Ambient AI speeds chart closing up to 39% at Jefferson Health, study finds
Oct 25, 2026
Listen to this post
read by AI voices · all episodesTHE SHORT READ
Clinicians at Jefferson Health who adopted Abridge's ambient AI cut their time to close charts by 25.3% to 39.3%, according to a study published Oct. 23 in NEJM Catalyst Innovations in Care Delivery.
The study covered 5,234 clinicians with access to the tool, about 34% of whom actively recorded visits, and 283,028 ambulatory encounters in the second half of 2025, Becker's Hospital Review reported. The authors described the results as associations and noted that adopters may have differed from clinicians who did not use the tool. Abridge had no role in the study and provided no funding, according to the report.
Elsewhere, Inova Health System said it had drawn up plans to switch off three AI features whose performance declined, a Lancet Digital Health experiment found that radiologists and other physicians both followed wrong AI advice and ignored correct advice on head CT scans, and a JAMA Psychiatry study found that adding imaging, genetics and microbiome data did not improve a model's prediction of depression relapse.
THE DETAILS
- FYIDeployment Oct 23, 2026The study, led by Dr. Jennifer Stephens, compared Epic usage data for the 12 weeks before and after adoption among 637 clinicians who met usage thresholds, in two cohorts for Jefferson and its Lehigh Valley Health Network. Time in notes per appointment fell 10.5% and 17.3%, after-hours documentation fell 8.9% and 16.6%, and weekly note volume held steady or rose, according to Becker's Hospital Review. Advanced practice providers gained 9.9 minutes of documentation time a day, compared with 3.8 minutes for physicians. Patient experience scores were equal or higher for ambient visits, but the advantage largely disappeared when each clinician's own ambient and nonambient visits were compared. Jefferson's earlier figure of about 1 million hours returned to patient care, reported in September, comes from executives and is not part of the study, Becker's reported.Source: Becker's Hospital Review
- FYIDeployment Oct 23, 2026Inova Health System, based in Falls Church, Virginia, now reviews more than 12 AI features a month for approval, up from about one a month two years ago, and is preparing governance to oversee as many as 500, according to Jon McManus, its chief data and AI officer. Its latest quarterly safety and reliability assessment flagged three features with declining performance, and the system developed decommissioning plans for all three, Becker's Hospital Review reported. Inova favors Epic's agent platform for projects that change the health record and may use OpenAI, Anthropic or other partners for others. "They are purposely turning AI off if it is not meeting performance standards," McManus said. The report did not name the three features.Source: Becker's Hospital Review
- FYIEvidence Oct 23, 2026In a randomized crossover experiment, 50 radiologists and 75 non-radiologist physicians and trainees each read 50 head CT scans for intracranial hemorrhage with no AI advice, with an AI prediction alone, or with an AI prediction plus scan annotations; 20% of the advice was deliberately wrong. Both over-reliance on wrong advice and under-reliance on correct advice occurred, and under-reliance did more harm to accuracy in absolute terms, the authors reported. Correct annotated advice improved diagnostic performance compared with no advice (odds ratio 1.84) but did not shorten reading time. The study, led by Susanne Gaube of University College London with authors including Marzyeh Ghassemi and Errol Colak, was published Oct. 23.Sources: The Lancet Digital Health, PubMed
- FYIEvidence Oct 21, 2026German researchers followed 564 patients with major depressive disorder for about two years, during which 270 relapsed, and built models from nine kinds of baseline data, including four MRI modalities, genetics, blood markers and the gut microbiome. Balanced accuracy ranged from 47.21% to 64.43%, and the best model used only a few clinical risk factors, the authors reported. They wrote that the results suggest "limited scope for guiding individual-level clinical decision-making." The study, from the Marburg-Münster Affective Disorders Cohort, was published Oct. 21.Sources: JAMA Psychiatry, PubMed
- FYIWorkforce Oct 23, 2026HIMSS Market Insights surveyed 525 clinical leaders with authority over AI purchases in the United States, Europe, the Middle East, Africa and Asia-Pacific in August, Healthcare IT News reported Oct. 23. Ninety-one percent said they were highly motivated to pursue specialized AI training, and 83% agreed AI will open new career pathways for clinicians. Thirty-two percent agreed completely that their organization has clear processes for evaluating and monitoring AI. In North America, 68% agreed their organization has a clear strategy for responsible clinical AI, compared with 84% in Europe, the Middle East and Africa and 85% in Asia-Pacific. The figures are the leaders' own views, not measured outcomes.Source: Healthcare IT News
- FYIMoney Oct 22, 2026Siemens Healthineers and Nvidia said Oct. 22 that they will embed AI agents in radiation treatment planning and use digital twins of patients to model treatment scenarios, with added computing power during delivery to adapt therapy to each patient, MassDevice reported. The work will use Nvidia's Nemotron open models and its Holoscan platform for real-time sensor data. Arthur Kaindl, head of the Varian cancer-care business, tied the effort to the single-breath radiotherapy system the company unveiled in September. The companies did not give a timeline or clinical data.Source: MassDevice