Patient perspectives on the use of artificial intelligence to support treatment decision making in renal cancer: findings from the KATY project.

backgroundArtificial-Intelligence (AI) empowered tools are increasingly being assessed for treatment selection in cancer. Patient perceptions on their use will be critical for uptake in clinical practice. We investigated patient willingness to endorse an AI-empowered Personalized Medicine (PM) system (the EU Commission funded “KATY”) to support treatment selection in renal cancer and mapped factors that impact on their perspectives.

methodsThis was a non-interventional, cross-sectional study piloted in Greece through the umbrella Hellenic Cancer Federation (ELLOK). Data were collected through anonymized electronic questionnaires between May and September 2024. Patients were recruited directly by ELLOK (convenience sampling) and provided their full consent prior to enrolling in the study. Results were collected by ELLOK and analysed using SPSS statistical software (version 27.0). For the comparison of proportions, chi-square and Fisher’s exact tests were used. Multiple logistic regression models were used to investigate the association between patient characteristics and willingness to endorse use of the KATY system. Statistical significance was set at p < 0.05.

results84 patients participated in the study. Most (89.3%) felt comfortable with an AI-empowered system supporting physicians with renal cancer treatment selection. Over 50% were willing or extremely willing to endorse the use of a tool with the characteristics of the KATY system. Willingness to endorse was significantly lower in participants who were 65 + years old and significantly higher in participants who were employed/self-employed. Factors that impacted on willingness to endorse were, in order of importance, the system’s contribution to treatment selection accuracy, cost and speed, followed by whether the system impacted on the human aspect of care, on the traceability of responsibility for treatment selection and on data privacy and security. Importance attributed to treatment selection cost was significantly associated with disease stage (p = 0.017), whereas treatment selection speed was significantly associated with time from diagnosis (OR = 0.90, 95% CI: 0.82–0.98, p = 0.013). Time from diagnosis was also significantly associated with privacy and data security, with patients diagnosed more recently attributing significantly greater importance to this aspect (OR = 0.93, 95% CI 0.86 ─ 0.99, p = 0.047).

conclusionsThis survey sheds light into patient attitudes and practices towards AI-powered systems to support treatment decision in renal cancer. It can, therefore, contribute to the wider discussion on ways to enhance patient understanding and potential endorsement of such systems, particularly in life-threatening or severely debilitating conditions such as cancer.