(Don’t Fear) the Regulator: FDA Clarifies When Software and Wellness Products Are Not Medical Devices

Key Takeaways

• FDA’s 2026 guidance clarifies when digital health tools fall outside device regulation. Updated policies for clinical decision support (“CDS”) software and general wellness products provide clearer, more predictable boundaries for when software and wearables are not regulated as medical devices.
• Transparency, validation, and human oversight are decisive. Non-device CDS must allow clinicians to independently review and understand recommendations, while wellness products must validate outputs and avoid claims that imply diagnosis, treatment, or substitution for FDA-authorized devices.
• Intended use and marketing still drive risk. Even lower-risk software or wearables can trigger FDA oversight if labeling, claims, or functionality suggest clinical decision-making, time-critical use, disease references, or reliance on unvalidated data.

“Don’t fear the Reaper,” the rock band Blue Öyster Cult famously advised. For many digital health companies, the Food and Drug Administration’s recently revised guidances affecting the classification of clinical decision support (“CDS”) software¹ and low-risk general wellness products² offer a similar message to not fear the regulator.

Through expanded discussion and examples, the January 2026 updates provide greater transparency into how FDA assesses the regulatory status of certain digital health technologies under existing law and policy. This Bulletin reviews the two guidance documents and considers takeaways for industry based on the agency’s revisions to the policies.

Guidance on CDS Software

FDA revised its CDS guidance (first issued in 2022)[3] to address the growing use of software in healthcare decision-making, particularly tools that provide recommendations or analyses based on clinical data. The update clarifies which types of software functions may qualify as non-device CDS. The guidance draws clearer boundaries around algorithm transparency, intended use, and the level of clinician oversight required.

• The core policy remains intact: to qualify as a non-device CDS and avoid medical device regulation, software must meet four criteria.
  1. The software cannot analyze medical images, diagnostic signals, or patterns from signal acquisition systems.
  2. It must display, analyze, or print patient data or general medical information, like clinical studies or guidelines.
  3. Recommendations can support healthcare professionals on prevention, diagnosis, or treatment, but cannot replace their judgment.
  4. The clinician must be able to review the reasoning behind any recommendation and not rely solely on the software for decisions.
• FDA clarifies that only repeated or sequential signal measurements constitute a “pattern,” while single, point-in-time measurements do not. Software relying solely on discrete measurements generally qualify as non-device CDS.
• The agency defines “medical information about a patient” as information used in clinical care or decision-making and supported by accepted sources, even if it is not routinely discussed in clinical conversations. This allows software developers to use a broader set of patient-specific data in tools while staying within the non-device CDS framework, so long as the information is clinically relevant and properly sourced.
• The guidance updates FDA’s approach for single clinically appropriate recommendation CDS tools. When only one option is clinically appropriate and the software meets all other criteria, FDA now intends to exercise enforcement discretion, signaling a more permissive stance than before.
• Several examples are provided to illustrate when single‑recommendation CDS software is subject to regulation. The following briefly summarizes only some of these illustrations.
  • Software predicting future cardiovascular risk based on established clinical data (e.g., weight, smoking history, blood pressure, and lab results) can qualify as non-device CDS, but adding unvalidated inputs or predicting imminent events would make it a regulated device.
  • Software that provides a differential diagnosis or a single-recommendation to support healthcare professionals decision-making qualifies as non-device CDS, but software that establishes a definitive diagnosis from images, waveforms, or other signal-level data is regulated as a medical device.
  • Software that summarizes a radiologist’s findings for HCP review qualifies as non-device CDS, but analyzing images directly or using unverified data makes it a regulated medical device.
  • Software that helps doctors make decisions by providing recommended orders, matching patient information to guidelines, sending safety alerts, offering preventive care reminders, or summarizing patient data may be considered non-device CDS.
  • A software function intended to support time-critical decision-making remains subject to regulation as a medical device.
• FDA highlights that automation bias can lead clinicians to over-rely on software recommendations, especially in time-sensitive situations, reinforcing the need for independent review of CDS outputs.

Guidance on Low-Risk General Wellness Products

FDA updated its guidance on general wellness products (first issued in 2016, previously revised in 2019)⁴ to address the rapid growth of wearables and digital health technologies. The previous frameworks focused primarily on activity and lifestyle metrics, while the 2026 revisions clarify when non-invasive physiologic sensing products, including wearables, may remain within the low-risk general wellness framework. The following summarizes the key updates most relevant to industry.

