Digital Health Technologies as Trial Endpoints: Fit-for-Purpose Evidence

Scenario Resolution: Under FDA's December 2023 final guidance on Digital Health Technologies (DHTs), establishing that a DHT is fit-for-purpose requires demonstrating that the tool generates accurate, precise, and clinically interpretable data within the specific context of the clinical investigation. Sponsors must execute sensor verification against bench standards, conduct analytical and clinical validation in the target patient population, evaluate participant usability, and pre-specify missing data algorithms in the Statistical Analysis Plan (SAP). Commercial 510(k) clearance and voluntary DDT qualification are separate regulatory determinations that neither confer nor preclude investigation-specific fit-for-purpose evidence.

In December 2023, the U.S. Food and Drug Administration (FDA) issued the final guidance Digital Health Technologies for Remote Data Acquisition in Clinical Investigations under docket FDA-2021-D-1128 (with website content current as of 29 July 2024), formally finalizing the draft issued on 23 December 2021. The guidance defines a Digital Health Technology (DHT) as a system that uses computing platforms, connectivity, software, and/or sensors for healthcare and related uses.

The primary objective of this guidance is to provide actionable recommendations for sponsors evaluating drugs, biological products, and medical devices who wish to utilize continuous or intermittent remote data capture from trial participants. Crucially, the guidance delineates several distinct regulatory boundaries:

• Trial Fit-for-Purpose vs. Device Marketing Clearance: A DHT may be a 510(k)-cleared commercial medical device, an uncleared investigational tool, or a consumer sensor. Whether a DHT meets the device definition in FD&C Act section 201(h) is a separate question from marketing authorization. Clearance or consumer wellness marketing does not establish that the DHT is fit-for-purpose for a specific clinical trial endpoint.

• Enforcement Discretion on Design Controls: Where a DHT is a device used only for remote data collection in a clinical investigation, and sponsors conduct the verification and validation activities recommended in the December 2023 guidance so that the DHT is fit-for-purpose, FDA does not intend to otherwise assess compliance with design control requirements in 21 CFR 820.30, when those requirements apply. That policy does not apply when a DHT that is a device is intended for use outside a clinical investigation, and it is not a finding of commercial device clearance.

• DDT / MDDT Qualification: Voluntary qualification under CDER/CBER Drug Development Tool or CDRH Medical Device Development Tool programs remains entirely independent of IND/NDA trial use. Sponsors are not required to utilize qualified DHTs.

FDA establishes that a DHT is fit-for-purpose when the level of validation associated with the technology is sufficient to support its use, including the interpretability of its data, in the clinical investigation. This determination requires three distinct evidentiary pillars:

graph TD
    A["Raw Physical Parameter: Acceleration / PPG / Pressure"] -->|Verification: Bench Testing| B["Accurate & Precise Sensor Measurement"]
    B -->|Analytical Validation: Signal Processing| C["Digitally Derived Metric: Step Count / Gait Speed / Heart Rate"]
    C -->|Clinical Validation & Usability: Target Population| D["Clinically Interpretable Trial Endpoint: Fit-for-Purpose"]

These verification and validation expectations apply regardless of whether the DHT meets the device definition in section 201(h). Verification is confirmation that the parameter the DHT measures (for example acceleration, temperature, or pressure) is measured accurately and precisely. Validation is confirmation that the selected DHT appropriately assesses the clinical event or characteristic in the proposed participant population (for example step count or heart rate). Usability covers whether participants can use the DHT for the duration and in the manner the protocol describes. Agreement statistics, if used, are a protocol-specific validation method; the December 2023 guidance does not mandate Bland-Altman limits.

Engineering verification should match the DHT's intended use, not a generic bench recipe. The December 2023 guidance asks whether the parameter is measured accurately and precisely; it does not publish frequency, temperature, humidity, or drift windows. Typical protocol questions include:

1. Parameter accuracy and precision: Accuracy and precision of the measured parameter against a reference standard appropriate to that parameter (for an accelerometer, a calibrated motion reference; for temperature or pressure, a calibrated physical standard).

2. Expected use environment: Whether environmental conditions expected in the investigation (home, clinic, storage, and shipping) could affect accuracy, precision, or wearability. Specify the conditions that matter for the device; do not import a generic chamber range as an FDA default.

3. Power, memory, and sampling duration: Whether power, memory, and sampling remain adequate for the protocol-specified wear period so that missingness is not an artifact of battery depletion or storage overflow.

4. Calibration stability over the collection period: Whether calibration remains stable over the duration of data collection, so that longitudinal change can be attributed to the clinical characteristic rather than sensor aging. The stability window is the trial's collection period, not a universal 6-to-12-month specification.

