FDA Final Guidance: Clinical Decision Support Software

The U.S. Food and Drug Administration (FDA) released an updated final guidance document titled “Clinical Decision Support Software | Guidance for Industry and Food and Drug Administration Staff” on January 29, 2026, superseding the 2022 version.

The latest guidance refines FDA’s interpretation of Section 520(o)(1)(E) of the 21st Century Cures Act, which defines the criteria under which certain software functions, including Clinical Decision Support (CDS) tools, are excluded from the statutory definition of a “medical device.” This interpretation affects whether such software must comply with medical device regulatory requirements.

Key updates include clearer boundaries between Non-Device CDS and Device CDS, along with expanded examples intended to resolve ambiguities that persisted after the 2022 guidance. The FDA also emphasizes transparency in data sources, exercises enforcement discretion for software that offers a single recommendation when only one option is clinically appropriate and relocates the discussion of time-critical functions to improve clarity and usability.

The guidance further clarifies that certain multiple-function device products may incorporate both Non-Device CDS software functions and device software functions. In such cases, the FDA evaluates the safety and effectiveness of the device software function using the framework set forth in its guidance Multiple Function Device Products: Policy and Considerations | FDA, in alignment with Section 520(o)(2) of the FD&C Act.

For more information about MedEnvoy’s US Agent service, click here.

What is a CDS Software? 

Clinical Decision Support (CDS) software is a type of health technology designed to help clinicians make more informed, evidence-based decisions about patient care. It does not replace clinical judgment; rather, it provides information, insights, or recommendations that support it.

CDS software is primarily used by physicians, nurses, pharmacists, and other licensed healthcare professionals. CDS software is intended to:

• Organize or display medical information, such as lab results, medication lists, or clinical guidelines.
• Analyze patient data to highlight patterns or risks.
• Offer recommendations related to diagnosis, treatment, or disease management.
• Alert clinicians to potential issues, such as drug–drug interactions or abnormal results.

CDS software can improve safety, reduce errors, and support consistent, evidence-based care. However, depending on how it functions, some CDS tools are regulated by the FDA as medical devices (Device CDS), while others are exempt (Non-Device CDS). This distinction hinges on factors such as transparency, risk, and whether the clinician can independently review the basis for the recommendation.

What Qualifies as Non-Device CDS? 

FDA reiterates that CDS is not a medical device when all four statutory criteria for exclusion from Device Definition are met: 

1. The software does not acquire, process, or analyze medical images or signals. 

2. It is intended to display, analyze, or print medical information. 

3. It supports or provides recommendations* to Health Care Professionals (HCPs) about prevention, diagnosis, or treatment. 

4. It allows the HCP to independently review the basis for the recommendation. 

*FDA clarified that software offering a single, clinically appropriate recommendation can still fall under enforcement discretion, as long as it satisfies the remaining exclusion criteria. This reflects an expansion of the agency’s prior policy position. 

Key Updates of the 2026 Version 

1. Clearer Definition of “Independent Review” 

FDA strengthened its explanation of what it means for a healthcare professional to “independently review the basis” of a recommendation and not rely primarily on the software. 

The update emphasizes transparency of logic (e.g., rule-based logic, visibility of inputs, traceability of rules or calculations) and ability for the clinician to validate or challenge the recommendation. This is especially important for AI/ML systems, which often obscure their internal reasoning. 

2. Expanded Examples Sections of Non-Device CDS and Device CDS 

FDA added or revised examples to reduce ambiguity:  

• Non-Device examples: CDS functions that meet all four Cures Act criteria
  • Drug-drug interaction checkers with visible rule sets. 
  • Risk calculators with transparent formulas. 
  • Clinical guidelines embedded in EHRs with citations. 

• Device examples: CDS functions that do not meet the Cures Act exemption 
  • AI/ML models that do not provide explainability. 
  • Predictive analytics (e.g., sepsis prediction) where the basis cannot be independently reviewed. 
  • Imaging based or signal processing based CDS. 
  • Software that recommends a specific treatment plan. 
  • Time critical triage or risk stratification tools.

3. Alignment With Broader Digital Health Policy 

This FDA support software guidance brings CDS terminology into line with the rest of FDA’s digital health ecosystem by using the same language found in the Digital Health Policy documents such as Policy for Device Software Functions and Mobile Medical Applications | FDA, Medical Device Data Systems, Medical Image Storage Devices, and Medical Image Communications Devices | FDA, General Wellness: Policy for Low Risk Devices | FDA. By aligning these terminologies, FDA makes it easier for developers to see how CDS functions fit within the larger regulatory landscape and to apply consistent expectations across different types of digital health tools 

What This Means for Developers 

• More predictable boundaries between device and non-device CDS. 

• Greater emphasis on transparency — for example, if clinicians cannot see how the software reached its conclusion, FDA will likely treat it as a device. 

• AI/ML CDS is still regulated unless it is fully explainable and meets all four Cures Act criteria. 

How MedEnvoy Can Support  

MedEnvoy can help manufacturers interpret and operationalize the updated FDA Clinical Decision Support (CDS) guidance by translating the agency’s clarified boundaries, particularly around AI/ML based tools, transparency expectations, and enforcement discretion, into concrete regulatory pathways. Our team supports developers in assessing whether their CDS functions meet the Cures Act exclusion criteria, determining when device requirements apply, and preparing the appropriate regulatory strategy. We provide end-to-end assistance across regulatory planning, risk classification, evidence expectations, and FDA engagement to reduce uncertainty and prevent downstream delays. If you would like to discuss how this guidance applies to your specific software or submission pathway, you can connect with our regulatory experts here. For more information about MedEnvoy’s US Agent service, click here.