagnosis and Forecasting Through Machine Learning Techniques: A Systematic Review Telecommunications and Smart Systems Laboratory, University of Djelfa, PO Box 3117, Djelfa 17000, Algeria Abdelaziz Abdelaziz Amity University Kolkata, India & Amity School of Engineering and Technology, India Pushan Kumar Dutta Tuberculosis (TB) is ranked as one of the leading causes of death from infectious diseases in the present world, causing important health and economic consequences in the different developing countries. The practices of traditional diagnostic approaches, although still expected, are associated with relativity, slowness, and organs, besides being confined to visual observations and touch. The new and increased capacity in advanced machine learning is a promising area that has shown potential in improving the diagnosis of TB, as well as identifying drug resistance and disease management. This review presents various aspects of using ML in diagnosing and managing TB disease based on its various categories of models, including deep learning, hybrid approach, and the metabolomics approach. Some of these methods have been very effective, with high diagnostic performance improvements in sensitivity, specificity and accuracy; Furthermore, ML has been used to analyze the molecular picture of TB and to find drug targets of the disease toward future targeted therapies. As seen with the integration of ML, substantial benefits are provided by the solutions proposed. However, questions concerning the quality of data, interpretations of ML models and ethical problems hinder further application. This review concludes with the idea that ML can transform the diagnosis and management of TB and calls for more investment in developing this field to overcome these barriers to global health. 2024 2024 35 44 10.54216/MOR.010103 https://www.americaspg.com/articleinfo/41/show/3365