• Phase Two Trial of AI Assistant for Documentation a Success

    Latest Innovation Lab Results Show Time Savings, Burnout Reduction

    December 17, 2021, 3:48 p.m. News Staff — How much time do you spend on documentation and other administrative functions each week? 10 hours? 15? More?

    Physician smiling at camera

    Turns out, physicians spend nearly 16 hours, on average, completing these tasks each and every week. For family physicians, the number is even higher.

    Is it any wonder, then, that for the past two years the overall percentage of physicians who report being burned out has hovered at 42%? Among family physicians in 2021, that figure is 47%.

    Granted, burnout can’t be traced solely to administrative burden, but it’s certainly a major contributor. Which is why the AAFP Innovation Lab has for years focused on investigating how the latest proven, advanced technological platforms and tools — cloud, artificial intelligence/machine learning, and voice and mobile technologies — can be harnessed to reduce administrative burden and family physician burnout.

    This week marks a major milestone in those efforts, with the AAFP releasing results from phase 2 of its Innovation Lab on the adoption, use and impact of an AI Assistant for Documentation by family physicians and other primary care physicians.

    Story highlights

    After partnering with Suki, an industry leader in voice AI technology for health care, in a phase one lab that successfully demonstrated that using an AI Assistant could reduce family physicians’ documentation burden and help ameliorate burnout, the Academy moved forward with phase two.

    “AAFP’s series of Innovation Labs identify and demonstrate innovations essential to optimizing the family medicine experience,” said Steven Waldren, M.D., M.S., AAFP vice president and chief medical informatics officer, in a news release. “Our initial lab provided proof that the use of an AI Assistant can greatly reduce documentation burden and family physician burnout, and Phase 2 of the lab series determined whether an AI Assistant is essential to and readily adoptable by family physicians.”


    The phase two lab examined the level of adoption and impact of an AI Assistant for visit note completion among family physicians and other primary care clinicians during a 30-day free trial. An initial cohort of 132 participants from 47 clinics in 18 states was recruited; of that number, 102 completed the trial and were included in the final results.

    The lab evaluated the following metrics:

    • Rate of adoption — with “adopter” being defined as a clinician who personally trialed the solution and then became a paying subscriber of Suki;
    • Time savings — measured by comparing documentation time per note before and after the 30-day trial;
    • Burden and burnout reduction — gauged using a standard question (i.e., one-item version of the Maslach Burnout Inventory Emotional Exhaustion scale) regarding level of burnout and stress before and after using the innovation; and
    • Professional satisfaction — measured across three domains (i.e., overall practice satisfaction, workload satisfaction and EHR satisfaction) before and after the trial.


    Of the 132 individuals who initially signed up to participate in the lab, 117 were considered to have successfully trialed the AI Assistant by completing at least five notes using the tool. Of those, 15 participants’ organizations have not finalized their decision-making or contracting process to convert to paying subscribers. As for the remaining 102 participants, 61 clinicians (60%) decided to continue using the tool and become paying subscribers and 41 (40%) opted not to continue.

    Participants who chose not to adopt fell into one of four categories:

    • they did not have a significant documentation burden,
    • their existing EHR workflow worked well for them,
    • their EHR did not yet integrate with the solution, or
    • they were too challenged to even fully trial the AI Assistant.

    The 60% of lab participants who adopted the AI Assistant saw a 72% reduction in median documentation time per note, for an overall calculated savings of 3.3 hours per week per clinician. Among those who chose not to adopt the tool, the median decrease in documentation time per note was 50%.

    Participants’ satisfaction was measured for each domain using a 10-point Likert scale, with 10 being most satisfied. In the adoption group, before and after values were as follows:

    • Overall practice satisfaction: mean value rose from 6.4 to 7.7;
    • Workload satisfaction: mean value rose from 5.3 to 6.9; and
    • EHR satisfaction: mean value rose from 6.4 to 7.4.

    Finally, participants were surveyed about their level of burnout before and after using the innovation according to a five-item scale that ranged from “I enjoy my work. I have no symptoms of burnout” to “I feel completely burned out. I may need to seek help.” Respondents overall ranked their burnout/stress levels across all but the most severe category. Although the majority of responses landed in the two middle ranks both before and after the lab, the “after” responses skewed lower.

    “We’re pleased to report that an AI Assistant for Documentation significantly reduced documentation time and burden while providing more flexibility and freedom,” Waldren noted. “Clinicians reported their notes were more meaningful and professional.

    “Our conclusion is that an AI Assistant for Documentation is an essential innovation for all family physicians who have documentation burden and experience burnout. It can help optimize their family medicine experience.” 

    The lab now enters phase three, in which the AAFP aims to build awareness of this essential AI Assistant among members and offer them the opportunity to try it themselves for free. Special AAFP pricing is available after the free trial for those who choose to subscribe.