855 - Optimizing Patient Understanding of Spine MRI reports using AI: A prospective Single Center Study

Objectives: This study aims to assess the ability of AI language models to translate spine MRI reports into plain language and examine its impact on patient comprehension.

Design: We conducted a prospective, single-center survey study. We recruited participants who had undergone a spine MRI. MRI reports were input into an AI language model within the secure hospital network using a standardized prompt requesting patient-friendly translations. Participants received both the original provider reports and AI-translated versions and then completed an online survey evaluating their understanding and satisfaction. Responses were categorized using a 10-point Likert scale.

Results:

We included 30 participants, 36.7% were female, and 53.3% were aged 71-80 years. Participants reported significantly better understanding of AI-translated spine MRI reports (8.2, ± 1.92) compared to standard provider reports (5.9, ± 2.29; p < 0.0001). Clarity of explanation was rated higher for AI-translated reports (8.27, ± 1.89 vs. 6.50, ± 2.19; P < 0.001). AI-translated reports were deemed more helpful in explaining their spinal condition's location, nature, and potential implications (7.87, ± 2.19 vs. 6.23, ± 2.50; p < 0.001). With AI-translated reports, participants reported a greater understanding of their medical conditions and potential treatment options (7.30, ± 2.18 vs 5.63, ± 1.97; p < 0.0001). Overall satisfaction was significantly higher for AI-translated reports (8.47, ± 1.98 vs. 6.77, ± 1.99; p < 0.01).

Conclusions: AI-translated spine MRI reports significantly improved patient understanding and satisfaction and were perceived as more helpful compared to standard reports. This technology shows promise in enhancing doctor-patient communication and patient engagement.