India accounts for nearly a quarter of the world’s tuberculosis (TB) cases, with more than 2.8 million new infections and around 300,000 deaths each year, according to the World Health Organization. Although the national TB elimination programme has made steady progress, identifying people who show no symptoms or live in remote areas remains one of its biggest challenges.
Artificial intelligence is beginning to address that gap. Working with state TB divisions, Wadhwani AI has developed two AI-powered tools that enable health workers to identify vulnerable populations and detect infections earlier, reducing the chances of the disease spreading unnoticed.
The approach marks a shift from waiting for patients to seek medical care to actively searching for those most at risk. “Finding missing TB patients requires proactive identification. AI can help public health systems focus efforts where the need is greatest and enable earlier detection at the last mile,” said Shekar Sivasubramanian, head of Wadhwani AI.
One of the solutions, Vulnerability Mapping for Tuberculosis (VMTB), uses AI-driven geospatial analytics to predict areas where TB is more likely to occur. Rather than relying only on notified cases, the platform analyses TB data alongside more than 20 geospatial and health-related indicators, including population density, air quality, proximity to healthcare facilities and other social and health determinants. The information is processed using a 500×500-metre grid to generate highly localised vulnerability rankings for villages and urban wards.
The predictions allow health authorities to identify high-risk pockets and concentrate active case-finding where undetected infections are most likely to exist, said Neeraj Agrawal, chief programme officer at Wadhwani AI. Instead of spreading limited resources across large areas, officials can improve micro planning and direct healthcare workers towards communities that need the greatest attention.
Integrated with the government’s Ni-kshay portal, VMTB generates vulnerability predictions across India. Evaluations in Haryana and Assam have shown a significant improvement in TB case detection after the platform was introduced.
While VMTB helps officials determine where to focus screening, the second tool – Cough Against TB (CATB) – supports frontline health workers in identifying people who may need further testing. The Android-based app analyses cough sounds along with basic symptoms to screen for presumptive pulmonary TB. Because it works offline, it is particularly valuable in remote villages where internet connectivity is unreliable.
Using the app, a health worker first obtains consent, registers the individual and records basic details. The person is then asked to cough three times into the phone. Trained on India-specific data, the AI analyses the cough sounds together with reported symptoms before classifying the individual as either presumptive or non-presumptive TB. Those identified as presumptive cases are referred for confirmatory tests, while their Ni-kshay records are created or updated to ensure continuity within the national TB programme.
Agrawal said the app extends screening beyond conventional symptom-based methods by identifying people who might otherwise be overlooked. Earlier diagnosis allows treatment to begin sooner while reducing the risk of further transmission, particularly in underserved communities.
CATB has now been deployed across 17 states and 148 districts, screening more than 175,000 people between the ages of seven and 100. One of its most significant outcomes is that between 12% and 16% of TB cases detected through the programme were asymptomatic – individuals who may otherwise have remained undiagnosed while continuing to spread the infection.
Together, the two AI tools make active case-finding more precise. VMTB identifies where health authorities should concentrate their efforts, while CATB helps frontline workers decide whom to refer for confirmatory testing. The combination enables limited public health resources to be deployed more effectively while increasing the chances of detecting infections before symptoms become apparent.
“Technology alone cannot eliminate TB, but it can help us find cases earlier and target interventions more effectively,” said Agrawal. “That is where AI can make the biggest difference.”