JMIR Publications

ABSTRACT

Background:

Many institutions and medical practices are in various stages of implementation of an ambient voice note generation system for documentation in electronic health records (EHR). In anticipation of University of California Davis Health’s rollout of an augmented intelligence (AI) scribe program, we surveyed current patients to understand and describe their perceptions and concerns about this technology’s integration into their care.

Objective:

To assess patient perceptions about current clinician EHR documentation practices prior to implementation of the AI scribe program, as well as preconceptions and impediments (positive and negative) related to the use of the AI scribe.

Methods:

A convenience sample of 9,171 patients 18 years and older who had a clinic visit within the last year were recruited through a post-visit survey. Demographics, including patient-identified characteristics (age, gender, race/identify) were collected. The survey included rating scales on questions related to the implementation of the AI Scribe program, plus open-ended comments. Summary data were collated and reported.

Results:

1893 patients completed the survey (20% response rate), with another 549 who partially completed the survey. 63% of the respondents were female, and most were 51 or older (87%). Most patients were self-identified as White (69%), multi-race (8%), Latinx (7%), and Black (2%). Patients reacted to the current system of EHR documentation, with 71% feeling heard or sometimes heard, but 22% expressing frustrations that their physician focused too much on typing into the computer rather than direct interactions. When asked about their anticipated response to use of an AI scribe, 48% were favorable, 33% were neutral, and 19% were unfavorable. Younger patients (18-30) expressed more skepticism than those who were 51 and older, but favorable responses were more frequent for all age groups. The favorability did not appear to differ based on race/ethnicity. Comments generally supported the view that use of an AI scribe would enhance the patient encounter experience by allowing the clinician to focus on the patient rather than documentation. However, concerns were raised about privacy, accuracy, and skepticism about an unfamiliar technology. Providing pre-visit patient education and obtaining permission were viewed as very important components to acceptance.

Conclusions:

This patient survey showed that there are valuable opportunities to enhance the patient experience with the implementation of an AI scribe program by allowing the clinician to focus on the patient during the actual encounter rather than the computer for documentation. Patient education and consent are important components to gain patient acceptance when using the AI scribe.