A MACHINE learning model is transforming tuberculosis (TB) screening for people with HIV (PWH), outperforming traditional diagnostic methods in predicting who will develop active TB. A new study analyzed data from the Swiss HIV Cohort Study (SHCS) and the Austrian HIV Cohort Study (AHIVCOS), demonstrating that AI-powered predictions could significantly improve early detection and preventive treatment strategies.
TB remains a leading cause of mortality among PWH, yet current screening methods often fail to identify those at the highest risk of progression to active disease. Researchers trained a random forest model using clinical data collected at HIV-1 diagnosis, incorporating information from 55 individuals who developed TB within six months and 1,432 matched controls from SHCS. The model achieved an impressive area under the receiver operating characteristic (ROC) curve (AUC) of 0.83 (95% CI 0.8-0.86) in the initial dataset.
External validation using AHIVCOS data—comprising 43 TB cases and 1,005 controls—confirmed the model’s robustness. After refining parameters and adjusting for patient demographics, the model maintained AUC values of 0.72 (SHCS) and 0.67 (AHIVCOS). Notably, the AI-based approach outperformed standard TB screening methods, achieving a lower number needed to diagnose (1.96 vs. 4), indicating greater efficiency in identifying high-risk individuals.
These findings suggest that integrating machine learning into routine HIV care could enhance TB prevention with minimal additional costs or data collection burdens. By refining risk stratification, this approach may allow for earlier and more targeted preventive treatment, ultimately reducing TB-related morbidity and mortality in PWH. As AI-driven healthcare solutions continue to evolve, their potential to transform infectious disease management becomes increasingly evident.
Reference: Bartl L et al. Machine Learning-Based Prediction of Active Tuberculosis in People with HIV using Clinical Data. Clin Infect Dis. 2025:ciaf149. doi: 10.1093/cid/ciaf149. [Online ahead of print].
Anaya Malik | AMJ
