A Concerted Eradication Effort through AI
At Wadhwani AI, we are creating technologies to address multiple challenges across the cascade of tuberculosis care.
Tuberculosis is caused by a bacterium called Mycobacterium tuberculosis. The bacteria can attack any part of the body such as the kidney, spine, and brain, but usually attack the lungs, in which case it is contagious. If not treated properly, tuberculosis can be fatal.
In 2016, there were 10.4 million new active tuberculosis cases and 1.7 million deaths worldwide. It is estimated that as many as 50% of tuberculosis cases may go unreported and unknown.
More people die of tuberculosis than of HIV and malaria put together. Despite a strong eradication programme, almost half of all tuberculosis cases in India go undetected.
TB hotspots are traditionally identified by income and population density markers. However, through AI we have found granular markers that can identify potential TB hotspots in non-obvious locations. Once identified, the typical action is to issue vouchers to residents so that they can be diagnosed using facility-based X-rays. This requires time and effort residents often cannot justify. Currently, Wadhwani Institute for Artificial Intelligence is working on a tool that uses cough-based sound identification to diagnose and triage patients on the spot. Once the patients have been identified, the next big challenge is adherence. Using lessons we have learned from HIV patients, we are developing tools to assess adherence likelihood for individuals that depend on several contextual factors. The combination of these three solutions can potentially solve problems across the cascade of care.
Impact through AI
As the official AI partner for India’s Central Tuberculosis Division, what works to our advantage is the fact that the circumstances are conducive to our work — there already exist a centralised strategy and execution; dedicated government staff and budgets; support from organisations such as WHO; strong digital infrastructure, and multiple touch-points across a patient’s journey.
Now, we are creating technologies to address multiple challenges across the cascade of TB care, starting with case load estimation at the district level using a variety of risk and transmission factors to help identify missing cases, and the prioritisation of TB patients for health workers through stratification of the risk of drop-off from treatment