Is Wearable Data Trustworthy? Data Quality and Validation in Clinical Trials

Wearable data is trustworthy in clinical trials when it is captured and validated well, and unreliable when it is not. Wearable data quality depends far less on the device brand than on wear time, completeness, and the validation work behind each measure. A validated sensor that participants stop wearing produces worse data than a humbler device worn every day.

That honest answer matters because the doubt is real. The first question most sponsors ask about a sensor is whether they can trust it for a regulatory endpoint. This guide answers that question the way a study team meets it. You will get the framework that decides whether a measure is reliable, the operational threats that quietly degrade data, the BYOD versus provisioned trade off, and a practical way to design a study for high quality wearable data. WeGuide has run wearable and mobile data capture at scale, including the BRACE trial with the Murdoch Children's Research Institute, so the focus stays practical rather than theoretical.

Is Wearable Data Trustworthy in Clinical Trials?

The short answer is yes, within limits that the sponsor controls. Wearable data is accepted in regulated trials when the device fits the population, the measure is validated, and the chain of capture is documented. The doubt sponsors carry is reasonable, because a sensor reading is only as good as the conditions it was collected in. Trust is earned at the study design stage, not assumed from a datasheet.

Regulators have moved past blanket caution. The FDA treats digital health technologies as acceptable sources of trial data when they are selected and validated for the specific measure. Industry bodies have built the methods that make this practical. So the live question is no longer whether sensor data can be trusted, but how you produce wearable data in clinical trials that a study team can defend later.

WeGuide sees this play out in practice. On the BRACE trial with the Murdoch Children's Research Institute, more than 6,000 participants across five countries recorded over 90% adherence in a six week deployment, with capture running through mobile and remote workflows rather than clinic visits. High adherence is what keeps wearable data complete, which is the first condition of trust. Our overview of wearable device studies shows how WeGuide brings devices into one participant app, and the broader wearables in clinical trials guide sets the wider context.

What Makes Wearable Data Reliable: The V3 Framework

Reliability is not a vibe. It has a method, and the most widely used one is the V3 framework from the Digital Medicine Society. V3 breaks the work into three questions that a measure has to pass before it can carry an endpoint.

Verification asks whether the sensor measures what it claims at the hardware level. Does the optical heart rate sensor report beats accurately against a reference under known conditions? This is bench work, and it is the floor, not the finish.

Analytical validation asks whether the algorithm turns that raw signal into an accurate measure in the people you are studying. A step counter verified on a treadmill still has to count steps correctly for older adults with a shuffling gait. This is where many consumer measures wobble, because performance shifts across populations.

Clinical validation asks whether the measure reflects something that matters to how a patient feels or functions. The clinical validation of a wearable measure is what lets you argue that a change in nocturnal activity, say, is clinically meaningful and not just a number that moved. Skip it and you have a precise measurement of something nobody can interpret.

Get all three right and you can prespecify the measure as a digital biomarker or endpoint and defend it. This is also why device brand is a weak proxy for quality. A measure rides on its validation evidence, not its packaging. We go deeper on the measures themselves in our guides to digital biomarkers and digital endpoints.

The Operational Threats: Wear Time, Missing Data, and Standardisation

A perfectly validated measure still fails if the data never arrives. This is where most real trust problems live, and they are operational rather than scientific. Three threats do the most damage, and a good study plans for all three from the start.

Wear time is whether participants keep the device on long enough to produce a usable signal. A sleep measure needs nights, not naps. Trials set minimum wear time rules, for example a set number of valid hours per day across a defined number of days, and treat data below that threshold with caution.

Missing data is the gaps that open when a device is not charged, a sync fails, or the person takes it off and forgets it. Wearable data completeness is the metric teams watch here, because scattered gaps can bias a result as much as outright dropout.

Standardisation is capturing the same measure the same way across participants and sites. Sensor data validation is hard to defend when one cohort wore the device on the wrist and another on the hip, or when sampling rates differed by phone model.

The table below maps the common risks to their cause and the mitigation that holds them down.

