Clinical decision support vs. diagnostic device software: what’s the actual difference?
Clinical decision support (CDS) software gives a health care professional information to help them make a decision, while diagnostic device software is regulated by the FDA because it analyzes medical images, signals, or data to produce a diagnostic finding on its own. Picture a routine visit: your doctor pulls up your recent lab results, and a small on-screen note suggests she consider a follow-up test based on your history. Whether that note comes from a regulated “medical device” or something closer to a smart reference chart depends on a specific framework FDA uses— and this guide walks through it.
Where the clinical decision support vs. device line comes from
In 2016, Congress passed the 21st Century Supports Act, amending the Federal Food, Drug, and Cosmetic Act to exclude certain decision-support software from the legal definition of a “device.” FDA issued guidance interpreting that exclusion, with a revised final version issued in January 2026.
That guidance covers software aimed only at health care professionals (HCPs)only —doctors, nurses, pharmacists, and similar clinicians. Software that gives recommendations directly to a patient or caregiver, rather than to a clinician, falls under a different set of FDA digital health policies, not this framework.
What kind of information is the software working with?
FDA first looks at the software’s input. Software intended to acquire, process, or analyze a medical image, a signal from an in-vitro diagnostic device, or a signal from something like an ECG or continuous glucose monitor falls outside the non-device category automatically, regardless of anything else the software does.
A tool that simply displays or organizes information a clinician already has — prior lab values, medication lists, published clinical guidelines — is working with a different kind of input than a tool that reads and interprets a raw image or signal itself.
What kind of output does the software produce?
Next, FDA looks at whether the software is limited to displaying, analyzing, or printing medical information about the patient or other established medical information — patient-specific data already in the record, or peer-reviewed clinical literature and guidelines. The software is expected to work with information a clinician could otherwise look up, not generate a new diagnostic signal on its own.
Does the software support the clinician, or replace their judgment?
This is where FDA’s most detailed line-drawing happens. Software is more likely to be regulated as a device when its automation is intended to replace or direct a clinician’s decision-making, or when it produces a time-critical output — like an alarm meant to trigger immediate action — where there isn’t real time for the clinician to independently evaluate it. Software intended to support, not direct, a clinician’s judgment, leaving room to review the basis for a suggestion before acting, sits closer to the non-device side.
Can the clinician see why the software reached its conclusion?
Finally, FDA looks at transparency. To fall outside device regulation, software generally needs to let the intended clinician user understand the basis for its recommendations — not function as an unexplained “black box” the clinician is expected to accept without being able to independently review the underlying information.
What changed in FDA’s January 2026 update?
The revised guidance added a narrower carve-out for software that recommends a single, clinically appropriate option rather than a range of choices. FDA said it intends to exercise enforcement discretion — meaning it does not intend to enforce device requirements — for tools that meet the other three criteria above and produce one recommendation because only one option is genuinely appropriate, such as software suggesting a specific FDA-approved drug for a clinician to consider based on a patient’s symptoms and history. The guidance does not extend this same treatment to more novel generative-AI tools, such as AI-enabled clinical chatbots, which it does not squarely address.
Why one factor can decide the classification by itself
It’s easy to miss this, so it’s worth repeating: the criteria above generally must be met together for software to fall outside device regulation. If a software function is intended to analyze a medical image or a diagnostic-device signal, that alone keeps it inside FDA’s device oversight, no matter how the rest of the product works. A partial product description isn’t enough to know how something would actually be categorized.
Does non-device status mean the software is safe or accurate?
No. Whether software counts as a device under this framework is a legal and regulatory classification question, not a stand-in for asking whether a specific tool is accurate, current, or appropriate for a given decision. Software on either side of this line can vary in quality. Non-device status also doesn’t tell you whether a specific device-classified product has completed FDA clearance or approval — that depends on its own classification and history. For a broader look at how we vet sources like FDA guidance before citing them, see our How We Research page.
Non-device CDS vs. device software: what tends to separate them
These are general tendencies from FDA’s framework above, not a checklist for classifying a real product yourself:
- Input: non-device CDS works with existing records, guidelines, or patient history; device software analyzes a raw medical image or diagnostic signal.
- Output: non-device CDS displays or analyzes existing medical information; device software often generates a new diagnostic finding.
- Automation level: non-device CDS supports a clinician’s judgment with time to review it; device software may replace judgment or issue time-critical alerts.
- Transparency: non-device CDS lets the clinician see the basis for a suggestion; opaque “black box” outputs point toward device status.
- Audience: this whole framework applies to HCP-facing tools; patient- and caregiver-facing tools are evaluated under separate FDA digital health policies.
A decision path for thinking through what you’re looking at
If you’re trying to understand what kind of software a description refers to, these are the general questions FDA’s framework raises:
- Who is the software talking to — a health care professional, or a patient/caregiver directly?
- Is it reading a raw medical image or diagnostic-device signal, or working with information already in the record?
- Is it limited to organizing or analyzing existing medical information, rather than generating a new diagnostic finding?
- Is it meant to support a clinician’s judgment with time to review it, or does it produce a time-critical alert meant to trigger immediate action?
- Can the clinician see the basis for the software’s suggestion, or is the reasoning hidden?
Answers pointing toward supporting a clinician’s independent review of existing information suggest a software function closer to non-device CDS. Any answer pointing the other way suggests device oversight. Only FDA, or a qualified regulatory professional evaluating a specific product’s intended use, can make an actual determination.
Common questions about clinical decision support and device software
Is an AI symptom checker for patients regulated the same way as clinician-facing CDS?
No. FDA’s non-device CDS framework only applies to software directed at health care professionals. FDA evaluates patient- and caregiver-facing tools, including AI symptom checkers, under separate digital health policies, not the criteria described in this guide.
Does “FDA cleared” mean a piece of decision-support software is accurate?
Clearance or approval reflects FDA’s review of a device’s safety and effectiveness data for its stated use, not a general accuracy commitment for every situation. FDA doesn’t review non-device CDS software under this framework because it falls outside the device definition.
Can software that analyzes a medical image ever qualify as non-device CDS?
Generally no. FDA’s guidance states that software intended to analyze a medical image or a signal from a diagnostic device or signal-acquisition system falls outside the non-device CDS category automatically, regardless of how the rest of the software functions.
Who decides whether a specific product is a device or non-device CDS?
FDA makes that determination, informed by the manufacturer’s own intended-use claims and a review process. Reading a product description isn’t enough to classify software yourself; a qualified regulatory professional would need to review the specific intended use for a real determination.
What this guide is not
This article explains FDA’s general framework in plain language. It is not a regulatory determination for any specific product, and it does not diagnose, evaluate, or recommend any device, app, or clinical tool. For our full editorial standards on sourcing and review, see Editorial Policy. If you have questions about specific software your care team uses, ask your health care provider or the software’s manufacturer directly. For a broader introduction to how we approach this topic, see our Start Here page.
Sources
- U.S. Food and Drug Administration, Clinical Decision Support Software guidance for industry and FDA staff, final guidance, January 2026.
- U.S. Food and Drug Administration, How to Determine if Your Product is a Medical Device, Medical Device Classification Guidance.
This article is for general education only and is not medical, legal, or regulatory advice. It does not evaluate or endorse any specific product. See our Medical Information Disclaimer for more. Last updated September 9, 2026.
By Connected Diagnostics Evidence Editorial Team
This content is for informational and educational purposes only and does not constitute medical advice. Always consult a qualified healthcare provider before making changes to your health routine, medications, or supplements.
Leave a Reply