Buzzworthy developments of the past few days.

• Brain specialists have trained a large-language AI model on the complete EHR of a sprawling healthcare system. The model ingests physician notes to predict rehospitalization, gauge infection risk and foresee other key clinical outcomes. It also has potential to help with operational tasks like billing and coding. The system was developed by neurosurgeons at New York University who state the achievement represents a solid step toward the “long-standing dream for physicians to have AI assistants observing care along with them and chiming in with predictions and advice.” Journal study here, NYU Langone news item here.
   
• Now comes an AI model that interprets electrocardiograms as language. Developed by researchers at Icahn School of Medicine at Mount Sinai, the system, called HeartBEiT, renders ECG readouts such that parts come out as individual words while the exam as a whole is a single document. Lead researcher Akhil Vaid, MD, says the model “understands the relationships between these representations and uses this understanding to perform downstream diagnostic tasks more effectively.” Peer-reviewed study. Mount Sinai news item.
   
• Royal Health of White Plains, N.Y., has established a subsidiary unit focused on radiological AI. The new unit, RoyalAI, is offering algorithms aimed at finding operational efficiencies. The product integrates with end-users’ existing workflows, broadening AI access with a simple software installation. Announcement.
   
• An Ivy League institution is opening a new center to advance precision medicine and healthcare AI. Dartmouth College in Hanover, N.H., says the facility will house medical researchers working on AI-powered diagnostics, therapeutics and public health policy. The center will reside on the campus of Dartmouth Hitchcock Medical Center in nearby Lebanon, N.H.
   
• Clarify Health of San Francisco lets “Clara” out to play. A generative AI aid for turning raw patient data into actionable clinical insights, Clara will take first steps that will be witnessed only by invited provider orgs in a closed beta launch. When all grown up and ready to roam, Clara will “enable organizations to uncover opportunities to influence and incentivize higher-value care decisions that directly result in better care and cost for patients.” Announcement.
   
• Got a smartphone? Check your blood oxygen level. Medical researchers in the Midwest have come up with an AI-driven smartphone app that lets users do just that. The team, based at Purdue University in West Lafayette, Ind., says the innovation can detect and diagnose conditions like anemia faster and more accurately than medical equipment dedicated to the task. Details.
   
• AI startup BigBear.ai (Columbia, Md.) is piloting a product for smart surgery scheduling. The AI offering, set for adoption at Children’s Hospital Colorado, helps manage suite utilization by computing OR capacity, resource availability, surgeon-specific metrics and surgical procedure trends. Announcement here.
   
• Reminder: ChatGPT wasn’t developed with medical research and practice in mind. And yet it has performed “surprisingly well in providing basic-level support in research and clinical settings,” researchers comment in a study published by the Journal of Multidisciplinary Healthcare. “This is a testament to its potential to revolutionize medicine and healthcare if further technological advancement of this model is made in conjunction with the medical field.”