FDA Issues Key Updates to CDS Software Guidance

By Emily Statham and Derek Flint

The U.S. Food and Drug Administration (FDA or Agency) recently issued a revised final guidance document addressing Clinical Decision Support (CDS) Software (the 2026 CDS Guidance). The 2026 CDS Guidance, which supersedes an earlier version of the guidance from 2022, clarifies regulatory boundaries, introduces targeted areas of enforcement discretion, and signals the Agency’s continued evolution in regulating (and de‑regulating) tools that incorporate machine learning and advanced analytics.

Statutory Background

CDS software refers to software that is intended to provide decision support for the diagnosis, treatment, prevention, cure, or mitigation of diseases or other conditions. It includes a variety of functions, such as “computerized alerts and reminders for providers and patients; clinical guidelines, condition-specific order sets, focused patient data reports and summaries, documentation templates, diagnostic support, and contextually relevant reference information.” Under Section 520(o)(1)(E) of the Federal Food, Drug, and Cosmetic Act (FDCA), as amended by the 21st Century Cures Act, CDS software is excluded from the statutory definition of a “device” if it satisfies the following four criteria (the Non-Device CDS Criteria):

1. It is not intended to acquire or analyze medical images, IVD signals, or physiological patterns;

2. It is intended to display, analyze, or print medical information about a patient or other medical information;

3. It is intended to support or provide recommendations to healthcare professionals (HCPs) about prevention, diagnosis, or treatment of a disease or condition; and

4. It is intended to enable HCPs to independently review the basis of those recommendations.

In 2022, FDA issued final guidance describing its interpretation of the Non-Device CDS Criteria (the 2022 CDS Guidance). The 2022 CDS Guidance drew substantial criticism from industry stakeholders and notable commentators who argued that FDA’s interpretation was far narrower than Congress intended and risked inhibiting CDS software innovation.

It is against this backdrop of criticism that FDA issued the 2026 CDS Guidance. While not a wholesale reversal of the 2022 CDS Guidance, the 2026 CDS Guidance materially clarifies several areas that have long generated ambiguity. It also signals FDA’s willingness to deploy enforcement discretion where the risk to patient safety is low, and a clinician remains firmly in control of the decision-making process.

Key Changes in the 2026 CDS Guidance

The 2026 CDS Guidance clarifies and expands upon the 2022 CDS Guidance in three primary ways.

1. More flexible approach to “singular recommendations”

The 2026 CDS Guidance evinces FDA’s willingness to permit single, clinically appropriate recommendations without automatically categorizing the CDS software as a regulated device. Historically, FDA took the position that CDS software must provide multiple treatment options or recommendations, rather than a single output, to satisfy Criterion 3. That approach forced many CDS software developers to deliberately design software to show multiple options, even in clinical scenarios where only one option made sense. The policy was widely seen as out of step with real‑world medicine, where treatment guidelines, best practices, and patient‑specific factors often converge on a single logical recommendation.

The 2026 CDS Guidance reverses this presumption, recognizing that some clinical scenarios produce only one medically reasonable option. FDA now acknowledges that the mere presence of a single recommendation does not, by itself, convert CDS software into a regulated device. Instead, the Agency recognizes that singular recommendations can still fall outside FDA’s device oversight if the software otherwise satisfies the Non-Device CDS criteria.

Importantly, FDA explicitly states it will exercise enforcement discretion for CDS software functions that produce a singular output when “only one clinically appropriate treatment option or recommendation” exists based on the patient‑specific information and clinical guidelines supplied. This reflects a practical understanding that software should not be artificially constrained to generate multiple options simply to retain exemption status.

2. Clearer boundaries around permissible “medical information” and regulated “signal” interpretation

FDA’s 2026 CDS Guidance offers long‑awaited clarity on what types of data CDS software developers may use without triggering device regulation. FDA reaffirms that CDS software may use structured clinical information already familiar to clinicians — including patient demographics, symptoms, discrete lab results, medication and problem lists, discharge summaries, and trusted external sources such as clinical guidelines, peer‑reviewed literature, and FDA‑required labeling. Software that displays, organizes, ranks, or cross‑references these inputs, while maintaining transparent logic, fits comfortably within the Non‑Device CDS framework.

At the same time, the Agency draws a much sharper line around data that requires interpretation to become clinically meaningful. The 2026 CDS Guidance explains that CDS software crosses into device territory when it processes raw, uninterpreted sensor or imaging data such as waveforms, pixel‑level images, spectroscopy output, PPG signals, or other physiologic measurements to generate a new metric or “signal.” Even a single numerical value may constitute a regulated signal if the software had to analyze raw inputs to derive it. By contrast, using a discrete value already interpreted and recorded in the patient’s chart (e.g., a lab result provided by a laboratory system) does not constitute signal processing.

This clarified boundary resolves longstanding ambiguity: analytics built on already‑interpreted clinical information will generally remain within the Non‑Device CDS exemption, while analytics that derive new parameters from raw data streams will likely fall within FDA’s device oversight. For developers, this distinction offers a practical architectural guidepost when evaluating data pipelines, algorithms, and intended use.

3. Strengthened expectations for transparency and clinician review

In its 2022 CDS Guidance, FDA emphasized that non‑device CDS software must let clinicians independently review the basis of any recommendation. But the Agency offered comparatively high‑level descriptions of what that meant in practice. The 2026 CDS Guidance operationalizes this requirement: developers are now expected to provide plain‑language explanations of the software’s logic, identify data sources and inputs, describe methods/assumptions, and disclose validation work and known limitations so that an HCP can verify the recommendation without relying primarily on the tool. In practice, this clarification makes it more difficult for complex or “black box” machine learning models to qualify for the Non-Device CDS exemption unless they incorporate explainability features that enable genuine independent clinical assessment.

Practical Considerations for CDS Software Developers

The 2026 CDS guidance provides important relief for developers who struggled to fit innovative software into the rigid interpretations presented in 2022. By softening its stance on single recommendations, clarifying definitions of medical information and signal analysis, and expanding transparency expectations, FDA has moved toward a more predictable, risk‑based, innovation‑supportive framework.

Organizations developing CDS tools, whether standalone software or features integrated into broader health IT platforms, should review the 2026 CDS Guidance closely and re‑evaluate product claims, labeling, logic transparency, and data‑handling practices to ensure alignment with the updated regulatory boundary lines.

*** Opinions expressed are those of the authors and not necessarily the firm’s or their colleagues’.