JMIR Publications

ABSTRACT

Background:

Physicians currently experience high levels of burnout and professional stress, and innovative tools, such as AI-based interventions, may hold promise for improving these outcomes. However, there is limited evidence of the effectiveness of these tools on improving physician professional wellbeing.

Objective:

The primary aim of this systematized review was to examine and synthesize the emerging evidence of the impact and implementation of AI-based interventions to determine whether these interventions effectively address physician burnout, stress, and workload and to map the existing evidence base.

Methods:

A systematized review was conducted and guided by the Preferred Reporting Items for Systematic Review and Meta-Analyses guidelines. Studies were included if they examined impacts of AI-based interventions on burnout, professional stress, and workload among physicians. Narrative synthesis was used to summarize the findings.

Results:

Seven studies met inclusion criteria with most studies conducted within United States health systems (5/7), evaluating AI-based scribes with EHR integration (5/7), and with a median study sample of 30 participants. Reviewed studies were a mix of randomized studies and non-randomized quasi-experimental studies. All study participants were physicians with a range of specialties across studies and within, including primary care, pediatric subspecialties, anesthesiology, radiology, or mixed outpatient physicians across specialties. The majority of studies focused on outpatient or patient-facing care settings (5/7). Significant improvements were reported in 4/5 studies on burnout and 3/6 studies on professional stress or workload. The majority of studies did not conduct formal safety or risk monitoring (6/7) and one study reported a mild adverse event.

Conclusions:

While the empirical literature examining the potential of AI-based interventions to effectively address physician burnout, stress, and/or workload may be limited and underdeveloped, preliminary findings suggest that AI-based scribes may hold promise in addressing physician burnout. However, these findings must be interpreted with caution due to the numerous methodological limitations identified across studies included in the present review. To build a stronger empirical evidence base, future research must prioritize more rigorous and implementation-informed designs to more effectively evaluate the complex, real-world contexts in which these tools are being tested.