India AI Impact Summit 2026

harnessing-ai-for-equitable-and-resilient-health-systems

10 min read

TRACK:Inclusion for Social Empowerment
SPEAKERS
  • Aparnaa Somanathan, World Bank
  • Geetha Manjunath, Niramai Health Analytix
  • Kalavathi GV, Siemens Healthineers
  • Karthik Adapa, WHO-SEARO
  • Kiran Gopal Vaska, National Health Authority, GoI
  • Mamta Murthi, The World Bank Group
  • Pramod Varma, India DPIs
  • Preetha Reddy, Former UNESCO Assistant Director-General for Social and Human Sciences
SESSION OVERVIEW

This session convenes policymakers, industry, technologists, and public health experts to explore how AI can be responsibly scaled to deliver health system impact. With reference to case studies and innovation platforms, it will examine practical pathways for moving from pilots to sustainable, system-level adoption. The session will explore governance, standards, and ethical considerations for responsible AI adoption and identify pathways for inclusive and equitable AI adoption, particularly in low- and middle-income countries.

VIDEO RECORDING

Harnessing AI for Equitable and Resilient Health Systems

Abstract

The panel brought together senior leaders from multilateral institutions, a health‑tech startup, a global medical‑devices firm, the architects of India’s digital public infrastructure, a leading hospital group, and a WHO advisor to explore how artificial intelligence can be responsibly scaled across India’s health system. Topics covered concrete use‑cases that have moved beyond pilots (e.g., AI‑driven TB and breast‑cancer screening, sepsis prediction in ICUs), the infrastructural and governance foundations needed for nation‑wide adoption, the role of public‑private partnerships, and the evidence gaps that still limit confidence that AI improves health outcomes. A rapid‑fire “future‑vision” round asked each panelist to identify a single measurable change that would prove AI has moved beyond pilots, and the concrete action they would start today to achieve it. The session closed with three audience questions on cost, primary‑care equity, and regulatory frameworks, followed by a summarising call to focus on real‑world outcomes for patients.

Detailed Summary

Aparnaa Somanathan (World Bank) opened the session by highlighting concrete AI pilots she has witnessed in low‑income settings such as Ethiopia, where AI‑enabled tools empower primary‑care workers. She then pivoted to India, noting the country’s large‑scale AI‑enabled TB‑screening tools that are already improving case detection.

Key points:

  • AI is not a threat of mass job loss; rather it drives job transformation and creates new roles across the value chain.
  • India’s digital public infrastructure (e.g., Aadhaar, DigiLocker, e‑Health IDs) provides a foundation for population‑scale AI integration.
  • Scaling must be paired with equity‑focused policies – skilling, institutional capacity, financing models, and robust governance.

Aparnaa concluded by stressing the panel’s role in unpacking how scaling can happen without widening inequality.

Mamta Murthi (World Bank Group) thanked the audience and reinforced AI’s potential as an “UHC accelerator.” She highlighted the need for an inclusive, resilient health system where India acts as a global learning platform.

2. Panelist Presentations

2.1 Geetha Manjunath – Niramai Health Analytics (AI‑driven Clinical Decision Support)

  • Scope of AI at Niramai: Focus on Clinical Decision Support Systems (CDSS) that improve both operational efficiency and clinical accuracy.
  • Key Use‑Cases:
    1. QR.ai – AI analysis of chest X‑rays for TB and lung cancer detection.
    2. Niram.ai – Thermal‑imaging AI for breast‑cancer screening; deployed to ≈400,000 women; provides risk categorisation (red/yellow/green) and detailed reports.
    3. Eye‑care AI – Diabetic retinopathy detection (e.g., Remidio).
  • Implementation Challenges:
    • Literacy & IT‑savviness of frontline health workers leads to protocol errors (e.g., improper image capture).
    • Iterative feedback loops were built into the tool to give real‑time corrective guidance.
    • Clinical validation must be performed both in hospitals and in field settings with intended users.
  • Future Outlook: Emphasised the need for offline capability, robust validation, and continuous training for health workers.

2.2 Kalavathi GV (Siemens Healthineers) – AI for Imaging Standardisation & Rural Access

  • Global Context: 4.6 billion people lack basic health services; projected 10 million clinician shortage by 2030.
  • Strategic Focus: AI must be embedded in the entire care continuum (early detection → referral → follow‑up).
  • Three Pillars of Impact:
    1. Imaging Standardisation – My Exam Companion: AI‑enabled software that guides technicians to acquire standardized scans, speeding diagnosis.
    2. Decision Support – AI‑enabled Companion: Assists radiologists/oncologists with rapid contouring and treatment planning.
    3. Access Expansion – Single Virtual Cockpit: Remote expert guidance for rural facilities lacking specialists.
  • Success Recipe: Augment clinicians, embed AI in workflow, and plan for scale.

