- Academics in the School of Computer Science visit Zambia to start a second project with the Centre for Infectious Disease Research in Zambia (CIDRZ), that will use AI analysis of cough sound to screen for tuberculosis.
- Further collaboration opportunities explored in the areas of apnea of prematurity and applications of AI in public health, including automatic speech recognition for Zambian vernacular languages.
TB remains a significant global health challenge, with over 10 million new cases and more than a million deaths reported annually. Most TB deaths occur among individuals with undiagnosed and untreated disease, highlighting the urgent need for systematic screening as a key strategy for early TB detection and treatment. In low-resource settings, where access to advanced diagnostic facilities is limited, the development of low-cost, scalable screening tools is essential to address this global health burden effectively.
The Sheffield team have been working closely with clinicians and research scientists at the Centre for Infectious Disease Research in Zambia (CIDRZ), a key non-governmental organisation (NGO) in the country, which is committed to improving access to high-quality healthcare in Zambia through extensive research and public health programmes.
In their initial project, the Sheffield team developed an AI-based screening tool that analyses cough sound recorded via a microphone. Using a state-of-the-art deep neural network classifier, they demonstrated the ability of the system to identify the characteristics of different cough sounds, and accurately classify coughs associated with TB, other respiratory diseases and those from healthy individuals. The project involved an extensive data collection effort from two hospitals in Lusaka, where sound recordings were complemented by chest X-rays and bacteriological testing to validate the results. The project was funded by an award from the Engineering and Physical Sciences Research Council (EPSRC) Impact Acceleration Account.
A follow-up project will extend data collection to a third hospital and investigate how TB may affect speech characteristics. Next year, the Sheffield team will also contribute to CIDRZ’s large-scale TB REACH project, funded by the Stop TB Partnership, which will assess their cough sound AI system as an approach for high-throughput screening of TB in the field.
This collaboration with CIDRZ demonstrates the transformative potential of AI to address pressing global health challenges like tuberculosis, especially in resource-constrained settings. By co-designing low-cost acoustic AI technology with local TB experts in Zambia, we aim to develop sustainable and impactful solutions to enhance early disease detection and improve health outcomes. I am particularly excited about the opportunity to evaluate this technology in real-world settings during the TB REACH project and explore its broader applications in public health."
Dr Ning Ma, principal investigator of the project, holding a joint position with the School of Computer Science and Sheffield Teaching Hospitals NHS Foundation Trust
Professor Guy Brown added: “CIDRZ is doing outstanding health research in Zambia, and our project on acoustic TB screening has been very productive. I’m delighted that our research collaboration is continuing to mature, and is developing the potential for significant impact.”
As part of capacity building and knowledge exchange, Dr Ma delivered a training session on AI for TB screening using cough sound to 25 CIDRZ clinicians and researchers. The Sheffield team and CIDRZ also discussed other opportunities for collaboration around AI for public health. These include the automatic transcription of field recordings spoken in Bemba, one of seven vernacular languages spoken in Zambia, and an AI system for monitoring infant breathing that could reduce mortality due to apnea of prematurity (AOP). The latter would build on previous approaches that the Sheffield team have developed for obstructive sleep apnea screening, using sound recordings and small sensors.
Follow the links to find out more about CIDRZ and the School of Computer Science.