Artificial Intelligence EHR Integration Cuts Down on EHR Screen Time

An artificial intelligence (AI) EHR integration helped physicians extract patient health information more efficiently, according to a study published in JAMA Network Open.

Researchers recruited 12 gastroenterology physicians/fellows at an academic medical center to analyze how an AI integration affected EHR use time and data accuracy.  

The AI tool extracted relevant patient health data and displayed it alongside the original record.

Each clinician received an AI-optimized patient record and a standard patient record. Using each record, clinicians had to answer 22 questions that required them to find clinically relevant information in the assigned record. Clinicians reviewed records from June 1 to August 30, 2020.

The AI system cut EHR use time by 18 percent; standard record review took 12.8 minutes on average, while AI-optimized record review took clinicians 10.5 minutes.

What’s more, there was no significant decrease in accuracy of patient information; data accuracy was 83.7 percent with the AI-optimized record and 86.0 percent without the AI-optimized record.

Survey results found that 11 of 12 physicians (92 percent) preferred AI-optimized record review to standard review.

While the clinicians noted a learning curve, 11 of 12 said they believed that the health IT solution would save them time and were interested in using the EHR integration in their clinic.

“Our results indicate a positive AI experience and desire by physicians to use such a system in their practice,” the study authors wrote. “The single clinician who expressed uncertainty about using the software had concerns regarding the amount of clicks it would take to go to various pages of the packet in our user interface and thought this could be an inconvenience.”

On average, clinicians reported that they thought the software could help them save 14.5 minutes per new patient encounter. The study authors noted that these potential time savings are notable, as new patient encounters typically last 30 minutes at the academic medical center.

While the time difference between answering the standardized questions with and without AI optimization was used as a marker of potential time savings, many factors such as referral packet size and data complexity would affect actual time savings, the researchers noted. 

AI-optimized record review could give providers more time to spend with existing patients, or even allow them to open new patient visit slots to decrease wait times, the authors explained.

The study also found that clinicians who spent more time on patient data extraction using standard record review would benefit most from the AI solution.

“This is an important association because it can estimate which physicians may gain the most from the use of such an AI system in actual clinical practice,” the study authors wrote.

As clinicians become more familiar with the user interface over time, the AI integration could lead to even greater time savings, they noted.

“In general, the supportive responses of our survey highlight the importance of this issue as an area of need that can likely be generalized and expanded to multiple other medical subspecialties that share similar challenges, because many referral records contain similar types of information (eg, progress notes, radiology reports, pathology findings, procedure notes, etc),” the study authors explained.

As the digital health transformation continues and new ONC regulations break down industry data siloes for increased interoperability, providers will gain access to more patient health information than ever.

While this data may aid in clinical decision support and care coordination, the influx of new patient health information could also add to clinician burnout, which is already a critical issue affecting the healthcare workforce. AI-based tools could help mitigate clinician burnout through EHR optimization.