Glossary
Clinical decision support meaning: software that advises The difference between recording a decision and shaping one, how CDSCO treats software that does the second, and what ICMR says about who is accountable.
Clinic software either writes down what the doctor decided or tries to change it. The second kind is clinical decision support, and in India it sits on the edge of medical device regulation. Here is the line, drawn from CDSCO's own guidance.
Published · Last reviewed · 5 min read
- CDSCO's 2026 guidance on medical device software says hospital and clinic information systems used for admission, scheduling, billing and storing patient information are not covered by the Medical Devices Rules, 2017.
- The same guidance adds that if such a system has functions such as "quantification of physiological parameters for clinical decision-making", it "may be considered as a medical device".
- ICMR's 2023 AI guidelines put responsibility for using an AI-based tool on the health professional and ask for human oversight and manual override.
Clinical decision support (CDS) is software that offers the doctor a suggestion, alert or score at the point of care, such as a drug interaction warning, a dose check or a likely diagnosis, so that the software's output influences the clinical decision rather than merely recording it. The common forms are interaction checkers, allergy and dose alerts, diagnostic suggestions and risk scores. What they share is direction: information flows from the software to the doctor before the decision, not from the doctor to the record after it.
What is clinical decision support?
A rule engine or a model, attached to the record. The simplest CDS is a table of interacting drug pairs: add the second drug and a box appears. More complex CDS reads the whole chart and proposes. HealthPlix's H.A.L.O page, read on 4 Nov 2026, says the product "comes equipped with built-in Drug-to-Drug Interaction (DDI) and Clinical Decision Support (CDS) systems, providing you with enhanced suggestions and valuable insights for improved decision-making". That is a usable definition of the category: suggestions offered to the doctor for the decision.
Where does the definition come from?
From the regulator that decides whether software is a medical device. CDSCO's Guidance Document on Medical Device Software, document CDSCO/MD/GD/MDSW/01/2026, released on 21 Jul 2026 according to the CDSCO medical devices page, restates the definition of "medical device" from the Medical Devices Rules, 2017: "All devices including an instrument, apparatus, appliance, implant, material or other article, whether used alone or in combination, including a software or an accessory, intended by its manufacturer to be used specially for human beings or animals ... for one or more of the specific purposes of (i) diagnosis, prevention, monitoring, treatment or alleviation of any disease or disorder" and five further purposes. CDSCO's page also lists the gazette notification that specified this definition, S.O. 648(E) dated 11 Feb 2020. Software meeting that definition "is regulated as 'medical device'".
Section 5.2 of the guidance then lists software that is not covered. Among the examples: "Hospital/Clinical Information systems (HIS/CIS) that support the process of patient data management (intended only for patient admission, for scheduling patient appointments/visits, for insurance and billing/invoicing purposes, enabling clinical communication such as voice calling, video calling, to store and transfer patient information ... generated in association with the patient's treatment)", and software that performs "transfer, storage, archive data, convert, format, communication, simple search, compression". It follows with a note: "If any HIS/CIS or LIS/IMS has any additional functions that allow its use for any medical purposes (e.g., image analysis/modification as an aid in diagnosis, quantification of physiological parameters for clinical decision-making, real-time patient monitoring, etc.), then it may be considered as a medical device." That note is the line between a record and CDS.
The guidance also asks applicants to state the "degree of autonomy of software", from "autonomous" through "supervised autonomy" to "non-autonomous", the last meaning "only intended as an aid for the user in clinical decision making", and to state "the extent to which clinical decision-making responsibility remains with healthcare professionals". On that last point the ICMR's Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare (2023) are direct. Section 2.3 says health professionals using AI-based solutions "may be responsible and accountable for the consequences arising out of their use", and that "human oversight and manual override mechanisms must be made compulsory for all AI technologies that have the capability to cause severe or bodily injuries". Section 1.5 says "AI for health should always be appropriately supervised".
What it is not
Record-keeping. An EMR that stores the diagnosis the doctor typed, prints the prescription and files the bill is doing what section 5.2 describes. A medicine list with each drug's composition, type and company is reference data, not advice; it does not fire when two drugs are combined. A template built from the doctor's own past prescriptions is the doctor's decision, saved.
| Function | Who decides | CDSCO guidance view |
|---|---|---|
| Storing the diagnosis and prescription the doctor entered | Doctor | Section 5.2: HIS/CIS for patient data management, not covered |
| Printing a prescription on the letterhead | Doctor | Section 5.2: format, transfer, not covered |
| Drug interaction alert when a second drug is added | Software proposes, doctor disposes | Section 5.2 note: additional medical-purpose function, may be a medical device |
| Suggested diagnosis from symptoms | Software proposes, doctor disposes | Section 5.1 example 7 pattern: information that may suggest or exclude conditions |
Does it apply to a small clinic in India?
The burden of licensing a medical device falls on the manufacturer, not the clinic. The clinic's exposure is different: if the doctor follows a wrong alert, or ignores a right one, the record shows what the software said, and ICMR puts responsibility for the tool's use on the professional. A CDS feature is something the doctor must understand, not something the vendor switches on. Before accepting one, ask whether the vendor holds a CDSCO licence or classification for it, which interaction database it uses and when it was last updated, and whether alerts can be switched off per doctor.
Where Vitels stands
Vitels has no clinical decision support, no drug interaction or dose checks and no diagnosis suggestions, by design and permanently. The software records the doctor's decision and never influences it. Its AI fills medicine details for review when a clinic builds its list and answers how-to questions about the app; neither touches a patient's chart. See security and Vitels AI.
Questions
Is a drug interaction checker clinical decision support?
Yes. It is the most common form: the software warns before the doctor finalises the prescription, so its output shapes the decision.
Is clinic software a medical device in India?
Not when it only manages patient data. CDSCO's 2026 guidance lists hospital and clinic information systems for admission, scheduling, billing and storing patient information as not covered by the Medical Devices Rules, 2017, but says added functions for clinical decision-making may make it a medical device.
Who is responsible if the doctor follows a wrong CDS alert?
ICMR's 2023 guidelines say the health professional using an AI-based solution may be responsible and accountable for the consequences of its use, with shared responsibility where the defect is in the tool itself. Confirm with your council or lawyer for your case.
Sources
- Central Drugs Standard Control Organization, Guidance Document on Medical Device Software, Doc No. CDSCO/MD/GD/MDSW/01/2026, sections 1.0, 2.0, 4.9, 5.1, 5.2 and 12.4 (h) and (i) (accessed 4 Nov 2026)
- Central Drugs Standard Control Organization, Medical Device and Diagnostics page (release date of the guidance, 2026-Jul-21; listing of S.O. 648(E) dated 11.02.2020) (accessed 4 Nov 2026)
- Indian Council of Medical Research, Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare, 2023, sections 1.5 and 2.3 (accessed 4 Nov 2026)
- HealthPlix Technologies, H.A.L.O product page (vendor page, read 4 Nov 2026) (accessed 4 Nov 2026)
This is operational guidance for clinics, not legal or tax advice. Confirm with your state authority, council or CA.
Drafted with AI assistance; every fact checked against the sources listed; reviewed by the Vitels team.