FDA establishes a clear dichotomy between digital technologies that replicate established clinic tests and those that introduce novel digital endpoints:

When a DHT remotely replicates an established clinical measurement (such as continuous ECG monitoring replacing intermittent in-clinic Holter monitoring, or remote spirometry replacing office-based spirometry), the fundamental clinical concept is already accepted. The sponsor does not need to justify the endpoint concept de novo, but must still provide comprehensive verification, analytical validation, and usability evidence for the remote DHT system.

When a DHT captures novel, continuous behavioral or physiological parameters (such as continuous nocturnal scratch duration in atopic dermatitis, free-living gait fluidity in Parkinson's disease, or daytime sedentary bout patterns in heart failure), the sponsor must assemble comprehensive evidentiary justification. The endpoint must be shown to directly reflect how a participant feels, functions, or survives (Meaningful Aspect of Health).

Digital measurements rely on complex data ingestion pipelines from wearable hardware through companion mobile gateways to cloud repositories. Sponsors must ensure data integrity across this pipeline:

• Derived-algorithm specification: Algorithms that derive the endpoint from DHT output should be specified before the trial uses them for confirmatory inference. Unplanned mid-trial changes to those algorithms can make the derived measure difficult to interpret across the study period.

• Source data in the durable repository: FDA considers electronic data in the first durable electronic data repository to which DHT data are transferred to be the source data. Those source data, including associated metadata, should be available for inspection. FDA generally does not intend to inspect individual DHTs for source data when the captured data and metadata are securely transferred to and retained in that repository, and generally does not intend to request machine-level raw voltages that require electronic processing to be understood.

• Clock and event-time alignment: When more than one DHT contributes to the same endpoint or to linked safety review, the protocol should specify how clocks and event times are aligned so that derived measures remain interpretable.

• Secure transfer and access control: Access controls, secure transfer into the durable repository, and protection of participant data should follow the investigation's applicable record-retention and privacy requirements. The December 2023 guidance does not prescribe named cipher suites or equate HIPAA or GDPR compliance with fit-for-purpose evidence.

In late-phase confirmatory trials, sponsors must embed digital health technologies directly into the ICH E9(R1) estimand framework. Continuous remote data acquisition introduces unique intercurrent events and missingness patterns that must be pre-specified in the protocol:

• DHT-Specific Intercurrent Events: Post-baseline initiation of assistive mobility devices (e.g., crutches, wheelchair) when measuring step count, concomitant medications that alter sensor signals (e.g., beta-blockers affecting PPG heart rate), device loss or hardware malfunction, and intercurrent hospitalizations.

• Missing Data Strategies: Continuous monitoring generates missingness from intermittent non-wear, battery depletion, transfer failures, and discontinuation. The SAP should distinguish data that would be meaningful under the estimand but were not collected from periods when the DHT was not worn. Wear-time algorithms, if used, are protocol-specified methods; the December 2023 guidance does not name Choi or Troiano as required procedures.

• Statistical Analysis of Continuous Streams: Analyses of DHT-collected data should be discussed in the SAP, including the definition of each endpoint and the source data from which it is derived (for example, average daily steps across the treatment period). Model choice is trial-specific; the December 2023 guidance does not require hurdle models or generalized additive mixed models.

• Non-Inferiority Trial Caution: FDA's December 2023 guidance explicitly cautions against utilizing DHT-derived endpoints in non-inferiority trials when the historical control effect size was established using conventional clinic visits. Continuous remote measurement may exhibit different variance and effect sizes, potentially compromising non-inferiority margin validity.

The regulatory landscape for digital health technologies continues to evolve actively under the Prescription Drug User Fee Act (PDUFA VII) commitments:

• FDA DHTs for Drug Development Program: FDA maintains an active central program page (content current as of 23 July 2026) coordinating agency-wide regulatory science research on digital measurement.

• FDA August 2026 Paper: Titled Key Considerations for the Development and Use of Digitally Derived Measures for Clinical Investigations (FDA media/194348). This document clarifies the relationship between parameter verification and analytical/clinical validation of digitally derived measures (DDMs). It is an exploratory discussion paper, not formal guidance.

• FDA Statistical Considerations Workshop (27 August 2026): FDA is convening a virtual public workshop on statistical considerations for digitally derived endpoints under PDUFA VII. Workshop deliberations focus on longitudinal functional data analysis, intermittent missingness, and estimand alignment, but do not constitute finalized statistical guidance.

By establishing rigorous verification, clinical validation, and estimand alignment, clinical trial sponsors can successfully deploy digital health technologies to capture objective, continuous real-world endpoints that accelerate drug development and reflect genuine patient outcomes.