Healthcare AI Daily Briefing

Operations & Workflow

Sponsored: A vendor advocates predictive billing tools.

Sponsored content: In sponsored content on Healthcare Dive, the author argues that integrating clinical and financial data could let predictive revenue-cycle tools flag eligibility issues and suggest codes before submission. The article also advocates human-in-the-loop automation for appeals; this is a vendor argument, not an independently measured outcome. [4]

Clinical Practice

AI diagnostic support for older adults shows high discrimination across 707 598 primary care encounters.

Journal article

A deep learning diagnostic decision support system trained on electronic health record data from adults aged 65 and older demonstrated excellent diagnostic discrimination with a micro c-statistic of 0.995 on a temporally held-out test set. The model used imitation learning and collective clinician intelligence to generate recommendations across 669 diagnoses without expert-adjudicated labels, based on 707 598 primary care encounters. [2]

Medical Imaging

Coronary CT AI showed 79.5% sensitivity against expert consensus.

Journal article

In a retrospective study of 1,193 emergency-department patients undergoing coronary CT angiography, AI was assessed against two reference standards. Among the 885 assessed against expert consensus, AI achieved 79.5% sensitivity and 95.4% negative predictive value for obstructive CAD; the other 308 were evaluated against invasive angiography. [1]

Operations & Workflow

PwC attributes 9% insurer medical cost increase to AI-backed billing tools.

Analysts at PwC attribute a 9% increase in insurers' medical costs next year to AI-backed billing tools, while Blue Cross Blue Shield Association reported about $942 million in extra spending over two years from similar tools, noting more diagnoses without corresponding treatment. Vendors acknowledge that more complete coding can raise payments, while disputing that legitimate documentation constitutes upcoding. [3]