A hierarchical framework for selecting reference measures for the analytical validation of sensor-based digital health technologies (Preprint)

UNSTRUCTURED Sensor-based digital health technologies (sDHTs) are increasingly used to support scientific and clinical decision-making. The digital clinical measures they generate offer enormous benefits, including providing more patient-relevant data, improving patient access, reducing costs, and driving inclusion across healthcare ecosystems. Scientific best practices and regulatory guidance now provide clear direction to investigators seeking to evaluate sDHTs and document claims related to their suitability for use in different contexts. However, the quality of the evidence reported for analytical validation of sDHTs —evaluation of algorithms converting sample-level sensor data into a measure that is clinically interpretable— is inconsistent and too often insufficient to support rigorous claims based on digital clinical measures. We propose a hierarchical framework to address challenges related to selecting the most appropriate reference measure for conducting analytical validation, as dictated by the medical claims related to the sDHT algorithm, and codify best practices and an approach that will help capture the greatest value of sDHTs for public health, patient care, and medical product development.