India AI Impact Summit Insights
All sessions

National AI Strategy for Health: Vision, Policy, and Impact

Contents

Executive Summary

This panel discussion convenes senior officials from India's Ministry of Health and Family Welfare, National Health Authority, state governments, and private healthtech companies to examine how AI-enabled solutions are being implemented in public health systems, with emphasis on scaling from pilots to sustainable deployment across diverse state health systems. The central thesis is that AI in healthcare must be viewed as an ecosystem-wide transformation—not isolated projects—with success contingent on addressing governance, regulatory capacity, change management, and alignment with public health objectives rather than purely commercial interests.

Key Takeaways

  1. Ecosystem design, not technology procurement, determines success. Solutions must align with government procurement norms, integrate into existing workflows, pass governance guardrails (privacy, transparency, bias mitigation), and address real state-level health challenges—not simply introduce cutting-edge AI.

  2. Clear, accessible government pathways are essential for innovation scaling. Startups need transparent guidance on accessing 180,000+ health facilities, program funding mechanisms beyond tenders, and collaboration models. The absence of this information actively prevents innovations from scaling despite awards and strong early evidence.

  3. Private sector impact requires shifting from transactional to programmatic partnerships. Moving beyond device sales to co-designed public health programs, post-deployment capacity building, utilization monitoring, and incentive alignment creates sustainable adoption and builds the ecosystem Sunil Kumar Banwal emphasized.

  4. Evidence + regulatory approval + implementation feasibility assessment = adoption. Government adoption accelerates when solutions have published evidence, regulatory clearance (CDSO), health technology assessment, and demonstrated implementability for frontline workers—not just clinical efficacy.

  5. AI in healthcare is a life-cycle commitment, not a one-time deployment. Governance, bias monitoring, performance tracking, and decommissioning must span the full operational life of solutions. This requires sustained funding and institutional accountability, shifting how both government and private sector plan AI initiatives.

Key Topics Covered

  • Digital transformation of India's health ecosystem — Evolution from record digitization to interoperable systems (Aayushman Bharat Digital Mission)
  • AI integration in clinical workflows — Clinical decision support systems, telehealth platforms, and AI-assisted diagnostic tools
  • State-level implementation challenges — Procurement timelines, data access, workforce capacity, and sustainability
  • Public-private partnership models — Roles, incentives, and collaboration mechanisms between government and startups/private sector
  • Scaling barriers for private innovators — Pilot-to-scale transition, funding gaps, unclear government procurement pathways
  • Governance and regulatory frameworks — Need for trust-by-design, privacy-by-design, bias mitigation, and lifecycle management of AI solutions
  • Health-specific AI use cases — Maternal mortality prediction, TB detection (handheld X-rays), diabetic retinopathy screening, breast cancer detection
  • Preventive care and health economics — Cost-benefit analysis, early detection economics, and reimbursement policy
  • Capacity building and workforce enablement — Training healthcare workers, reducing administrative burden on clinicians, and knowledge transfer

Key Points & Insights

  1. Ecosystem over projects: AI must be embedded into healthcare delivery as an integral component, not treated as isolated initiatives. Success requires coordinated participation from government institutions, private health providers, startups, innovators, and academic centers—with open standards and interoperability as foundational.

  2. Problem-driven deployment is key: The most scalable AI solutions are those addressing identified state/district-level problems (e.g., specialist shortages, maternal mortality, TB detection, administrative burden). Interventions using "Problem-Driven Iterative Adaptation" (PDIA) with early feedback loops show stronger adoption and sustainability.

  3. Measurable impact drives adoption: Quantified outcomes—50% reduction in maternal deaths (Meghalaya), 85-99% cost reduction in screening (Kerala diabetic retinopathy program), 110 TB cases identified with cost savings of ₹4 crores—create political will and frontline worker motivation necessary for continuation.

  4. Private sector faces structural barriers: Startups encounter fragmented decision-making (health is a state matter), undefined procurement pathways, expectation of free pilots, limited knowledge of government infrastructure access points (e.g., 180,000+ Aayushman Arogya Kendras), and unclear program funding mechanisms beyond individual tenders.

