The National Clinical Decision Support System: How AI Is Being Rolled Out Across 70,000 Hospitals

In late December 2025, India’s National Health Authority directed states and Union Territories to activate an AI-powered Clinical Decision Support System across nearly 70,000 public and private hospitals — one of the largest single deployments of clinical AI anywhere in the world. Built by AIIMS New Delhi and distributed through the Ayushman Bharat Digital Mission’s existing infrastructure, the tool (referred to informally as “Smart Doctor”) is now live or being activated in hospital software across the country. This article explains what it actually does, what it doesn’t, and what a practising doctor encountering it should understand.

What the System Actually Is

The AIIMS CDSS is a rule-based clinical decision support system, developed with support from the Ministry of Health and Family Welfare, focused specifically on non-communicable diseases (NCDs) — conditions like diabetes and hypertension that account for a substantial share of India’s disease burden. By analysing a patient’s medical history, the system provides evidence-based recommendations on appropriate treatment plans, including guidance on drug selection and dosage, and flags potential contraindications, all designed to cross-reference a patient’s symptoms and history against a standardised, guideline-driven database rather than the more opaque machine-learning approach some other AI diagnostic tools use.

How It’s Being Distributed: Riding on Existing Infrastructure

Rather than requiring hospitals to adopt an entirely separate application, the CDSS ships inside existing hospital software, distributed through the ABDM technology stack already covered elsewhere in this series. The NHA has directed hospital software vendors to activate the CDSS module within their existing platforms, and facilities running software that is not ABDM-approved have been directed to upgrade their systems to enable integration — with the central government offering technical support to facilitate this transition. This distribution approach is precisely why a government-built tool can reach 70,000 sites essentially at once: it plugs into infrastructure that was already being rolled out for other ABDM purposes, rather than requiring a completely separate procurement and installation process at each hospital.

The Explicit Design Principle: Assist, Never Override

This is the single most important thing for a doctor to understand about the system, and it is stated consistently and explicitly by NHA officials: the CDSS is fundamentally a decision-support mechanism designed to assist clinicians, not to override their judgement. The treating doctor keeps final authority over diagnosis and treatment decisions — the tool cross-references symptoms and protocols against a standardised database and surfaces relevant, evidence-based information, but does not make the decision itself. This design choice, as covered in the companion article on AI diagnosis liability, has direct legal significance: because the doctor remains the final decision-maker by explicit design, the doctor also remains the primary locus of professional and legal responsibility when using the tool.

Where the System Builds On Existing Experience

The AIIMS CDSS is not being deployed into a vacuum — an earlier, related version has already been used within the eSanjeevani teleconsultation platform, covered in more depth elsewhere in this series, where a prospective implementation study (conducted between 2022 and 2024, in collaboration with AIIMS Rishikesh, Wadhwani AI, and CDAC Mohali) evaluated an AI-CDSS specifically designed to assist eSanjeevani’s high-volume teleconsultation workflow. That earlier deployment addressed a genuine, documented problem: eSanjeevani’s original free-text symptom recording produced unstructured, inconsistent data that was difficult to interpret consistently across a huge volume of remote consultations — the CDSS was intended to standardise and structure that process. The current 70,000-hospital rollout represents a significant scale-up of this same underlying approach, extended from the teleconsultation context into general hospital practice.

Why Non-Communicable Diseases Specifically

The initial focus on NCDs is a deliberate choice, not an incidental limitation. Non-communicable diseases — diabetes, hypertension, and related chronic conditions — are high-volume, guideline-driven, and prone to real variability in how consistently they are managed across different doctors, facilities, and regions of India. A rule-based system referencing standardised national guidelines (ICMR and Ministry of Health and Family Welfare protocols) is particularly well suited to this kind of condition, where the goal is reducing unwarranted variation in care quality between, say, a specialist in a metro tertiary hospital and a general physician at a rural primary health centre — rather than attempting the more open-ended, harder-to-standardise task of complex differential diagnosis in acute or unusual presentations.

What This Means for a Doctor Encountering the Tool for the First Time

Understand that the CDSS output is a recommendation grounded in standardised national guidelines, not a definitive diagnosis or an order you’re obligated to follow without your own clinical assessment.
Apply the same documented clinical correlation you would with any other decision-support information — note your own reasoning, especially where you agree or disagree with what the system suggests.
Recognise this tool as a rule-based system referencing structured protocols, distinct from more opaque, pattern-learning AI diagnostic tools — its recommendations should be traceable back to the specific guideline or protocol behind them, which is worth understanding if you want to evaluate a suggestion critically.
If your hospital’s software has not yet activated the CDSS module, or is running non-ABDM-approved software, this rollout is a relevant prompt to raise with your hospital’s IT and administrative leadership, given the national directive already issued to vendors.
Treat this as part of the broader shift toward ABDM-integrated, digitally structured clinical workflows covered throughout this series — the CDSS module is one piece of a considerably larger digital health infrastructure transition already underway.

The Regulatory Backdrop This Sits Within

This deployment does not exist in a regulatory vacuum — it arrives against the backdrop of CDSCO’s new Software as a Medical Device guidance (covered in a companion article), which introduces exactly the kind of risk classification and Algorithm Change Protocol framework that a large-scale clinical AI tool like this would eventually need to operate under as India’s regulatory structure for medical software matures. It also connects to the ICMR’s 2023 Ethical Guidelines for AI in healthcare, discussed in the companion article on AI diagnosis liability, which call for exactly the kind of ongoing validation and audit that a tool operating at this scale should be subject to.

Frequently Asked Questions

Is the Clinical Decision Support System mandatory for doctors to use?

The rollout directive is aimed at activating the module within hospital software systems nationally; whether and how individual doctors are expected to engage with its recommendations in daily practice is being implemented at the facility and state level, but the tool is explicitly designed as an assistive aid, not a mandatory override of clinical judgement.

Does the CDSS replace the need for specialist consultation?

No. It is designed to support standardised, guideline-based management, particularly useful for extending consistent NCD care quality to lower-resource or rural settings — it does not replace the clinical judgement or specialist referral pathways a doctor would otherwise use.

How is this different from a general AI chatbot giving medical advice?

The CDSS is a rule-based system built on standardised national clinical guidelines and deployed specifically within a clinical workflow for use by registered doctors — it is not a consumer-facing conversational AI tool, and it is not intended for direct patient use without a treating clinician involved.

What happens if a doctor disagrees with a CDSS recommendation?

The doctor retains final clinical authority by explicit design; disagreeing with and overriding a recommendation, based on independent clinical judgement, is expected and should simply be documented with the doctor’s own reasoning, consistent with good clinical practice generally.

Is my hospital’s software required to support this tool?

The NHA has directed hospital software vendors to activate the CDSS module, and facilities on non-ABDM-approved platforms have been directed to upgrade for integration, with central technical support offered to facilitate this — hospitals should check with their software vendor on current activation status.

Researched Sources

Disclaimer

This article is for general informational and educational purposes and reflects the national CDSS rollout as understood at the time of writing, which is an ongoing and evolving deployment. It is not clinical or regulatory guidance; doctors should refer to their hospital’s specific implementation documentation and the official ABDM/AIIMS CDSS resources for operational details.

Vivek Chaudhary

Vivek Chaudhary is a Technical Content Developer specializing in healthcare, health technology, and digital healthcare business solutions. He creates research-driven, SEO-focused content for doctors, clinics, hospitals, healthcare professionals, and patients, covering topics such as healthcare technology, patient engagement, clinic management, digital communication, and online visibility.

Smart Doctor CDSS: AI Rolling Out Across 70,000 Hospitals