219 - A Pilot Study Incorporating AI-Based Avatar Digital Engagements for Patients in a Community Radiation Oncology Practice

Purpose/Objective(s): Patient-reported teach back and satisfaction scores in our large community-based radiation oncology department, and across the nation, have declined, despite annual refinements in education materials, in large part due to reduced engagement between doctors and patients. To address this challenge, we sought to explore “first touch” digital engagements. The growing capabilities in artificial intelligence led us to conduct a pilot study to assess the feasibility of utilizing videos, including AI-generated video avatars of our treating physicians, to engage with patients before their consultations. We hypothesized that pre-consultation digital engagements, with multiple choice teach back, might improve communication, patient understanding, and patient satisfaction.

Materials/Methods: We collaborated with ASTRO and a commercial patient engagement company to generate pre-consultation digital engagement materials about radiation treatment options for consecutive patients. We assessed patient comprehension via teach back quizzes and collected satisfaction scores. We used ASTRO patient education videos in the digital engagement, and subsequently evolved to an AI-avatar based video generated for specific physicians.

Results: 312 patients were offered video digital engagements prior to their initial radiation oncology consultation, with 290 (93%) completing both the video and the survey. Of those, 267 received an ASTRO video; subsequently, 23 received a digital avatar-based video. Teach back scores were collected with 100% patient compliance. 93% completed the engagement prior to entering our department. Among the rest, patients age > 75 were less likely to complete the digital engagement at home, usually due to technology barriers. Patient satisfaction scores (mean 4.32 / 5) were high among patients receiving ASTRO video engagements and remained high (mean 4.33 / 5) among patients receiving the personalized AI avatar videos, within the limitations of small sample size for the latter cohort. AI avatar videos use personalized scripts and illustrations, which can be updated as clinically needed. Further data collection is ongoing.

Conclusion: In collaboration with ASTRO, we successfully implemented pre-consultation digital engagement education materials for our patient population, with outstanding patient teach back scores and satisfaction rates. To our knowledge we are the first community radiation oncology practice to implement such a program using AI avatars to educate patients. Furthermore, we demonstrate bidirectional engagement whereby patients immediately provide feedback about their understanding of treatment options. Pre-educated patients have yielded more focused and higher value engagement in the clinic. The study is expanding to all radiation oncologists in our health system which will facilitate a larger dataset and more granular data analysis.