• The core policy remains intact: when a product is intended only for general wellness use and presents low risk to users, FDA will generally exercise enforcement discretion to not classify the product as a regulated medical device.
• The revised guidance expands the scope of low-risk wellness devices. FDA now expressly recognizes certain non-invasive physiologic sensing products (e.g., optical sensing) as eligible for the general wellness policy when outputs are limited to wellness uses, including measures such as blood pressure, oxygen saturation, blood glucose, and heart rate variability.
• Qualifying products:
  • must be non-invasive and not implanted;
  • should not diagnose, treat, cure, mitigate, or prevent disease;
  • cannot substitute for FDA-authorized devices;
  • do not prompt clinical action or provide treatment guidance; and
  • may not present clinical values unless appropriately validated.
• FDA confirms that eligible products may display values, ranges, trends, baselines, and longitudinal summaries, and may contextualize data with sleep, activity, stress, recovery, or similar wellness domains.
• The guidance adds new exclusions for products with clinical features, specifying that products may fall outside the general wellness policy if products reference diseases or diagnostic thresholds, prompt clinical action or treatment, claim equivalence or substitution for FDA devices, or are intended for disease-related use.
• FDA now allows products to include notifications suggesting healthcare evaluation when outputs fall outside appropriate ranges. The notification should not name a disease, characterize outputs as diagnostic, include clinical thresholds, or provide ongoing disease monitoring alerts.
• The guidance now includes additional examples of how FDA would classify products under its policy. The three new examples focus on wearables: a wrist-worn wearable outputting blood pressure, a microneedle-based glucose, and a non-invasive athletic wearable.
  • Wrist-worn wearable: The product measures sleep, pulse, and blood pressure using an accelerometer and photoplethysmography monitor. FDA considers it a low-risk general wellness product if blood pressure values are validated. However, any claims implying medical or clinical use would place it outside the policy.
  • Microneedle glucose monitor: The product estimates blood glucose using minimally invasive microneedles and is marketed for tracking nutritional impacts. Although the claims are wellness-focused, the device does not meet the low-risk requirement because it penetrates the skin.
  • Non-invasive athletic wearable: The product monitors electrolytes, lactate, and hemoglobin for fitness use, using validated values and including disclaimers that the device is not for diagnosing any condition. FDA considers it a low-risk general wellness product.

AGG Observations

• Both guidances highlight FDA’s intent to provide digital health companies with clearer boundaries on when products fall outside the scope of FDA’s oversight. Industry can use the guidances and illustrative examples to evaluate whether their clinical decision software or wellness products are subject to medical device regulation.
• FDA emphasizes validation and transparency. CDS software must let clinicians independently review recommendations, while wellness products must validate outputs and avoid intended use claims that imply medical or clinical use to stay within the low-risk framework.
• Firms should carefully review labeling, promotional materials, and user-facing messaging because intended use and marketing claims can affect FDA’s evaluation of the product’s regulatory classification. For wellness products, references to diseases or clinical equivalence can trigger increased oversight. For CDS, unvalidated inputs, imminent-event predictions, or definitive diagnoses from images or signals may make software a regulated device.
• Developers should be mindful of AI and other automated CDS tools, designing them to reduce automation bias and ensure clinicians remain in control of any recommendations. FDA expects these tools to support, not replace, independent clinical judgment.
• In summary, the revised guidance documents reflect the agency’s efforts to modernize and right-size its regulatory approach for digital health. The policies provide for more predictable pathways for lower-risk products while maintaining oversight for higher-risk products. Companies that design products with clear regulatory positioning can expedite time-to-market, reduce enforcement risks, or both.

[1] The guidance, “Clinical Decision Support Software,” issued on January 6, 2026, can be accessed here.

[2] The guidance, “General Wellness: Policy for Low Risk Devices,” issued on January 6, 2026, can be accessed here.

[3] AGG’s Bulletin on FDA’s 2019 guidance on clinical decision software can be accessed here.

[4] AGG’s Bulletin on FDA’s 2016 guidance can be accessed here. Our review of the 2015 draft guidance can also be accessed here.