August 2, 2025Open Access

The Development and Evaluation of AI-based Tuberculosis Screening with a Digital Stethoscope used to Capture Lung Sounds. A Case-Control Study.

View Full Paper

Ask AI

Bookmark

Authors

Discussion

Case-control study finds AI screening achieves 89.9% sensitivity in TB detection, suggesting it could improve access and efficacy.

AI-based screening for tuberculosis showed a sensitivity of 89.9% and specificity of 50.4%.
A dataset of 49,770 lung sound recordings was collected from 1,169 participants across South Africa.
The model utilized K-fold cross-validation to evaluate its performance against the sputum Xpert Ultra standard.
This AI method may enhance TB screening accessibility due to its low cost and portable design.

Rath et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark