State of the Art Survey on Artificial Intelligence-Based Approach for Early Diagnosis of Tuberculosis

Tuberculosis (TB) has been affecting people and high deaths from it has been recorded globally. Techniques around use of artificial intelligence (AI) based approaches are gaining momentum in early diagnosis of TB. This research has managed to uncover current approaches that are being used in TB diagnosis. It was found out that the AI-based approaches are offering better accuracy especially if the data sets used to train the models are high. Combining various AI-based models was observed to offer better accuracy, i.e., the ensemble approach. The pitfall of the current AI-based models are that they lack interpretability since they are black box and thus cannot offer decision on how the have arrived at the decision, which is a critical area to consider in the medical field. This pitfall can thus be addressed by utilising explainable artificial intelligence, and this has been identified as a future area in this study.