Artificial Intelligence Documentation Assistant Shows Promise for Healthcare Charting

By Jonathan Springston, Editor, Relias Media

An artificial intelligence (AI) documentation assistant provided some relief to physicians who were struggling with electronic health record (EHR) and other healthcare documentation burdens, according to the results of an American Academy of Family Physicians (AAFP) Innovation Labs Phase II report.

Several companies have developed an AI documentation assistant, including Suki, Notable, Robin, Saykara, and more. The AAFP partnered with Suki to use its solution, a voice recognition tool similar to Siri or Alexa. Through deep learning, this tool listens to a physician’s voice commands continually, learning and adapting along the way, becoming accustomed to desired documentation needs and patterns. Imagine a virtual or digital medical assistant who completes medical charting instead of physicians spending hours every week planted in front of computer screens managing these tasks.

After a tiny study group reported favorably in a Phase I report, AAFP expanded the pool of users for a Phase II run. Here, AAFP recruited a cohort of 132 participants from 47 clinics in 18 states to try the Suki AI assistant for free for 30 days. Of these, 102 completed the trial and were included in the results. Investigators measured four metrics: rate of adoption, time savings, burden and burnout reduction, and professional satisfaction.

Of the final 102 participants, 61 decided to continue using the AI assistant after the study and become paying subscribers. Of these adopters, there was a 72% reduction in median documentation time per note, for an overall time savings of 3.3 hours per week per participant. On a satisfaction scale of 1 (lowest/worst) to 10 (highest/best), overall practice satisfaction mean values rose from 6.4 to 7.7, workload satisfaction mean values rose from 5.3 to 6.9, and EHR satisfaction mean values rose from 6.4 to 7.4. Overall stress/burnout indicators skewed lower among participants after the trial ended.

AAFP says it is moving on to Phase III, where the group will work to build awareness of this AI assistant by offering the solution widely to its membership to try for free.

AI solutions are creeping into healthcare, and some believe AI ethics should be taught in medical school to help future physicians and nurses learn about what is ahead. Although juries might be sympathetic to a physician who made a mistake while using an AI solution, liability exposure and malpractice risks abound.

For more on this and related subjects, be sure to read the latest issues of Healthcare Risk Management and Medical Ethics Advisor.