2.3 Pramod Varma – Networks for Humanity (Digital Public Infrastructure)

  • Perspective: As the only non‑health‑sector expert, he framed AI as a solution to the systemic shortage of clinicians for 1.4 bn Indians.
  • Learning from Aadhaar/ABDM:
    • Avoid “one‑size‑fits‑all” solutions; instead, build a “forest” of interoperable building blocks that enable thousands of innovators.
    • Separate foundational infrastructure from application layers – the fabric (e.g., UID, health IDs) must be stable, while apps can evolve.
    • Don’t confuse scaling a pilot with scaling what works – pilots that succeed in a few villages rarely survive at national scale without adaptivity.
  • Action Items: Continue to lay the digital groundwork (foundational models, open data ecosystems) that allow private‑sector innovators to plug‑and‑play AI solutions.

2.4 Preetha Reddy – Apollo Hospitals (Clinical Evidence & Outcome Focus)

  • Current Deployments:
    • Sepsis prediction in large ICUs (≈2,000 beds, with a plan to reach 10,000). Early detection is life‑saving and ready for scale.
    • Geriatric care AI – Predicting falls and nutrition‑related risks.
    • Drug‑interaction detection – AI reduces polypharmacy and adverse events while cutting drug costs.
  • Scalability Driver: Emphasised the need for robust clinical validation and integration into existing hospital workflows to move from isolated pilots to system‑wide adoption.

2.5 Kiran Gopal Vaska – National Health Authority (Public‑Sector Partnership & Regulation)

  • Digital Foundations: Cited the ABDM (Ayushman Bharat Digital Mission) and earlier DPI as the “rails” for a national AI ecosystem.
  • Ecosystem Building:
    • Multi‑stakeholder collaboration among state governments, academia, and private firms.
    • Clear pathway: Lab → Pilot → Scale, with an emphasis on bridging knowledge gaps between innovators and regulators.
  • Regulatory & Safety Measures:
    • MOUs with IASC Bangalore & IIT Kanpur for research collaboration.
    • Unveiling a national AI‑for‑health strategy and a benchmarking platform for AI solutions.
    • Emphasised “innovation over restraint” while maintaining ethical, safe use.
  • Visionary Note: Cited the film Interstellar (Indian‑made drone) as a metaphor for India’s ambition to build its own foundational AI models.

2.6 Karthik Adapa – WHO‑SEARO (Global Evidence & Standards)

  • Evidence Gap: Only ~20 % of AI research studies progress from lab to clinical setting; of those, ~10 % reach population‑scale deployment.
  • Root Causes: Lack of standards, auditability, regulation, and governance – core WHO responsibilities.
  • Pilotitis Problem: Donors favor “shiny pilots” over building foundational guardrails.
  • Strategic Recommendations:
    1. Define governing principles for AI in health.
    2. Leverage existing national digital platforms (e.g., Nikshay for TB) instead of creating siloed apps.
    3. Develop data‑centric ecosystems that enable evidence generation and scalable implementation.
  • Global Context: Even high‑resource health systems (e.g., Mayo Clinic, Stanford) struggle to translate AI from research to bedside, highlighting that the scaling challenge is universal.

3. Rapid‑Fire Future‑Vision Round

The moderator posed a single question to each panelist:

“If we reconvene three years from now, what single measurable change would prove AI in health has moved beyond pilots to delivering equitable system‑wide impact? What is the one action you can start today to make that change happen?”

3.1 Pramod Varma (≈1.5 min)

  • Measurable Change: Existence of an open AI‑trial network (a “Play Store” for AI health tools) that enables rapid, regulated trials with controlled, ethical data access.
  • Immediate Action: Invest in portable health‑record infrastructure and open‑network layers that allow startups to test and iterate AI solutions within months, not years.

3.2 Kiran Gopal Vaska (≈1 min)

  • Measurable Change: Universal validation of any AI tool before patient use – a certification badge guaranteeing safety, equity‑by‑design, and auditability.
  • Immediate Action: Establish multidisciplinary validation teams (clinical + technical) that review each AI tool’s context‑specific performance and ethical compliance.

3.3 Preetha Reddy (≈1 min)

  • Measurable Change: A national tally of patients whose outcomes improved because AI was used, captured through a mandatory reporting field in electronic health records.
  • Immediate Action: Integrate an “AI‑used” flag into hospital information systems and start aggregating outcome data.

3.4 Kalavathi GV (Siemens) (≈1 min)

  • Measurable Change: Reduction in diagnostic turnaround time by ≥30 % across rural imaging centers, translating into measurable cost savings and increased access.
  • Immediate Action: Deploy My Exam Companion at a pilot set of district hospitals and track time‑to‑diagnosis.

3.5 Geetha Manjunath (Niramai) (≈1 min)

  • Measurable Change: Screening coverage of ≥70 % of eligible women for breast cancer in selected states, with AI‑driven risk stratification integrated into ASHA workflows.
  • Immediate Action: Co‑design an ASHA‑friendly protocol and deliver targeted training plus a low‑tech UI on mobile devices.

