Contemporary physicians act not only as caregivers but also as administrators. A significant portion of their workday is dedicated to patient consultations, with the rest spent on creating and entering clinical notes from these appointments. This exacts a cost. Over 45% of doctors experience burnout, according to the American Medical Association—with the administrative duties, frequently extending into their personal time, contributing to this issue.
However, a potential remedy has emerged: A study released in JAMA Network Open indicates that artificial intelligence systems, specifically ambient documentation technology, which log patient interactions and prepare notes for medical practitioners, can lower burnout rates by almost 31%.
“This remarkable decrease in burnout is truly restoring satisfaction to the medical profession,” states Dr. Rebecca Mishuris, a primary care physician, chief medical information officer at Mass General Brigham, and a co-senior author of the research. “I aim to ensure that everyone has an equitable opportunity to gain from this advantage and to collaborate with and assist individuals in maximizing the technology’s utility.”
For their investigation, the researchers enlisted 873 doctors from Mass General Brigham and 557 from Emory Healthcare in Atlanta. These medical professionals represented diverse specialties, such as surgery, urgent or emergency care, pediatrics, infectious disease, among others. Their professional experience varied from only one year to over two decades. The physicians completed surveys assessing burnout and overall well-being at several intervals over approximately three months.
The majority of physicians did not continue participation throughout the study period; by its conclusion, the response rate stood at only 22% for doctors at Mass General Brigham and 11% for those at Emory. Nevertheless, the findings were promising for the subset of medical practitioners who did provide feedback. For the Mass General Brigham physicians, the AI assistant—recording patient interactions unobtrusively—was linked to a 21% decrease in burnout, whereas at Emory, it resulted in a 30% improvement in well-being.
“I believe obtaining high response rates is consistently challenging, especially when dealing with clinicians who have demanding schedules,” Mishuris comments. “From our perspective, part of the indication of its effectiveness is that people continue to utilize the system currently, and its adoption has rapidly expanded.”
A number of physicians expressed enthusiasm for ambient documentation. “It undeniably enhances my professional satisfaction because I can engage with patients and maintain eye contact without concern about forgetting their statements later,” an infectious disease specialist noted in a survey reply. “As these tools advance, I anticipate they will profoundly alter the reality of being a doctor.”
“Extremely beneficial,” a neurologist remarked. “It certainly enhances my interaction with patients and their families and unequivocally simplifies clinic operations.”
However, dissenters were also present. “I attempted using it but discovered it appended 1 to 2 hours daily to my note-taking process,” a pulmonologist stated. “I am not yet prepared to delegate my documentation tasks to AI,” a primary care and internal medicine specialist concluded.
Mishuris acknowledges these critiques. “Evidently, this technology will not prove advantageous for every individual,” she remarks. “Each clinician possesses a somewhat distinct workflow and a unique method for their documentation. Certain individuals are prepared to entrust tasks to AI—to relinquish that perception of absolute control, recognizing that they retain full authority over the ultimate output, though not over the initial draft.”
Further investigation is still necessary. In their publication, the authors concede that the comparatively low participation rate could imply that primarily the most ardent users completed the follow-up surveys, thus potentially biasing the results positively, while those who derived no advantage from the technology did not respond. Furthermore, not all medical disciplines are well-suited for an AI assistant. A pediatrician noted that most patient consultations in that area involve physical examinations, which AI cannot record. A hospice and palliative care physician voiced dissatisfaction, stating the system performed inadequately concerning psychosocial and spiritual well-being matters.
A year has passed since the initial findings were reported, and Mishuris and her collaborators are currently engaged in a subsequent study that expands upon their preliminary research.
The researchers are convinced that artificial intelligence can address a significant void within the medical field. “The ongoing utilization of the system by individuals…indicates that these outcomes are probable to remain consistent across a considerably larger demographic,” states Mishuris.