Madeline Baumann
Research Mentor(s): Sekhar Sripada
Mentor Department: Psychiatry/Philosophy
Authors: Madeline Baumann, Sekhar Sripada
Session: Session 6 (3:00pm – 3:50pm)
Presentation Type: Poster 38
Abstract
Artificial Intelligence (AI) is rapidly being implemented into many aspects of healthcare, however, its utilization in psychiatric clinical care has been slower. Traditional in person sessions between patients and clinical providers rely heavily on patient retrospective reports which are often unreliable and consume valuable time. Patients may struggle to recall events from previous weeks and heavily rely on recent events, creating a large gap in their stories that may leave out key information. Through the inMyVoice app, patients record a daily voice diary for about 1-2 minutes about events of their daily life noting specific stressors and concerns. Generative AI then transcribes the voice diaries and summarizes clinically relevant information. The clinician can then access the AI generated summaries before patient meetings and develop a comprehensive understanding of relevant events and stressors in the patient’s life. Then, during the in person session, the time can be utilized dissecting the events and building a stronger connection with the patient. In this study, the AI summaries were validated through comparison to established rubrics that grade the summaries on how concise, informative, and organized they were. The AI summaries were rated compared to clinical summaries written by humans trained to read the voice diary transcriptions and summarize issues of clinical significance. The findings of this study show the potential of generative AI to optimize psychiatric care by improving clinical data collection and enhancing the productivity of clinical interactions.