  5. AI must enable, not replace, human judgment: Solutions should reduce clinician/worker burden (administrative tasks, workflow optimization), amplify specialist capacity, and maintain human-in-the-loop decision-making. AI should mitigate—not reinforce—existing health inequities and biases.

  6. Regulatory and governance frameworks lag deployment: Current frameworks lack pre-deployment governance mechanisms. Solutions require: trust-by-design and privacy-by-design; health technology assessment beyond cost-effectiveness (implementation feasibility); lifecycle monitoring post-deployment; and decommissioning protocols.

  7. Data fragmentation and standardization remain critical: While ABDM provides open standards and interoperability, achieving data integration across public facilities, accessing real-world patient data for model training, and ensuring consent/security mechanisms remain substantial challenges.

  8. Post-implementation sustainability neglected: Many solutions focus on pilot outcomes but lack programmatic approaches to capacity building, utilization monitoring, stakeholder incentive alignment, and registry systems necessary for long-term operation. Private sector must build ongoing support into business models.

  9. Publication and evidence strengthen procurement decisions: Evidence generation and peer-reviewed publication (e.g., diabetic retinopathy algorithms validated on 1.2 million UK NHS images) facilitate policy-level adoption and reimbursement. Governments prefer evidence-backed solutions with demonstrated generalizability.

  10. Preventive care funding and reimbursement gaps: Current health financing (PM-JAY, ABDM) emphasizes treatment; preventive screening solutions struggle to find sustainable funding mechanisms despite demonstrated cost-benefit. Policy must integrate preventive care budgets and reimbursement pathways.

Notable Quotes or Statements

Puna Shvastava (Secretary, Ministry of Health & Family Welfare):

"The cornerstone of this discussion has to be the principles laid down by the Prime Minister: digital public infrastructure is a tool for inclusion and equity."

"AI should reduce the burden on our healthcare workforce, complementing the physician-patient relationship that remains at the heart of healthcare delivery."

Dr. Sunil Kumar Banwal (CEO, National Health Authority):

"If we just take it as a project, I don't think we will be able to actually leverage the full potential of AI. We need to move from project implementation to enable ecosystem development."

"At population scale, governance has to be by design—not deployment and then governance. It must have trust-by-design and privacy-by-design."

Satkunmar (Principal Secretary, Meghalaya):

"By effectively using data and analysis, we could able to see tremendous motivation for frontline functionaries in saving lives of mothers. We could able to reduce maternal deaths by over 50% in the last 5 years." (PDIA approach)

Dr. Gita Manjunath (Founder, Nira Health Analytics):

"The pathway [from innovation to deployment] is not clear. We don't know how to get into those 180,000 facilities...that pathway is not clear to our minds and probably there is one but it's not communicated to the right people who are innovating."

"What is killing so many innovations from scaling is we're just stuck doing pilots, pilots, pilots."

Dr. Anand Saraman (CEO, Remedio Innovative Solutions):

"We need to work with the government not transactionally in terms of a tender. We have to work in terms of how do you build programs and sustain programs."

"When you design under constraints in India, you are able to get the design right so that it is globally relevant."

Speakers & Organizations Mentioned

Government Officials:

  • Puna Shvastava — Secretary, Ministry of Health & Family Welfare, Government of India
  • Dr. Sunil Kumar Banwal — CEO, National Health Authority (NHA), IAS 1997 batch
  • Satkunmar — Principal Secretary & Development Commissioner, Government of Meghalaya, IAS 1997 batch
  • Dr. Push Singla — Secretary, Information Technology Department, Government of Jammu & Kashmir, IAS 2012 batch
  • Kiran Gopal Vasca — Joint Secretary & Mission Director, Aayushman Bharat Digital Mission, National Health Authority, IAS 2008 batch

Private Sector:

