Abstract
BACKGROUND: A large amount of data are generated in health care facilities, yet it is rarely made available for secondary research use. The reasons are manifold. Most importantly, different stakeholders' needs must be balanced. However, there are currently hardly any feasible solutions for this. OBJECTIVE: This study aimed to develop a data trust model with supporting user interface applications to provide a legally and ethically sound framework for secondary use of medical data. The development was based on extensive surveys of various stakeholders. METHODS: Semistructured interviews were conducted with researchers (data users) and institutional representatives of the Medical Center-University of Freiburg, and online questionnaires were administered to patients (data subjects), data users, and institutional representatives. The questionnaire for data subjects covered the dimensions of trust (measured with a 5-point Likert scale), quality of interaction and involvement (measured with a 4-point Likert scale), subjective and objective understanding, and usability (measured with the user version of the Mobile Application Rating Scale). For all other stakeholder groups, the questionnaire focused on usability measured using the user version of the Mobile Application Rating Scale. The surveys comprised a requirement elicitation followed by two rounds of evaluation. Independent-samples Welch t tests were used to compare group means between the first and second evaluations. RESULTS: We devised SouveMed, a framework for secondary use of medical data, applied to the use case of sleep research data. The model includes secure onboarding of data subjects and using digital consent and a digital interface for data users for onboarding, defining research aims, querying the amount of available data, and, finally, either downloading data or having algorithms run on it. At its core is a data trust entity that matches descriptions, consents, and constraints of all stakeholders using digital representations and constraint-solving techniques. Fourteen participants took part in the requirements elicitation, 22 in the first evaluation, and 16 in the second evaluation. In both the first and the second evaluations, data subjects showed a high level of trust in the concept, with mean ratings on the trust scale of 4.23 (SD 0.46) in the first and 4.23 (SD 0.68) in the second evaluation (t15.78=0.03, P=.97). Regarding usability, the mean functionality score of the data user system increased from 3.56 (SD 0.77) in the first to 4.58 (SD 0.38) out of 5 points in the second evaluation (t10.69=-3.28, P=.008). The mean functionality score of the data subject system increased from 4.30 (SD 0.41) in the first to 4.50 (SD 0.74) in the second evaluation (t13.99=-0.75, P=.46). CONCLUSIONS: The SouveMed concept provides a comprehensive framework for the secondary use of medical data. The developed processes can be adapted to other areas of medical research.