Hospital AI Lab Practical information on medical AI for Indian hospitals

Updated:

Is your hospital AI a medical device in India? CDSCO says

CDSCO’s 2026 guidance on medical device software says what counts. A plain reading for administrators and nurse leaders before they sign a proposal.

When a vendor says its AI tool is “CDSCO approved”, or says it needs no approval at all, an administrator has no easy way to judge. CDSCO’s Guidance Document on Medical Device Software (Doc No. CDSCO/MD/GD/MDSW/01/2026) is the first place to look. CDSCO’s public-notices page lists it with a release date of 21 July 2026.

What the guidance covers

The document applies to software that meets the definition of a “medical device” in the Medical Devices Rules, 2017. It calls this Medical Device Software, and says mobile apps, AI/ML-based software and cloud or network-based software that meet the definition are considered Medical Device Software. Software that controls or adjusts a medical device is considered a medical device, and standalone software intended for a medical purpose is covered too. Software merely intended for general wellness or fitness tracking, with no medical purpose, is outside the document.

The guidance says it reflects current practice under the Rules and should not be misconstrued as a new regulatory control.

The guidance also carries a notice. It is meant to create public awareness and is not meant for legal or professional use. The legal texts are the Drugs and Cosmetics Act, 1940 and the Medical Devices Rules, 2017.

What decides the class

Rule 4 of the Rules sets four risk classes: A (low risk), B (low moderate), C (moderate high) and D (high risk). The guidance says the class of this kind of software is fundamentally based on its intended use and the applicable rules in the First Schedule. In the Schedule, software that drives a device or influences its use automatically falls in the same class as that device.

For a hospital, the practical point is the intended use statement. The same model can sit in different places depending on whether the manufacturer says it informs, supports or makes a clinical decision.

Three questions to put to the vendor

  1. Does the vendor treat this product as a medical device under the Rules, and in which class?
  2. What licence or permission does it hold for this software and this intended use?
  3. Does your planned use match the intended use on that permission?

Ask for the answers in writing and attach them to the contract.

Key takeaways

  • Software that meets the medical device definition is regulated under the Medical Devices Rules, 2017, and CDSCO’s guidance considers AI/ML-based software that meets it to be Medical Device Software.
  • Class A to D depends on the intended use the manufacturer states.
  • The guidance is for public awareness, not legal use; read the Rules for a real decision.
  • Ask the vendor for class, licence and intended use in writing.

Sources: CDSCO, Guidance Document on Medical Device Software, Medical Devices Rules, 2017, CDSCO public notices

This article is part of a series

All series

← Blog