Key Details
The knowledge paper “AI in MedTech: Revolutionizing Healthcare Through Artificial Intelligence”, developed by Praxis Global Alliance with FICCI, sets out five priorities:
Start with diagnostics: Imaging, pathology, ECG and point-of-care screening offer measurable and relatively scalable applications.
Build AI-ready health data: Models require interoperable, consent-based and representative Indian datasets for development and validation.
Regulate throughout the product lifecycle: Oversight must cover software updates, model drift, cybersecurity and post-market performance—not only initial approval.
Recognise AI in procurement and reimbursement: Hospitals have limited incentive to adopt validated tools when payment and tender systems recognise hardware and procedures but not AI-enabled value.
Integrate AI into clinical workflows: Providers need trained clinicians, clear responsibility and systems for monitoring AI after deployment.
These are recommendations from a jointly developed knowledge paper, not measures already approved by the Government.
Diagnostics Could Extend Specialist Capacity
India's healthcare constraint is not only equipment shortages but also the concentration of specialist expertise in large urban hospitals. AI-enabled devices could extend screening, preliminary interpretation and risk assessment to primary and secondary facilities while retaining clinician oversight.
Diagnostics are a particularly suitable early application because they:
generate structured digital data;
follow relatively standardised workflows;
produce measurable clinical outcomes; and
can assist rather than replace clinical decision-making.
Applications include detecting abnormalities in medical images, identifying arrhythmias from ECGs, prioritising suspicious pathology slides and guiding less-experienced operators during ultrasound examinations.
The paper cites MadhuNetrAI, which had screened more than 7,100 patients across 38 facilities by December 2025, as an example of AI-assisted screening extending specialist capacity without requiring an ophthalmologist at every site.
India Has Digital Foundations but Limited Evidence for Scale
The Ayushman Bharat Digital Mission, IndiaAI Mission, Strategy for AI in Healthcare in India and Benchmarking Open Data Platform for Health AI provide foundations for interoperability, benchmarking and responsible deployment.
But digital infrastructure is not the same as AI-ready clinical evidence. Health data remain fragmented across hospital records, imaging systems, laboratories and medical devices, often without common standards.
Scaling requires datasets that are:
representative of India's population and disease burden;
drawn from multiple institutions and care settings;
clinically labelled and quality-checked;
governed by consent and privacy safeguards; and
available for post-deployment monitoring.
This is important because algorithms trained predominantly on foreign populations may not perform consistently across India's demographic diversity and healthcare settings.
Regulation Must Account for AI That Changes Over Time
AI-enabled medical devices fall under the Medical Devices Rules, 2017, while CDSCO's draft guidance distinguishes between standalone medical software and software embedded within a device.
The paper argues that one-time approval may be insufficient for AI systems that can be updated or retrained after deployment. It proposes:
risk classifications and validation requirements;
approved protocols for algorithm changes;
monitoring for deterioration or model drift;
cybersecurity requirements;
periodic revalidation using real-world evidence; and
clearer accountability across developers, hospitals and clinicians.
It also recommends a dedicated technical expert committee and a public registry of authorised AI-enabled medical devices, including their approved uses and supporting evidence.
Payment and Procurement Remain the Missing Adoption Link
Even clinically effective AI may struggle to scale if providers have no financial incentive to adopt it. Hospitals may receive the same reimbursement whether diagnosis uses conventional or AI-enabled equipment, leaving them to absorb the additional technology cost.
Public procurement creates a similar constraint: tenders often emphasise acquisition cost and hardware specifications rather than software quality, workflow improvements or longer-term clinical outcomes.
The paper therefore proposes:
pilot reimbursement for validated AI diagnostics targeting high-burden diseases;
faster health-technology assessment for AI-enabled devices;
procurement based on clinical and health-system value, not price alone;
recognition of domestic software and AI research as local value addition; and
use of public hospitals as anchor users to generate real-world evidence.
These remain policy proposals, requiring subsequent decisions by government, public purchasers and insurers.
What Is Lifecycle Regulation?
Lifecycle regulation monitors an AI-enabled device from development and clinical validation through deployment, software updates and post-market use. It recognises that an algorithm’s performance may change as the software, patient population or clinical environment evolves.
Policy Relevance
AI-enabled diagnostics could extend screening and decision support to district hospitals and primary-care facilities where specialists are scarce.
ABDM interoperability can support deployment, but regulatory decisions require clinically representative and independently validated Indian evidence.
Public procurement can create an adoption pathway if it measures outcomes, workflow gains and total cost rather than hardware price alone.
Reimbursement will determine whether validated AI becomes routine care or remains dependent on temporary pilot budgets.
Offline operation, multilingual interfaces and compatibility with non-specialist workflows will influence whether AI reduces or reproduces existing access inequalities.
Human oversight, post-market monitoring and clear responsibility remain central because an approved algorithm can deteriorate or behave differently after deployment.
Follow the Full Paper Here: Knowledge Paper on AI in MedTech Released