3.6 Karthik Adapa (WHO) (≈1 min)

  • Measurable Change: Global adoption of a WHO‑endorsed AI standards framework with ≥50 % of member states reporting compliance.
  • Immediate Action: Publish a concise, actionable standards checklist and run pilot workshops with ministries of health.

4. Audience Q&A

4.1 Question – Cost & Equity (Rohit, American Express)

  • Concern: How can AI lower the cost of healthcare tools that remain expensive, ensuring accessibility for low‑income populations?

  • Panel Synthesis:

    • Scale economies – Cloud‑based AI APIs reduce per‑user costs as usage rises (Kiran).
    • Workflow integration – Embedding AI into existing processes (e.g., ASHA’s routine tasks) avoids additional infrastructure costs (Geetha).
    • Regulatory simplification – Tiered certification reduces time‑to‑market for low‑risk tools, cutting development expense (Kiran).

4.2 Question – Primary‑Care Focus (Priyanka Shrestha, WHO)

  • Concern: AI tends to concentrate in secondary/tertiary hospitals; how to ensure it strengthens community‑based primary health care (PHC)?

  • Panel Response:

    • Digital public infrastructure (ABDM, DigiLocker) already reaches primary‑care level; AI built on these layers can be deployed via ASHA workers (Kiran).
    • Low‑risk AI tools (screening, triage) need lighter regulatory pathways, making them easier to certify for PHC use (Kiran).
    • Evidence‑based pilots should deliberately target PHC metrics such as dropout reduction in chronic‑disease follow‑up (Preetha).

4.3 Question – Regulatory Framework (Rajeeb Sekka, Tech Founder)

  • Concern: India lacks a comprehensive AI‑in‑health policy akin to HIPAA; many apps ignore existing telemedicine guidelines.

  • Panel Response:

    • DPDP Act (Data Protection) provides a privacy umbrella; the Health Data Management Policy is under development.
    • Capacity‑building is essential: regulators (e.g., CEDESCO) need resources to review AI solutions efficiently (Kiran).
    • Risk‑based approach – Class‑C devices (e.g., AI‑enabled screening) undergo a lighter, faster clearance compared with autonomous surgical robots, enabling quicker market entry (Kiran).
    • Benchmarking platform announced by the NHA will serve as a pre‑market assessment repository, fostering transparency.

4.4 Additional Points from the Discussion

  • Economic Argument: Cloud deployment and higher throughput (e.g., 70 % faster scan interpretation) lower per‑patient cost, making AI financially viable for public health programs.
  • Equity‑by‑Design: Emphasised throughout—AI models must be trained on diverse Indian datasets to avoid bias.
  • Governance Need: Continuous auditability, drift monitoring, and safety certification (Health‑CERT) were highlighted as essential for long‑term trust.

5. Closing Remarks

The moderator returned to the opening vignette of a community health worker whose effectiveness hinges on data standards, training, and trust. The final message underscored that the true measure of AI is not algorithmic sophistication, but tangible health outcomes for patients, such as early detection of pre‑eclampsia or sepsis. Gratitude was expressed on behalf of the World Bank and WHO, and the session was adjourned.

Key Takeaways

  • Foundational Infrastructure First: Successful AI scaling in health requires stable, interoperable digital public infrastructure (ABDM, Aadhaar) that can host a multitude of AI applications.
  • Evidence‑Based Scaling: Only a small fraction of AI pilots reach population scale; robust clinical validation, real‑world field testing, and outcome measurement are non‑negotiable.
  • Equity‑by‑Design: AI tools must be developed and validated on diverse Indian datasets, with built‑in safeguards for bias, privacy, and accessibility.
  • Public‑Private Ecosystem: Partnerships between government (NHA, MoHFW), multilateral bodies (World Bank, WHO), academia, and startups are essential to create clear pathways from lab → pilot → scale.
  • Regulatory Evolution: A risk‑based, capacity‑enhanced regulatory framework (e.g., tiered certification, CEDESCO reviews) will accelerate safe deployment while preventing “pilotitis.”
  • Metrics for Success: Panelists agreed on measurable indicators such as (i) an AI‑trial marketplace, (ii) an “AI‑used” flag in health records, (iii) ≥30 % reduction in diagnostic turnaround, and (iv) nation‑wide coverage thresholds for specific AI‑driven screenings.
  • Actionable Next Steps: Immediate investments in portable health records, open trial networks, multidisciplinary validation teams, and cloud‑based scaling are the concrete actions each speaker pledged to start today.
  • Cost‑Reduction Pathways: Economies of scale from cloud APIs, faster workflows, and integration into existing primary‑care tasks can substantially lower per‑patient costs, addressing equity concerns.

These takeaways capture the consensus that AI must be anchored in solid digital foundations, governed responsibly, and measured by real health outcomes before it can fulfill its promise of equitable, resilient health systems in India and beyond.

See Also:

Graph View

Backlinks