Tuberculosis (TB) is the world’s deadliest infectious disease, causing 1.5 million deaths and 10 million new cases annually, despite being both preventable and treatable. This project proposes a low-cost, non-invasive device that leverages cough sound analysis for real-time TB screening and triage. The system automatically detects and segments coughs for embedded analysis while preserving patient data privacy and operating on low power for extended use. Applications include
- community-based active case finding,
- deployment in public spaces for early alerts,
- hotspot surveillance for resource planning,
- clinical triage in low-resource settings, and
- monitoring patient progress during and after treatment.