The Wearable Gap: Why Sensors Are Outrunning the Trials That Use Them

Picture a Phase 2 Parkinson’s trial. The participant is at home, wearing a wrist-based accelerometer that captures tremor frequency 200 times per second — continuous, passive, ecological. The data stream is richer than anything a clinician could capture during a quarterly site visit. And yet, somewhere in the statistical analysis plan, the primary endpoint is still a modified Unified Parkinson’s Disease Rating Scale score, administered in person, at month six. The wearable data is relegated to exploratory status — not because it lacks precision, but because nobody has agreed on what it means yet.

Three signals over the past 90 days point to something the industry hasn’t yet named: the Sensor-Validation Scissor. The hardware is scaling faster than the evidentiary standards required to use it, with wearable integration across interventional drug trials, digital biomarker validation, and emerging sensing modalities, making this gap impossible to ignore. Alongside the FDA’s existing regulatory posture on digital health technologies, a structural crisis comes into focus.

The Validation Floor Doesn’t Exist Yet

The Movement Disorders review catalogs established wearable classes — accelerometers, photoplethysmography-based heart rate monitors, electrodermal activity sensors, continuous glucose monitors — and maps their deployment in interventional trials. The findings are not reassuring to sponsors who assumed the regulatory path was clearer than it is. Across therapeutic areas, wearable-derived endpoints are overwhelmingly used as secondary or exploratory outcomes. The primary endpoint remains anchored to conventional clinical assessments in the vast majority of registered trials covered by the review.

FDA’s 2023 guidance on digital health technologies for remote data acquisition in clinical investigations (the DHT guidance) articulated the agency’s position: sponsors must demonstrate that a DHT-derived measure is fit for purpose, meaning it measures the concept of interest with sufficient accuracy and reliability for the intended use. The language is defensible in principle. In practice, it creates a recursive problem: a measure cannot be fit for purpose until it is validated in a trial, but it cannot anchor a trial’s primary endpoint until it is validated.

Verily’s Project Baseline study, a large-scale longitudinal health study involving continuous wearable monitoring, demonstrated what population-level physiological sensing looks like at scale — thousands of participants, continuous passive data capture, cardiovascular and metabolic endpoints. But even that program’s regulatory translation into pivotal drug trials has been slow, because continuous wearable signals require analytical frameworks that FDA reviewers have not yet standardized. The validation floor doesn’t exist yet. Sponsors are building on it anyway.

The Regulatory Scaffold Is Three Frameworks Behind

FDA’s posture here is understandable — but incomplete. The agency has published its DHT guidance. It has issued the 2021 guidance on patient-focused drug development, which explicitly endorses the use of clinical outcome assessments meaningful to patients in their natural environment. The International Medical Device Regulators Forum has published frameworks for software-based medical devices that touch wearable outputs. On paper, the scaffolding exists.

But open the Nature Reviews Drug Discovery review and look at the emerging sensing technologies section — sweat-based electrochemical sensors measuring cortisol and interleukin-6, textile-embedded biopotential sensors capturing continuous ECG without a rigid device, implantable-adjacent subcutaneous optical sensors reading tissue oxygenation in real time. None of these have a defined regulatory pathway for clinical trial use as primary endpoints. FDA’s DHT guidance was written for the generation of wearables that preceded them.

Consider what this means for CNS drug development specifically. Depression trials have long suffered from the limitations of clinician-administered rating scales — the Hamilton Depression Rating Scale captures a snapshot, typically 17 items, during a structured interview that might last 20 minutes. Passive wearable data on sleep fragmentation, activity rhythms, and heart rate variability can, in principle, provide a 168-hour-per-week behavioral phenotype. Sage Therapeutics’ Phase 3 trial of brexanolone in postpartum depression, for instance, used the Hamilton scale as its primary endpoint while generating a body of wearable-adjacent digital data that could not be brought forward as primary evidence. The sensor knew things the scale did not — and it was excluded from the headline read.

I’ve submitted these packages myself. The reviewer’s question is always the same: “What is the analytical validation for this digital endpoint?” When the honest answer is “we are generating that validation through this very trial,” the endpoint gets demoted to exploratory. The regulation is eating its own tail.

Who Breaks First — and Who Builds on the Wreckage

The operational implications split sharply across stakeholders, and the counterintuitive read is this: wearable technology vendors are not the primary beneficiaries of wearable adoption in trials. The primary beneficiaries are biostatisticians and clinical pharmacologists who can translate sensor streams into regulatory-grade endpoints — and there are not enough of them.

For sponsors running Phase 2 CNS, cardiovascular, or metabolic trials right now, the Nature Reviews Drug Discovery review’s analysis of accelerometry in movement disorders is a direct strategic signal. Accelerometry is the most mature wearable modality with the longest validation trail in drug trials. Roche’s Phase 3 trial of risdiplam in spinal muscular atrophy used wrist-worn accelerometry to capture motor function endpoints in a population that cannot reliably complete pen-and-paper assessments. That trial generated one of the cleaner regulatory acceptance cases for a wearable-derived supportive endpoint in a pivotal submission — not because the technology was novel, but because the analytical validation was built into the protocol from the start, and the concept of interest (upper limb motor activity) mapped cleanly to what the sensor measured.

For CROs, the structural challenge is that wearable data integration is being sold as a site-burden-reduction play — fewer in-person visits, more remote capture, lower per-patient cost. The Nature Reviews Drug Discovery review’s data on adherence rates complicates that narrative. Wearable adherence in trials drops significantly beyond eight weeks of continuous wear, with some studies reporting less than 60% device wear-time compliance at the 12-week mark. Remote data capture is only cheaper if the data is actually captured. Sites that lose visibility into whether a participant is wearing the device — which is what happens when you remove in-person touchpoints — face a new category of protocol deviation that existing GCP frameworks were not written to address.

Technology vendors face the sharpest near-term disruption. The emerging sensing modalities catalogued in the Nature Reviews Drug Discovery review — electrochemical sweat sensors, smart textiles, continuous biomarker patches — represent the next product generation. But FDA’s Breakthrough Device Program, which has been used to accelerate digital health tools in diagnostics, has not been applied consistently to wearable endpoints in interventional drug trials. Without a defined accelerated pathway, vendors selling next-generation sensors into the clinical trial market are doing so against a regulatory approval horizon that is effectively undefined. Several companies building continuous cortisol monitoring platforms are already in early-phase feasibility studies with academic medical centers, but the commercial pivot to pivotal trial use will require FDA to update its DHT guidance to address non-traditional biofluid sensing — a document revision the agency has not yet signaled.

The Sensor-Validation Scissor closes in one of two ways. Either FDA moves proactively — issuing updated DHT guidance that creates fit-for-purpose validation frameworks for next-generation sensing modalities, potentially through a qualification pathway analogous to the Drug Development Tool qualification process established under PDUFA commitments — or the industry builds validation evidence outside of pivotal trials and arrives at the agency with completed dossiers that reviewers have no precedent to evaluate. The second scenario has happened before. It is what produced years of regulatory uncertainty around patient-reported outcomes before the 2009 PRO guidance finally provided a framework. The wearable endpoint is following the same arc — and the sensors are moving faster than PROs ever did.

Watch for the first NDA or BLA where a wearable-derived endpoint is proposed as a primary efficacy measure in a Phase 3 submission. That filing — whenever it arrives — will force FDA’s hand in a way that guidance development timelines cannot. The sponsor who builds that dossier correctly will not just win an approval. They will write the template everyone else uses for the next decade.