Machine Learning Models Predict Drug Resistance in Tuberculosis with High Accuracy

Strategic Insights -

Research from the Sri Ramachandra Institute of Higher Education and Research has shed new light on the use of machine learning models to predict drug resistance in tuberculosis, a global health crisis that claimed 1.25 million lives in 2023. The study found that ensemble machine learning models can accurately predict resistance to key anti-TB drugs, with accuracy varying from 66% to 91.37% across different genes associated with TB resistance. The research holds promise for developing more effective diagnostic tools and treatments for tuberculosis.

Key Takeaways:

  • Researchers at the Sri Ramachandra Institute of Higher Education and Research developed machine learning models to predict drug resistance in tuberculosis, using a comprehensive dataset of genomic variants and mutations associated with resistance phenotypes.
  • The study evaluated the performance of ensemble machine learning models (Stacking, Bagging, and Voting Classifiers) on six TB resistance genes (gyrA, gyrB, inhA, katG, rpoB, and pncA), achieving accuracy ranging from 66% to 91.37% and ROC scores varying from 0.69 to 0.92.
  • The best-performing model for each gene was chosen, emphasizing a gene-specific approach to maximize resistance prediction.
  • The study concluded that gene-specific ensemble models supported by a comprehensive feature set can provide valuable predictions of drug resistance in M. tuberculosis, but requires further validation on larger and more diverse clinical datasets.
  • The research has been peer-reviewed and published in the journal Computational Biology and Chemistry, with a corresponding author A. T. Subalakshmi from the Department of Bioinformatics at the Sri Ramachandra Institute of Higher Education and Research.

Statistics:

  • 10.8 million cases of tuberculosis worldwide in 2023 (NewsRx, 2025).
  • 1.25 million deaths from tuberculosis in 2023 (NewsRx, 2025).
  • 66% to 91.37% accuracy in predicting resistance to key anti-TB drugs using ensemble machine learning models (Computational Biology and Chemistry, 2025).
  • 0.69 to 0.92 ROC scores for different ensemble machine learning models (Computational Biology and Chemistry, 2025).

Sources:

  • NewsRx. (2025, October 21). Researchers at Sri Ramachandra Institute of Higher Education and Research Report Findings in Tuberculosis (Machine learning approaches to predict drug resistance in tuberculosis). TB & Outbreaks Week, 3905.
  • Computational Biology and Chemistry. (2025). Machine learning approaches to predict drug resistance in tuberculosis, 120, 108705.

Elsevier Sci Ltd. (Elsevier - www.elsevier.com; Computational Biology and Chemistry - www.journals.elsevier.com/computational-biology-and-chemistry/)