5. Operationalize your tech

If the product under investigation is a drug or biologic:

When an sDHT is used to collect data as part of a clinical trial, the overall trial design (e.g., randomization scheme, visit schedule) is reviewed as part of the drug or biologic application. The FDA review division generally does not require a separate review of the full protocol solely because an sDHT is included, unless the sDHT itself qualifies as a medical device.

If the sDHT is a component of the evidence package for a medical device:

When a clinical investigation is conducted to support the authorization or clearance of a medical device, the design and conduct of the protocol are directly relevant to FDA’s review of that device. Whether the investigation is subject to Investigational Device Exemption (IDE) requirements depends on the medical device under evaluation, the study design, and the associated risk profile—not on the inclusion of an sDHT. When sDHT-derived data contribute to the evidence package, FDA will evaluate the appropriateness of the sDHT and the methods used to generate those data as part of its review.

Design for blinding (masking):

In double-blinded trials, if an sDHT displays real-time data to the participant (e.g., a mobile app showing step count or heart rate), it risks unblinding the participant to their treatment status, which can bias subjective reports. To mitigate this, the protocol design must plan to maintain blinding where appropriate (e.g., preventing the participant from viewing data or ensuring the measurement is objective).

Considering practice and learning effects:

For performance outcome measures (PerfOs) using sDHTs (e.g., timed cognitive tasks), repeated testing can artificially improve a patient’s score due to familiarity or learning rather than a true treatment effect. The trial design might incorporate a run-in period or a sufficient number of baseline training assessments to ensure patients have plateaued on the learning curve before receiving the intervention, ensuring any observed change is clinical, not technical.

Considering data yield and adherence:

The study protocol must clearly define and justify expectations for data completeness and adherence (e.g., minimum wear time). This links directly back to your threshold-setting in the monitoring plan and dictates how missing data will be handled in the final statistical analysis plan (SAP).