CAGE-TB
Automated smartphone-based cough audio classification for rapid tuberculosis triage testing (Cough Audio triaGE for TB)
Objective
- To promote the adoption of mobile health-based cough radio triage testing for active pulmonary tuberculosis in health facilities in high burden settings.
- To generate and separately validate a cough audio classifier that meets WHO triage test TPP sensitivity and specificity criteria.
- To produce data on potential cost savings of mHealth triage testing.
- To package the underlying technology into an easy-to-use smartphone app built using human-centred design ready-for-use in large-scale clinical evaluations.
- To facilitate significant and diverse capacity building, including training North-South co-supervised African students, post-graduates and post-doctoral fellows in digital signal processing for respiratory health, diagnostic accuracy evaluations, costing analyses and implementation research, and mHealth.
Description
In Africa, many tuberculosis (TB) cases remain undiagnosed. One of the highest TB control priorities is thus a triage test in priority populations to systematically identify people requiring essential confirmatory testing. Mobile health (mHealth)-based cough audio classification represents a potential holy grail for triage testing. No specimens are collected, costs per patient is negligible, and inexpensive smartphones have high-quality microphones and computational power to rapidly analyse audio on-device. Although ambitious and at an early stage in the technology lifecycle, CAGE-TB is predicated on our proof-of-concept work, which shows that TB patients have a distinct sounding cough compared to healthy people, as well as that from other respiratory diseases. CAGE-TB’s three key research components (1. cough audio signature discovery; 2. independent audio classifier validation and calculation of potential provider costs averted; 3) implementation research to identify cross-cutting barriers and facilitators) will inform the development of a mobile application that will be capable of on-device (offline) classification of TB in clinical trials.
Partners
Stellenbosch University, South Africa
Georg-August-Universitat Gottingen Stiftung Offentlichen Rechts (UGOE), Germany
Makerere University, Uganda
Funders
European & Developing Countries Clinical Trials Partnership (EDCTP)
Countries
Germany
Netherlands
Uganda
South-Africa