Multifactorial Analysis for Predicting Drug Resistance in Tuberculosis: Integrated Approaches to Improve Treatment

Tuberculosis (TB) remains a significant threat to global health, exacerbated by drug resistance, socio-economic, and environmental factors. These challenges make the treatment of the disease not only prolonged but also costly. With these issues in mind, this study introduces a multifaceted predictive model that integrates clinical, environmental, and socioeconomic factors to predict drug resistance in TB patients. Utilizing a dataset of 103,846 patient records from São Paulo, Brazil, spanning from 2006 to 2016, we applied machine learning techniques, including clustering, segmentation, and predictive modeling, to identify patterns and predictors of drug resistance. The proposed methodology enabled us to predict, analyze, and anticipate the clinical outcomes of patients in relation to TB drug resistance, with the XGBoost model achieving an accuracy of 84.65% and an F1-score of 85.34%. Moreover, the clustering of data allowed for the analysis of how comorbidities influence the likelihood of developing drug resistance in TB. Through this approach, it becomes possible to foster a better understanding of the clinical picture and diagnosis of patients, thereby facilitating more targeted and effective interventions.