  • Dr. Gita Manjunath — Founder, CEO, CTO, Nira Health Analytics (thermal imaging-based breast cancer screening)
  • Dr. Anand Saraman — CEO, Remedio Innovative Solutions (smartphone-based retinal imaging and diabetic retinopathy screening)

Government Programs/Initiatives:

  • Aayushman Bharat Digital Mission (ABDM) — National interoperable digital health ecosystem
  • Aayushman Bharat — Large-scale health insurance and public health program
  • Aayushman Arogya Kendras — Upgraded primary health centers (180,000+ across India)
  • e-Sanjeevani — Telemedicine platform integrated with AI-assisted clinical decision support
  • PM-JAY (Pradhan Mantri Jan Arogya Yojana) — Health insurance scheme
  • National Health Authority (NHA)
  • National Center for Disease Control (NCDC)
  • National Health Mission (NHM)
  • ICMR (Indian Council of Medical Research) — Health technology assessment

Academic/Research Institutions:

  • AIIMS Delhi
  • PGI Chandigarh
  • AIIMS Rishikesh
  • IIT Jammu
  • Skims (Sher-i-Kashmir Institute of Medical Sciences), Kashmir

AI/Health Tech Solutions Mentioned:

  • Madu Netra AI — Diabetic retinopathy screening (non-specialist deployment)
  • AI-enabled handheld X-rays — TB detection
  • Cough against TB — Screening tool (1.6 lakh+ screened)
  • Media Disease Surveillance System — AI-assisted disease alert integration
  • Clinical Decision Support Systems (CDSS) — Integrated with telehealth platforms
  • Nira Health Analytics thermal imaging — AI-detected breast health assessment (1.3% mammography screening baseline in India)

Technical Concepts & Resources

AI/ML Frameworks & Approaches:

  • Problem-Driven Iterative Adaptation (PDIA) — Experimental, feedback-based approach to solution design and deployment
  • Privacy-preserving algorithms — On-device AI inference to avoid cloud costs and data transmission (e.g., Remedio's smartphone-based retinal analysis)
  • Human-in-the-loop decision-making — AI as clinical decision support, not autonomous decision-making
  • Bias mitigation and equity assessment — Built-in fairness mechanisms for population-scale deployment

Standards & Governance:

  • Open standards and interoperability — ABDM framework enabling cross-facility AI application deployment without data centralization
  • Trust-by-design and privacy-by-design — Governance principles for AI deployment
  • Lifecycle management — Deployment, monitoring, and decommissioning protocols

Health Technology Assessment:

  • Cost-effectiveness analysis — 85-99% cost reduction in screening programs (diabetic retinopathy example: ₹4 crore saved in TB handheld X-ray deployment)
  • Health economics evaluation — Publications demonstrating clinical sensitivity, specificity, and real-world generalizability
  • Regulatory approval pathways — CDSO (Central Drugs Standard Organization) predicate device mechanisms

Data & Implementation:

  • Digital health records integration — 859 million ABHA (Aayushman Bharat Health Account) IDs with 878+ million linked health records
  • Telehealth utilization metrics — 449 million consultations through e-Sanjeevani, 2.2 lakh registered providers
  • Registry systems — Post-implementation disease registries for chronic disease management and loop closure
  • Utilization dashboards — Real-time monitoring of health facility technology adoption and performance

Published Evidence Referenced:

  • UK NHS validation study on diabetic retinopathy algorithms (1.2 million images) — Best clinical balance among international models
  • Kerala government diabetic retinopathy screening program outcomes — 85% cost reduction, 99% disease detection rate
  • Maternal mortality reduction case study (Meghalaya) — 50% reduction using predictive data analytics + socioeconomic determinants

Measurement & Scale Indicators:

  • Facility coverage: 180,000 Aayushman Arogya Kendras; 15,000+ OPD capacity per major facility
  • Population reach: 300 million at-risk for diabetic retinopathy; 7,000+ maternal mortality reduction beneficiaries; 90,000 TB screening in 2 years
  • Specialist distribution gap: 1 ophthalmologist per 200,000 population; 1 radiologist for TB screening in resource-constrained areas

End of Summary