Half of clinicians report positive ambient AI experience

Approximately half of clinicians who used an ambient AI clinical documentation tool reported a positive outcome, according to a study published in JAMA Network Open.

Ambient AI tools use automatic speech recognition and natural language processing to draft EHR documentation based on patient-provider conversations.

The nonrandomized clinical trial recruited family medicine, internal medicine and general pediatrics clinicians from all outpatient clinics in North Carolina and Georgia within Atrium Health.

Clinicians who accepted the invitation received an hour-long training on the AI tool -- Dragon Ambient eXperience (DAX) Copilot -- from health IT vendor Nuance.

The researchers assembled a comparison group of clinicians through two approaches to identify practice locations and specialties like those of the intervention group.

First, service line leaders encouraged clinicians to participate as controls. Second, researchers invited clinicians who expressed initial interest in the AI tool but opted out after the informational meetings to participate as controls.

Researchers sent out a seven-question survey to 230 participants pre-implementation and five weeks after post-implementation to assess the impact of the AI implementation on clinicians' EHR experience.

The study found that 47.1% of intervention respondents reported decreased time on the EHR at home compared to 14.5% in the control group. Additionally, 43.5% of intervention respondents reported decreased time on clinical documentation after visits compared to 18.2% in the control group.

Moreover, 44.7% of intervention respondents reported less frustration using the EHR in comparison to 14.5% in the control group.

However, an average of 44.7% of participants in the intervention group and 68.7% in the control group reported that their EHR experiences were comparable before and after the intervention.

The researchers noted that study limitations include potential selection bias and recall bias in both groups.

"Further research is needed to identify opportunities for improvement and understand the impact on different clinician subsets and health systems," the study authors concluded.

Hannah Nelson has been covering news related to health information technology and health data interoperability since 2020.