Healthcare AI Daily Briefing

Operations & Workflow

EHR‑derived ML AUROC 0.709 matches clinician VTE risk scores.

Preprint

In 577,904 admissions across five NHS hospitals, machine‑learning models using EHR‑derived VTE risk factors available within 14 hours achieved an AUROC of 0.709, compared with 0.704 for clinician‑recorded factors. The analysis assessed completion, concordance with EHR data, and clinical validity of the risk assessment forms. [1]

Research & Evidence

AI‑guided wearable reduced instability but not HF rehospitalization in LINK‑HF2.

Journal article

In a prospective randomized trial of 171 heart‑failure patients at five Veterans Affairs medical centers, participants wore a continuous multisensor patch for up to 90 days and received AI‑guided wearable monitoring or usual care; the intervention lowered AI‑detected physiological instability but did not reduce HF rehospitalization. [2]

Operations & Workflow

Baptist Health early adopter of assistive coding, 20% faster billing.

Podcast from Lifers (Second Opinion)

The interview describes that Baptist Health leveraged automated assistive coding to reduce the meantime of submitting bills and obtaining prior auth by 20%, which the interviewee said amounts to almost a one‑million‑dollar savings. The head of coding said she was reluctant at first but later saw its value in filling staffing gaps. [3]

Industry & Products

Artera’s AI‑native platform self‑hosts LLMs and claims 40% effectiveness gain.

Artera’s AI‑native platform self‑hosts six LLMs and, in a Federally Qualified Health Center with a 200‑person call center, identifies and closes patient care gaps during the same call, boosting provider quality measures and yielding a claimed 40 % increase in effectiveness. [4]

Industry & Products

AI chatbot summaries aided physicians in three‑quarters of visits.

In a study at BIDMC’s ambulatory primary care clinic, 98 patients consulted the AMIE AI chatbot before urgent‑care visits. Clinicians reported AI summaries helped prepare visits in 75% of cases and influenced care in more than half, while AMIE’s diagnoses matched doctors’ final diagnoses 90% of the time. Findings suggest AI could enhance patient‑physician relationships pending larger trials. [5]