By Gianna Melillo
Researchers laid out the latest findings in ambient documentation by artificial intelligence (AI) and the effects of weight loss drugs, as well as ACP's new breast cancer screening guidance, during a plenary session Friday morning at Internal Medicine Meeting 2026.
Christine Laine, MD, MPH, FACP, moderated the event, which featured Ashok Reddy, MD, MSc, ACP Member; Carolyn Crandall, MD, MS, MACP; and John Batsis, MD, FACP.
Dr. Reddy, a primary care physician and health services researcher at Veterans Affairs Puget Sound in Seattle, shared results of his research comparing the quality of AI-generated and human-produced clinical notes. The study was published by Annals of Internal Medicine on April 17.
“We all know the problem,” Dr. Reddy told attendees. “Every one of us has probably stayed late at clinic, or more likely, headed home” to finish writing clinical notes. While ambient AI promises to help reduce this workload, previous research into its value has only focused on a single vendor of the technology at a single institution and offered little insight into the quality of notes produced, he noted.
To address this gap, Dr. Reddy and colleagues created five recordings of primary care cases using standardized patients and a real clinician in a simulated clinical environment. Some of these cases incorporated real-world acoustic challenges, such as background noise or non-native English accents.
Investigators then tasked 11 AI vendors and three clinicians with creating notes for each case, for a total of 14 notes per case. Thirty clinicians, blinded to which notes were by humans or AI, reviewed them all and assessed the quality of each.
“Across all five clinical cases, human-generated notes received higher overall quality scores compared to AI generated,” Dr. Reddy said. The largest difference in quality ratings was seen in a case that had significant background noise and a masked patient and physician. “For this case, the note produced by AI was essentially half as good as the one produced by humans,” he said.
What’s more, researchers found the AI-generated notes rated lower in all 10 domains of quality measured. The findings led Dr. Reddy to conclude that additional quality evaluation in different clinical environments is needed to better understand the tools’ benefits and limitations.
“I do think that the measure of these AI scribe evaluations should be done at every point pre-implementation, during the implementation, and after,” he said.
Next, Dr. Crandall, an internal medicine physician and current Chair of ACP's Clinical Guidelines Committee, walked attendees through the College's new guidance statement on screening for breast cancer in asymptomatic, average-risk adult females. The statement was also published by Annals on April 17.
“It's critical to discuss breast cancer screening in the U.S., [as] there were more than 316,000 new breast cancer cases in the year 2025 alone,” said Dr. Crandall. The new guidance is designed to reconcile several conflicting clinical guidelines. It incorporated recommendations from the Canadian Task Force on Preventive Health Care, the European Commission Initiative on Breast Cancer, and the United States Preventive Services Task Force.
The committee defined asymptomatic, average-risk females as those with no “personal history of breast cancer or high-risk breast lesion, no known genetic mutation like BRCA, and no childhood chest radiation exposure,” Dr. Crandall explained.
A total of five guidance statements were offered based on patients’ age and breast density. “In females 40 to 49 years, use shared decision making,” Dr. Crandall said. “That's always going to include values and preferences of that individual.”
A key discussion point for this demographic is that breast cancer mortality is marginally reduced with mammography in 45- to 49-year-old females, but they face an increased risk of overdiagnosis, false negatives, additional testing, false positives, and related psychological distress. If discussion with the patient results in a decision to screen, physicians should initiate biennial screening.
For females age 50 to 74 years, the guidance calls for initiating biennial mammography because it decreases breast cancer mortality, Dr. Crandall explained. In those who are 75 years and older or have limited life expectancy, shared decision making should again be used. The discussion should include the fact that available data show breast cancer mortality is not reduced by screening for this population.
Dr. Crandall also noted that recent highly rated guidelines haven't specifically reviewed whether breast cancer mortality is different with annual or biennial screening. “We really need more data regarding best frequency, especially for females with dense breasts, and for the two ends of [the] age spectrum, 40 to 49 and beyond 74 years.”
Finally, Dr. Batsis, a geriatrician and associate professor in the schools of medicine and public health at the University of North Carolina at Chapel Hill, discussed results of a study comparing the effects of incretin-based therapies versus lifestyle modifications or placebo on body composition.
Previous research on weight loss with therapies like semaglutide and tirzepatide has shown that “even when the relative composition appears to be favorable, there is still meaningful absolute decline in lean mass,” Dr. Batsis explained. “This raises an important clinical question: How do we balance that [desired weight] loss with potential muscle loss?” His systematic review of 36 randomized clinical trials appeared in Annals on April 17.
The results showed that nearly 70% of incretin-treated study arms exceeded the expected proportion of muscle loss, Dr. Batsis told attendees. “The key point here is that it's not simply that muscle-based parameters are lost, but that the proportion of weight loss attributable to muscle was higher than anticipated,” he continued. Similar patterns were observed in about half of the comparator arms achieving weight loss.
None of the included studies assessed measures of physical function, he cautioned, and future trials should incorporate this assessment as well as other patient outcomes.
“We need to know beyond weight loss, beyond morbidity, what's happening and then the functional consequences of these changes,” he said.
Physicians should be regularly monitoring patients taking these therapies in clinical practice. “You don't want to have a patient who was playing golf and the next thing you know, six months later, has lost so much weight, is using a walker," Dr. Batsis said. ■