Early experience with a patient-facing AI chatbot integrated in a patient portal

Background and significance Many patients struggle to interpret lab and test results. Objective To evaluate patient experience, usability, and intention to reuse an AI assistant integrated into a patient portal’s test results interface, aimed at enhancing patient comprehension of laboratory and diagnostic findings. Methods A cross-sectional study with 131 adults using a non-production portal instance. Participants accessed their lab results via the embedded AI assistant and completed a post-use survey. Outcomes included System Usability Scale (SUS), reuse intention (5-point Likert), and Net Promoter Score (NPS). Multivariable regression examined associations of outcomes with patient characteristics. Results Mean SUS was 81.0 (SD 16.1). Most participants (76.9%) were very or extremely likely to reuse (mean 4.1/5.0). NPS was 43.9%. Higher reuse intention was significantly associated with more diagnoses [mean difference = 0.017, 95% CI (0.002, 0.032), p = .032]. Conclusions A secure, portal-integrated AI assistant showed strong usability and satisfaction, particularly among patients with greater comorbidity.