Open Notes in Brain Injury Rehabilitation Part 2: Provider and Patient Preferences for Artificial Intelligence Solutions

Abstract

Objective: To characterize opinions and experiences related to use of artificial intelligence (AI) in the context of patient-accessible electronic medical records in rehabilitation (PAEHR or "open notes").

Design: Rehabilitation providers and patients with acquired brain injury (ABI) completed a web-based survey.

Setting: Provider participants were recruited via publicly-available email lists published by licensure boards and via internal listservs within a large Midwestern healthcare system. Patient participants were recruited via medical records search within the same healthcare system.

Participants: 677 providers and 270 patients with ABI completed the survey. Providers (physical therapy, occupational therapy, speech-language pathology) were working in settings that utilize electronic health records. Adults with ABI had received rehabilitative care between 2021 and 2024.

Interventions: Not Applicable.

Main outcome measures: Participants responded to multiple choice, Likert-scale, and open-response questions about their experiences with and opinions about AI in the context of clinical documentation in rehabilitation.

Results: Both groups reported little experience with AI in daily life or healthcare contexts. Both groups reported concerns regarding AI accuracy and privacy. Most providers expressed interest in AI tools for supporting clinical documentation; patients with ABI expressed more hesitancy. Both groups agreed that a patient-directed summary would be helpful for improving the utility of open notes. Both groups rated Goals, Progress, Activities, and Home Exercise Programs as the most important note components for patients to understand.

Conclusions: Rehabilitation providers and patients with ABI agree that patient-directed note summaries may be one way to improve open notes. Generative AI provides one potential pathway for creating such a summary without undue clinician burden, but both providers and patients express significant concerns about accuracy and privacy. Future work should continue to evaluate the potential of AI solutions for improving patient-provider communication in the open notes era, while mitigating potential drawbacks.

Keywords: Artificial Intelligence; Brain Injuries; Electronic Medical Records; Rehabilitation.