Overcome The 5 Most Common Hurdles To Turning Wearable Data Into Usable RWE

By Rashmi Gurnani

Most clinical teams I work with no longer debate if trial participants should be using wearables. A more difficult question arrives after collecting data for several months, when a statistician opens a folder holding billions of wearable data samples and asks what exactly the study measured. We’ve witnessed that collecting wearable data has become easy, but turning it into evidence that a regulator, payer, or safety committee will accept is still a big challenge.

Regulators have signaled that they are open to accepting wearables data as evidence. The FDA finalized its guidance on digital health technologies (DHTs) for remote data acquisition in December 2023.¹ In July 2023, the EMA qualified stride velocity 95th centile (SV95C), a wearable-derived measure, as a primary endpoint in trials of ambulatory patients with Duchenne muscular dystrophy (DMD), the first digital endpoint to reach that status.² Although they are embracing wearables as generators of RWD, regulators still expect sponsors to show that the measure is valid, the data is trustworthy, and the gaps are explained.

Across clinical research and digital health programs, I see the same six hurdles keep wearable data from becoming usable RWE. Thankfully, there are practical solutions, offered here so that other teams can build those fixes into their protocols today.

1. Teams Choose The Device Before The Measure

Too many teams start a program with a familiar device brand or an existing vendor contract then search for an endpoint the device can produce. The result is a metric that is easy to collect but loosely tied to anything patients benefit from. Daily step count, for example, says little about a patient whose real limitation is climbing stairs or standing up from a chair.

How teams can solve it:

• Start from the concept of interest. Write one line that links a symptom or functional limitation to an observable behavior, then to a digital measure, then to its context of use. If the team cannot write that line, the measure is not ready.
• Ask for evidence layer by layer. The verification, analytical validation, and clinical validation (V3) framework separates sensor performance, algorithm accuracy, and clinical meaning.³ Request evidence for each layer from the technology provider and treat a gap in any one as a study risk.
• Check the algorithm against your population. For example, step and gait algorithms built on healthy adults can undercount slow walkers or people who use walking aids. Confirm performance in a population that resembles the one you’re studying.
• Lock versions in the protocol. Record the device model, firmware, and algorithm version, and define how the study will handle updates mid-trial.

2. Missing Data Are Rarely Random

Participants take devices off when they feel worst: during a flare, a hospitalization, or a week of fatigue. Those gaps correlate with the very outcome the study measures, so simply analyzing the days with data biases the result toward healthier days.

How teams can solve it:

• Define a valid day before the first participant enrolls. Specify wear-time rules in the statistical analysis plan. Many accelerometry studies follow the convention of at least 10 hours of wear for at least four days, used in U.S. national survey data, ⁴ but justify the threshold for your population.
• Separate non-wear from inactivity. A participant lying still and a device sitting on a nightstand can look alike in raw counts. Use wear-detection signals such as skin temperature or capacitance where the device supports them.
• Monitor adherence centrally. Flag participants whose devices have not been synced for 48 hours and route the alert to the site coordinator, not just to a dashboard that the coordinator may not check.
• Plan for intercurrent events. Use a pre-defined framework to state how the analysis treats hospitalizations or device loss and pre-specify sensitivity analyses that assume data are missing, and not at random.

3. The Data Are Too Big And Too Derived

A wrist-worn accelerometer sampling many times per second over a long study produces gigabytes of data per participant. Meanwhile, many consumer platforms deliver only summary metrics calculated by proprietary algorithms that can change without notice. Teams end up with either more data than they can process or less transparency than they need.

How teams can solve it:

• Contract for raw data access. Negotiate access to raw or high-resolution data and algorithm version logs before the study starts. Raw data lets the team re-derive endpoints if an algorithm changes or a regulator asks.
• Separate raw storage from derivation. Store raw files unchanged, compute derived features with version-controlled code, and document every transformation.
• Adopt standards early. The FDA recommends public data exchange standards for DHT data.¹ Decide how derived endpoints will map into submission datasets while the protocol is still in draft.
• Budget for data engineering. Device costs are visible in the budget; pipeline costs often are not. Fund both.

4. Provenance Breaks Between Wrist And Database

An inspector who asks where a value came from expects a complete answer: which device, worn by which participant, recorded when, and transformed how. ICH E6(R3), adopted in January 2025, expands GCP expectations for data governance across the full data lifecycle and for computerized systems.⁵ Wearable platforms now sit squarely inside that scope.

