Abstract 4142032: Hospital Discharge Education For Heart Failure Patients Delivered by Conversational Agent Technology Improves Patient Experience and Is Preferred to a Doctor or Nurse

Introduction: . The hospital discharge process is non-standardized and contributes to post-discharge adverse events, re-hospitalization and high cost. To address this, we developed the Re-Engineered Discharge (RED) program designed to provide high quality discharge education, shown to lower readmission rates and improve patient satisfaction. Integrating digital technologies could facilitate its implementation into hospital workflow. Methods: We designed “MayaRED”, a digital conversational agent that simulates face-to-face human interactions through personalized, conversational dialogue to deliver discharge education. MayaRED employs the “teach-back” method to document patient understanding. Upon completion, a report documents what was taught by MayaRED and understood by the patient. Participants were asked to complete a survey regarding their perceptions of the acceptability and usefulness of MayaRED. Participants were matched to controls subjects to evaluate the comparative effectiveness. Results: HF patients (N=18) using MayaRED averaged 69 years, 13,003 pg/ml pro-BNP, 42% ejection fraction and 3 mg/dl creatinine were compared to 17 matched controls. On a 1 to 7 point Likert scale (1=strongly disagree; 7=strongly agree), patients using MayaRED reported that they had “their questions answered” (mean 6.47 vs. 4.76; P<.0001), “understood their follow-up appointments” (6.73 vs. 5.59; P<.0001), “understood their medicines” (6.87 vs 4.59; P<.0001), and understood “how to care for themselves at home” (6.07 vs 4.65; P<.0001) (Figure 1). Intervention patients reported that they trust MayaRED (6.80), “are likely to follow Maya’s advice” (6.66), and are “confident in Maya’s ability to help me” (6.66). When asked “If you were hospitalized again at SVH, would you like to interact with Maya again?,” all patients responded “yes”. When asked “would you rather have the nurse/doctor or Maya give you your information about your discharge, all 18 responded MayaRED (Figure 2). Preliminary data show the 30 day all-cause readmission rate for those receiving MayaRED was 16.6% vs 29% for controls. Conclusion: The MayaRED conversational agent system is well-accepted by HF patients, demonstrating its feasibility as a tool for delivering discharge education. These findings support the continued implementation and rigorous testing of MayaRED in diverse clinical settings.