BYOD vs Provisioned and Data Consistency

One operational choice shapes data consistency more than any other: whether participants use their own phone and wearable, known as bring your own device or BYOD, or whether the study ships a provisioned device to each person.

BYOD widens reach and cuts cost, because most people already own a phone and use a device they know. The trade off is variability. Different phone models, operating systems, and wearables sample and process data in slightly different ways, which makes a single clean dataset harder to assemble and validate.

Provisioned devices give the opposite balance. Everyone has the same hardware and settings, so wearable data accuracy is easier to characterise and defend, and you can reach people who do not own a suitable device. The cost is logistics, from shipping and support to recovering devices at the end.

Most studies land on a mix, using BYOD where it is good enough and provisioning hardware where the primary measure or the population demands consistency. Our deep dive on BYOD vs provisioned devices covers the full cost, equity, and compliance decision.

How to Design a Study for High Quality Wearable Data

High quality wearable data is designed in from the protocol, not rescued at analysis. A few decisions, made in the right order, do most of the work.

Start with the endpoint and its validation evidence, then choose the device. Deciding the measure first tells you how much accuracy you need and which validation steps the FDA digital health technologies guidance expects you to document. Picking a device first and hunting for an endpoint later is how teams end up with interesting data they cannot use.

Set wear time rules and define completeness up front. State the minimum valid wear time, what counts as a valid day, and how you will handle missing data in the statistical plan. Writing this before enrolment keeps later decisions honest and supports a defensible analysis.

Monitor data in real time. The cheapest way to protect completeness is to notice gaps while you can still act. Real time monitoring lets a coordinator reach a participant whose device went quiet a week ago, rather than a month after the visit window closed.

Design for the participant. Reminders, simple syncing, comfortable devices, and a single app that people already check keep wear time high and gaps low. This is the quiet engine behind clean sensor data, and it is the same lever that lifts retention. WeGuide is the patient facing layer that brings devices, eConsent, ePRO, and reminders into one branded app, so wearable data quality is built into daily use rather than chased after the fact.

Do these four things and wearable data in clinical trials stops being a leap of faith. It becomes a measure you selected, validated, and captured on purpose, with the wear time and completeness to back it up.

Frequently Asked Questions

Is wearable data accurate enough for clinical trials?

Often yes, when the measure is validated for the specific use and population. Wearable data accuracy depends on the sensor, the algorithm, and who is wearing it. For exploratory or supportive measures, consumer devices can be fine. For a primary endpoint, teams validate the measure carefully or choose a research grade device.

What is the V3 framework for wearable data?

V3 is the Digital Medicine Society method for showing a digital measure is trustworthy. It has three steps: verification, that the sensor is accurate, analytical validation, that the algorithm produces an accurate measure, and clinical validation, that the measure reflects how a patient feels or functions. All three are needed for an endpoint.

How much wear time is needed for valid wearable data?

There is no single number, because it depends on the measure. A sleep metric needs valid nights, while an activity measure needs valid days. Trials set a minimum, often a defined number of valid hours per day across several days, and write it into the protocol before enrolment. Clear wear time rules protect completeness.

What causes missing data in wearable studies?

Missing data usually comes from flat batteries, failed syncs, or participants taking the device off and forgetting it. Less often it comes from app or connectivity faults. Real time monitoring and automatic syncing catch most gaps early, which protects wearable data completeness and keeps the dataset defensible at analysis.

Does BYOD reduce wearable data quality?

Not on its own, but it adds variability. Different phones and wearables sample and process data in slightly different ways, which complicates a clean, comparable dataset. Many trials use BYOD for supportive measures and provision identical devices for primary endpoints, where sensor data validation and consistency matter most.

Does the FDA accept wearable data in clinical trials?

Yes, within its framework. The FDA's guidance on digital health technologies explains how sponsors select, verify, and validate a device for remote data collection. Wearable data is accepted when the device fits the population, the measure is validated, and data handling is documented, so the sponsor can defend the endpoint.