How teams can solve it:

• Treat the wearable platform as a computerized system. Include it in the system inventory with its validation status, access controls, and audit trail.
• Get linkage right at dispensing. Record which device serial number went to which participant, and log every swap, loss, or replacement.
• Standardize time. Store timestamps in a single reference time zone with the participant's local offset. Time-zone errors corrupt sleep and circadian measures.
• Qualify the provider properly. Audit subprocessors, data residency, and change control, not just the device itself.

5. Participant Burden Undermines Data Quality

Charging, syncing, and smartphone dependence all fall on participants, and the burden is not equal. Older adults, people in rural areas, and participants with unreliable connectivity drop off first, which skews both the data and the population the evidence represents. In a survey of clinical research professionals in India, 71% named operational challenges as their main concern with remote data capture and wearables.⁶

How teams can solve it:

• Pilot in the target population. A two- to four-week usability run before the protocol is locked surfaces charging problems, confusing instructions, and connectivity gaps while they are still cheap to fix.
• Choose for the real world. Favor long battery life and on-device storage that tolerates days offline. Where connectivity is uneven, provide phones and data plans rather than relying on participants’ own devices.
• Collect only what the endpoint needs. Turn off location tracking and other streams the analysis will not use and say so in the consent form.
• Write consent for continuous collection. Explain what the device records, how long data are kept, and whether they may be reused, in line with applicable laws such as GDPR, HIPAA, and India's Digital Personal Data Protection Act, 2023.

6. Nobody Owns The Whole Chain

Clinical operations manages devices, data management owns the EDC system, biostatistics writes the SAP, and IT runs the cloud environment. Wearable data is cross all four, and problems surface in the gaps between them, usually at database lock.

How teams can solve it:

• Name a digital measures lead. Give one person accountability from endpoint selection through submission.
• Write a short digital endpoint charter. Before protocol sign-off, agree in one document on the concept of interest, V3 evidence, wear-time rules, data flow, and analysis approach. Every function signs it.
• Engage regulators early. The FDA encourages sponsors to discuss planned DHT use with the relevant review division early.¹ In Europe, scientific advice and qualification procedures offer a similar route.

Start Small, Document Everything

Wearable data become evidence through discipline not device sophistication. Pick one wearable-derived secondary or exploratory endpoint in your study and run it with full rigor: a written concept of interest, V3 evidence, prospective wear-time rules, a traceable pipeline, and a named owner. Each study then adds to an evidence package the organization can reuse.

The sponsors who treat the sensor as a measurement instrument, and the data pipeline as a regulated system, will be the ones whose wearable data become usable RWE.

References:

1. U.S. Food and Drug Administration. Digital Health Technologies for Remote Data Acquisition in Clinical Investigations: Guidance for Industry, Investigators, and Other Stakeholders. December 2023.
2. Servais L, Strijbos P, Poleur M, et al. Evidentiary basis of the first regulatory qualification of a digital primary efficacy endpoint. Scientific Reports. 2024.
3. Goldsack JC, Coravos A, Bakker JP, et al. Verification, analytical validation, and clinical validation (V3): the foundation of determining fit-for-purpose for Biometric Monitoring Technologies (BioMeTs). npj Digital Medicine. 2020;3:55.
4. Troiano RP, Berrigan D, Dodd KW, et al. Physical activity in the United States measured by accelerometer. Medicine & Science in Sports & Exercise. 2008;40(1):181–188.
5. International Council for Harmonisation. ICH E6(R3) Guideline for Good Clinical Practice. Adopted January 6, 2025.
6. The role of remote data capture, wearables, and digital biomarkers in decentralized clinical trials.

About The Author:

Rashmi Gurnani, MSc, MBA, is a seasoned healthcare technology leader and the CEO of AKT Health Analytics in Hyderabad, India. With nearly 15 years of experience across the life sciences, pharma IT, and healthcare IT sectors, her expertise spans clinical research, pharmaceutical analytics, digital health innovation, and RWE generation. She holds a master’s degree in clinical research and an MBA in clinical research management. Throughout her career, Rashmi has successfully led cross-functional programs and driven pre-sales, business development, and client engagement initiatives for pharmaceutical companies and CROs across India, the U.S., Japan, and the MENA region.