gms | German Medical Science

Background and state of research: Artificial intelligence (AI) can provide valuable information for doctors and patients when it comes to patient care. For example, in nephrology it can help to identify patients at risk for rejection after kidney transplantation. Due to a lack of empirical data on technical, clinical, and ethical issues, however, the impact of AI on doctor-patient interactions remains under-studied. Our study PRIMA-AI aims to explore the effects of artificial intelligence-based decision support (AI-DSS) on the shared decision-making process in post-kidney transplant care in a German kidney transplant center (KTC), as perceived over time by the patient, their support person (SP), and the treating clinician.

Method: A two-year prospective, randomized, two-armed, parallel-group, single-center trial is conducted at the KTC. The study population will consist of 100 adult kidney transplant recipients (KTR) and SPs who will be randomized to either routine care or AI-supported care. Patients will be interviewed at randomization as well as three, six, twelve, and 24 months after randomization. The primary endpoint will address differences in individual key outcome parameters (e.g. patient and graft survival, patient-reported outcome measures). Three AI-based prediction models will be tested regarding real-time risk estimation for rejection, grafting loss, and urinary tract infection considering a period of 12 months. The models are based on the semi-structured EHR (electronic health records) data of the KTC, including, for instance, clinical, laboratory, immunological, and histopathological data. This study is registered with ClinicalTrials.gov as NCT022663. A nested qualitative study will assess in-depth changes in patients’, SPs’ and clinicians’ views on the use and consequences of AI on SDM.

Implication for care: To the best of our knowledge, PRIMA-AI is the first systematic longitudinal study of physicians, patients, and SPs’ views on AI's impact on shared decision-making following kidney transplant. By using a robust, interdisciplinary, mixed methods approach, this project will also provide guidance on an ethics and governance framework for the use of AI-informed SDM in clinical practice to allow for a better understanding and informed discourse among different stakeholders from policy, science and society on the role and impact of AI on physician-patient interaction.

Funding: BMBF-Strukturförderung Versorgungsforschung/Nachwuchs; 01GP2202A