Overcoming unique, interlinked data and human factors is key to scaling AI adoption in India’s healthcare system, argues one AI expert.

AI and advanced data analytics are reshaping the global healthcare landscape, and India is no exception. The potential is immense: faster, more accurate diagnostics, streamlined hospital workflows, and deeply personalized patient care.

Yet, despite promising pilot projects in leading hospitals, broad adoption remains slow. Three systemic barriers stand out.

  • First, healthcare data remains fragmented in India, with institutions operating in silos that limit cross-platform insights.
  • Second, regulatory and governance frameworks are still evolving, leaving uncertainty around trust, accountability, and fairness in AI models.
  • Third, many AI solutions are designed for conditions that are very different from India’s clinical realities, limiting the tech’s local relevance and impact.
  • Overcoming these challenges requires more than just technology. It calls for policy alignment, a shared data infrastructure, clinician-led tool design, and the upskilling of the healthcare workforce to use AI responsibly. Only with these foundations in place can AI move from isolated use cases to becoming a core driver of equitable, patient-centric healthcare across India, according to Ankit Shrivastava, founder and Managing Partner, Enventure, in this Q&A with DigiconAsia.net.

DigiconAsia: What strategies, technologies, and governance frameworks are needed to overcome data fragmentation in Indian healthcare, and to enable equitable, interoperable AI?

Ankit Shrivastava (AS): India’s data fragmentation problem has become more than just a technical issue; it has spilled over into being an incentive problem as well.

Hospitals, labs, and insurers still see their datasets as competitive assets, not public infrastructure. Initiatives such as the Ayushman Bharat Digital Mission (ABDM) are crucial, but unless there is a tangible benefit for providers to share clean, structured data, uptake will be patchy.

The real breakthroughs will come when interoperability is tied to reimbursement rates, accreditation, or procurement eligibility, making participation a competitive advantage, not just a compliance task.

Technology can make sharing possible, but policy and economics make it probable. For AI, this shift is critical: without rich, longitudinal datasets that cross institutional boundaries, models will keep producing insights in silos. ABDM can be the technical backbone, but unless it is paired with aligned incentives and strict data quality standards, we will continue to have “interoperable” systems that in practice speak different dialects of the same language.

DigiconAsia: From explainable AI to generative outputs — how can we ensure that AI-driven recommendations in Indian healthcare are both trustworthy and ethically deployed?

AS: Trust in AI recommendations comes from relevance, not just transparency. Models must be trained on India-specific datasets, grounded in local protocols, and have to provide concise, context-aware explanations that integrate seamlessly into clinical workflows.

Generative AI tools can accelerate documentation, treatment plan drafting, and language translation — but without safeguards, risks of hallucinations/errors, over-reliance, and bias remain. Ethical deployment requires governance baked into workflows: mandatory human validation, audit trails for AI-generated content, and patient record transparency about where and how AI was used.

In short, AI should remain a “first word” assistant, never the “final word” authority.

DigiconAsia: How can predictive analytics and remote monitoring be adapted to India’s diverse patient populations to truly deliver proactive and equitable care?

AS: For predictive analytics to be truly proactive and equitable in India, the models must account for social determinants such as income, housing, and diet — not just medical data.

  • Upskilling programs should prioritize medical professionals, empowering them as informed co-pilots in AI-assisted care.
  • True collaboration means involving clinicians from the earliest design phases, ensuring tools are usable, context-appropriate, and clinically relevant.
  • Policymakers can accelerate this by tying funding, approvals, or incentives to demonstrable clinician participation in AI development.
  • Safety, scale, and effective deployment emerge when human expertise guides technological ambition.

DigiconAsia thanks Ankit for sharing his professional insights with our readers.