“A Novel Regularized Deep Learning Approach for Tuberculosis Risk Prediction”

Effective treatment of tuberculosis (TB) requires prompt and precise risk assessment, since it continues to be a major worldwide health concern. Labor-intensive and subjective approaches are frequently used in traditional methods. The neural network model for Tuberculosis risk categorization which includes the following categories: No_Risk, Low_Risk, Moderate_Risk, and High_Risk is found in this paper. Regularization, Batch normalization and the use of Reversed Linear unit (ReLU) activation functions is adopted by the model to enhance learning process. After successfully conducting the sophisticated and optimal simulation the proposed model attains 91% classification accuracy using a variety of physiological and clinical parameters. This potentially presents future deep learning techniques to automatically rate people for possible tuberculosis occurrence thus delivering a strong tool to medical practitioners to offer distinctive intervention techniques. With deep learning, the model is capable of detecting a very complex pattern and correlation between data sets inputs and its output, thus enhancing its predictive precision as well as offering recommendations that may assist in the early detection of actual clinical situations.