Details of the new mobile-enabled intelligent expert system, which has been developed by a team led by scientists from Anglia Ruskin University, have been published in the journal Expert Systems with Applications.
The study shows that this new mobile phone testing system has a 98.4% accuracy rate in detecting tuberculosis antigen-specific antibodies.
The app uses a regular smartphone camera to digitally capture a biosensor with sputum sample and then uses artificial intelligence to analyse the images and automatically detect TB-specific antibodies, delivering results to users within a couple of seconds.
The high rate of mobile phone use in developing countries means that it can reach a wide population, particularly benefiting those who have limited access to advanced laboratory facilities. And because the system works offline, it can be used in the most remote areas of the world.
Professor Alamgir Hossain of Anglia Ruskin University, a co-author of the paper, said: “This new system has been developed with colleagues in Malaysia which, in common with many areas of south east Asia, has a high rate of tuberculosis and extensive, remote rural areas with few medical facilities for testing.
“This system wouldn’t be used instead of seeing a physician but would be a first port of call. If the app detects tuberculosis it would instruct the user to contact a medical professional. One major advantage of our app system is that it doesn’t require internet access or any additional hardware to be added to a smartphone.
“By using the phone’s camera to capture the sample, rather than manually using colour charts, it eliminates human error and avoids any subjectivity around interpretation. It also means that positive and negative samples do not need to be distinguishable to the human eye or depend on perfect colour vision.
“Our system is a portable, cheap and accurate automatic TB diagnosis tool which can benefit millions of people, especially in remote locations where few health experts are available. The technique we have developed for this test, using a mobile phone camera, can be applied to other colourimetric qualitative tests, so it has a number of other possible uses.”
The research and development of the app has been supported by a £129,000 Institutional Links grant, funded by the Newton-Ungku Omar Fund, managed by the UK Department for Business, Energy and Industrial Strategy and delivered by the British Council. The developers hope it will be in use within the next two years.
The work also involved academics from the University of East Anglia and Universiti Putra Malaysia. The full open access journal article can be read here.
Watch a video explaining the work: