JMIR Publications

ABSTRACT

Leprosy is a neglected tropical disease (NTD) caused by Mycobacterium leprae. It predominantly occurs in areas with poor socio-economic conditions and affects the skin and peripheral nerves. Without proper treatment, leprosy can lead to severe physical deformities, making it a highly stigmatizing disease. This study evaluated four machine learning models that predict the progression of the grade of physical disability related to leprosy based on clinical and socio-demographic data sourced from a Brazilian database. The database contained notifications of leprosy cases spanning from 2001 to 2023. The objective was to predict the likelihood of an increase in the disability grade caused by the disease. After preprocessing the data, a total of 199,924 records and 12 clinical and socio-demographic variables were selected for analysis. The study evaluated the performance of four machine learning models: Decision Tree (DT), Random Forest (RF), Adaptive Boosting (AdaBoost) and Gradient Boosting (GB). The RF model had the highest performance achieving a recall of 82.59% (±0.02), followed by GB with 77.27%, AdaBoost with 73.11% and DT with 72.09%. These results suggest that the models are proficient at identifying instances of the positive class, which in this context means predicting the progression of leprosy-related disability, thereby reducing the number of false negatives. This ability is crucial in predicting the progression of the disease and potentially improving patient outcomes. This study's findings indicate that machine learning can be a valuable tool in managing leprosy, particularly in predicting the worsening of physical disabilities. By leveraging clinical and socio-demographic data, healthcare providers can better identify patients at higher risk and tailor their treatment strategies accordingly.