On July 22, 2026, FDA selected Dexcom as the first participant in a pilot allowing certain digital-health devices to be used in chronic care while manufacturers collect evidence toward marketing authorization.
The Dexcom Glucose Health Program is intended to combine monitoring of metabolic and nutritional status, tailored guidance, and AI insights. Its pilot use is as an aid in screening for prediabetes and type 2 diabetes; FDA adds that it may also contribute to improving glycemic control and lowering HbA1c in people with prediabetes.
Selection did not mean FDA had established the program’s effectiveness for those uses. FDA’s participant page explicitly says effectiveness for the intended uses participating in the pilot has not yet been evaluated by the agency.
How can a device be used in patient care, connected to Medicare payment, before that evaluation is complete?
FDA’s answer is a bounded pilot. It can decline to enforce specified requirements for a selected device and use while the manufacturer gathers real-world evidence. CMS separately determines participation and payment for the care in which the device is used.
Two Programs Address Different Barriers
FDA’s Technology-Enabled Meaningful Patient Outcomes pilot, or TEMPO, operates in connection with the CMS Innovation Center’s Advancing Chronic Care with Effective, Scalable Solutions (ACCESS) model.
TEMPO addresses the regulatory pathway for selected devices. ACCESS addresses how eligible care organizations are paid for managing qualifying chronic conditions.
ACCESS provides recurring payments linked to measurable patient outcomes. It gives participating organizations flexibility to deliver technology-supported care, rather than paying only for a prescribed set of individual activities.
The manufacturer does not receive CMS payment merely because FDA selects its device for TEMPO. It can supply the device to an ACCESS organization. Alternatively, it can itself participate in ACCESS if it separately meets that model’s requirements.
The distinction is fundamental. TEMPO participation is not a Medicare coverage determination or a general right to bill for a device. ACCESS payment supports qualifying condition management by an eligible participating organization.
FDA Can Relax the Pathway Before Use
The Federal Register notice describes how manufacturers request participation. They submit a statement of interest identifying the device and its proposed use and asking FDA not to enforce specified legal requirements.
Those requirements can include premarket authorization, investigational device exemptions, informed consent, and institutional review boards. FDA works with participants to determine when enforcement discretion is appropriate for the particular device.
The manufacturer therefore need not necessarily complete a full marketing application before entering the pilot. The initial request begins a selection and evaluation process. FDA can ask for additional information before deciding whether the device and proposed arrangement are suitable.
Enforcement discretion does not repeal the requirements or constitute marketing authorization. It means FDA does not intend to enforce particular requirements under the identified circumstances. The notice adds that participation is not an indication of whether FDA will issue a positive decision on a later marketing submission.
That scope matters. The pilot does not provide an unrestricted route to market a medical device or a general exemption for healthcare AI.
Clinical Supervision Is One Part of the Arrangement
Eligible devices must be intended for use with clinician-supervised outpatient treatment. Manufacturers must be based in the United States, and their devices must not present potential for serious risk to patient health, safety, or welfare.
TEMPO covers specified conditions within four clinical areas: early cardio-kidney-metabolic, cardio-kidney-metabolic, musculoskeletal (chronic musculoskeletal pain), and behavioral health (depression or anxiety). A digital-health product outside the pilot’s eligibility does not gain access merely because a physician agrees to supervise it.
FDA also considers evidence that the device is adequately safe, functions as designed, and has a reasonable expectation of patient benefit. Additional information can include the manufacturer’s quality system, plans to mitigate risk, and proposed performance goals and statistical analysis.
CMS imposes a separate layer of accountability. ACCESS organizations must meet Medicare participation requirements, designate a physician Clinical Director, and comply with applicable licensure, privacy, and FDA requirements or be subject to FDA enforcement discretion. CMS monitors performance and can terminate organizations that fail its quality, safety, or outcome standards.
The controls therefore extend beyond having a clinician available. They concern the product, the patients, the care organization, and the evidence that will be collected.
Real-World Use Must Produce Usable Evidence
TEMPO manufacturers collect, monitor, and report performance data relating to the intended uses under evaluation. FDA considers their plans for data collection, analysis, reporting, and alignment with ACCESS outcomes.
The notice also identifies proposed interim reporting of adverse events, new risks, and progress against timelines. Manufacturers are expected ultimately to seek appropriate marketing authorization using pilot data together with other information needed for the submission.
The care setting matters because it provides information about performance in actual use. That information still needs a defined question, suitable outcome measures, and an analysis plan. The fact that data came from real patients does not, by itself, establish effectiveness.
FDA’s discretion is limited to the ACCESS-connected arrangement. Offering the same intended use outside ACCESS falls outside TEMPO. The path toward broader authorization remains separate.
The Selected Devices Show Why the Boundaries Matter
FDA’s current participant list includes four products addressing different clinical tasks, two of them in behavioral health.
Cadence’s HypertensionOS is intended to support clinician-supervised, protocol-bound antihypertensive medication initiation and titration when predefined eligibility criteria and deterministic safety checks are satisfied. FDA lists it for use by licensed prescribers such as physicians or advanced practice registered nurses, managing outpatient adults with diagnosed Stage 2 hypertension at initiation.
Limbic’s Unpacked is intended to deliver structured cognitive behavioral therapy-based treatment through an AI voice agent within an outpatient service model, for Medicare beneficiaries with clinically significant depression and anxiety who are already receiving a course of talking therapy. Its listed use includes outcome monitoring and real-time safety flagging. FDA lists extensive contraindications, including suicidal or homicidal ideation and several complex psychiatric conditions. SonderMind’s Adjunctive Care Application occupies the same clinical area on a different model: a smartphone application used as an adjunct to psychotherapy or medication for adults 22 and older, with its own contraindications for current suicidality, mania, psychosis, and PTSD.
These descriptions establish the proposed boundaries of use. They do not prove that the safeguards succeed or that the devices are effective. That distinction remains part of the evaluation.
The examples also show why oversight must be specific. Supporting medication adjustment and delivering a behavioral-health intervention involve different risks, exclusions, and clinical responses. A generic statement that a human supervises the technology cannot describe both arrangements adequately.
