FDA General Wellness Guidance 2026: How to Prototype an AI Health Wearable Without Becoming a Medical Device

On 6 January 2026 FDA reissued its General Wellness: Policy for Low Risk Devices guidance (the FDA general wellness guidance) and, for the first time, spelled out how wearables that use non-invasive sensing (such as optical PPG sensors and accelerometers) to estimate blood pressure, oxygen saturation, heart rate variability or similar values can stay low risk general wellness products instead of regulated medical devices.

The product can display validated values, ranges, trends and baselines tied to sleep, activity, stress or recovery. It cannot name diseases, show diagnostic thresholds, push clinical actions, claim medical grade accuracy, or penetrate the skin.

For founders building an AI health wearable, the practical answer is: decide the intended use and the exact claims before the first prototype, design the app screens, alerts and firmware to match those claims, and plan a real validation study for any number that looks like a clinical measurement.

This guide is for founders, product managers and engineers building fitness trackers, smart rings, sleep and recovery bands, sports wearables and other consumer health gadgets that use sensors and AI models to estimate what is happening in the body.

It is about the line between a wellness product and a medical device, and how that line shapes the prototype.

If your product is clearly a medical device, our guides on FDA QMSR 2026 and FDA medical device cybersecurity cover what to build in.

If you are deciding where the AI model runs, see edge AI hardware prototyping in 2026.

The 6 January 2026 guidance says FDA may treat products that use non-invasive sensing to estimate, infer or output physiologic parameters (FDA lists blood pressure, oxygen saturation, blood glucose and heart rate variability as examples) as general wellness products when the outputs are solely for wellness and the product:

Such products may display values, ranges, trends, baselines or longitudinal summaries in the context of sleep, activity, stress, recovery or similar wellness domains. King & Spalding called this new section the most significant update in the guidance and noted FDA published it without a draft or comment period (9 January 2026).

A wellness product may tell a user that evaluation by a healthcare professional may be helpful when outputs fall outside ranges appropriate for wellness use.

Per the guidance, the notification should not name a disease, should not call the output abnormal or diagnostic, should not include clinical thresholds or treatment advice, and should not become ongoing monitoring of a condition.

In its town hall on the update, FDA gave an example of acceptable wording: "Your resting heart rate was high over the past few days. You should check with your doctor if this is an unexpected result." FDA also said such notifications should not use colors or sounds that characterize the output as abnormal.

FDA's examples for blood pressure (Example 7) and for an athlete wearable that reports electrolyte balance, lactate and hemoglobin (Example 9) both depend on the product having validated values.

In the town hall, FDA said validation can come from manufacturer testing or peer-reviewed publication, that FDA encourages making results public and may request a summary, and that there are no specific validation requirements under the policy, while pointing manufacturers to relevant FDA guidance, recognized consensus standards and professional society criteria.

FDA also revised its Clinical Decision Support Software guidance. The current version was issued on 29 January 2026 and supersedes the 6 January 2026 version (FDA). That guidance covers software that gives recommendations to healthcare professionals, so it matters if your wearable data will later feed a clinician-facing tool.

The update was not just paperwork. FDA's 14 July 2025 warning letter to WHOOP had said its Blood Pressure Insights feature was a device, pointing to messaging such as "Higher blood pressure may be an indicator of poor sleep" and a color-coded gauge.

On 17 June 2026 FDA issued a close-out letter saying that, based on changes WHOOP made to the feature and its labeling and consistent with the updated wellness guidance, it does not intend to enforce device requirements for the feature as modified. FDA stressed the letter is specific to that modified feature.

MedTech Dive reported on 24 June 2026 that the changes included adjusting the dial boundaries so the feature would not appear to classify blood pressure clinically.

Oura took a different design route. MedTech Dive reported on 28 May 2026 that Oura planned a wellness feature to track nighttime blood pressure patterns that does not show systolic and diastolic readings, while a separate hypertension risk feature under study would be regulated by FDA.

What it means for a prototype: two companies with similar sensors made different choices about what to show the user. Those choices live in the app screens, the alert logic, the firmware outputs and the marketing copy, so they need to be decided while you are still prototyping, not after launch.

Most modern health wearables turn raw PPG, accelerometer or temperature signals into estimates with machine learning. The model is not what makes a product a medical device. The intended use and claims are.

Apple's Hypertension Notification Feature is a useful contrast: Apple says it is powered by a machine learning algorithm developed with data from more than 100,000 study participants that analyzes PPG signals over 30-day windows, and its pivotal validation study enrolled 2,229 subjects against a cuff reference (Apple validation paper, September 2025).

Because it notifies users of possible hypertension, it was cleared through FDA 510(k) K250507 on 11 September 2025 as a Class II device.

If your AI feature crosses into a medical claim, FDA's device software policies apply.

FDA's draft guidance on AI-enabled device software functions (7 January 2025) is still listed as a draft, and its guidance on predetermined change control plans for AI-enabled device software functions is final (FDA digital health guidance list).

If you stay in wellness, you still need disciplined model versioning, because a model update that changes what the product outputs or implies can change which side of the line it sits on.

Wording that fits the 2026 guidance (as general patterns, not a guarantee):

Wording and design that pushes a product toward device status, per the guidance:

FDA's 2026 general wellness update gives AI health wearables more room to show useful physiologic data, but the room is defined by claims, alerts, validation and non-invasive design. Founders who write the intended use first, prototype the sensor and model with real data, and review screens, alerts and firmware outputs together will ship faster and avoid rebuilding the product around a regulatory surprise.

This article explains public FDA guidance for product planning. It is not legal or regulatory advice.

See also our health and wellness product development services, medical device prototyping and the wearable GPS bracelet case study.