India pioneers a Health AI benchmarking platform to ensure AI models are reliable and safe for healthcare applications within the country, setting a precedent for the Global South.
India is among the first countries in the Global South to develop an open Health AI benchmarking platform to evaluate artificial intelligence models on diverse Indian datasets before deployment at a population scale, National Health Authority (NHA) CEO Dr Sunil Kumar Barnwal said on Friday.
Officials said the initiative aims to strengthen AI-driven claims management under the Ayushman Bharat Pradhan Mantri Jan Arogya Yojana (AB PM-JAY), which processes nearly 40,000 claims daily across more than 1,900 treatment packages.
Speaking at the AB PM-JAY Auto-Adjudication Hackathon showcase at the Indian Institute of Science (IISc), Bengaluru, Barnwal said AI applications in healthcare must be validated in the Indian context to ensure reliability, safety and trust.
"India is among the first countries in the Global South to come up with a Health AI benchmarking platform," he said.
Barnwal said healthcare AI requires greater responsibility than any other sector because "in finance, you may lose money but in health, you risk trust and lives".
He stressed that AI systems trained on foreign datasets may not work effectively in India due to differences in disease patterns, demographics and healthcare practices.
The NHA chief said the platform would allow AI applications to be benchmarked transparently on Indian datasets before large-scale deployment, helping policymakers, healthcare providers and citizens assess whether such tools are suitable for Indian conditions.
According to NHA, the benchmarking initiative is being developed in collaboration with IIT Kanpur under a federated architecture in which data remains at the source and moves only with patient consent, ensuring privacy and data sovereignty.
Barnwal said India's digital public health infrastructure created under the Ayushman Bharat Digital Mission (ABDM) provides the foundation for scaling responsible AI solutions in healthcare.
The two-day hackathon, organised by the NHA in collaboration with the IndiaAI Mission and IISc Bengaluru, has drawn over 3,500 participants. Finalists presented AI and machine learning solutions for clinical document classification, radiology-based validation and detection of forged or AI-generated medical records.
The event also featured discussions on AI-driven claims adjudication, fraud detection, transparency, accountability and data privacy in public health insurance systems.