Healthcare AI Daily Briefing

Medical Imaging

Aidoc CTPA pulmonary embolism model sensitivity 86.8% outperforms incidental model 73.5%.

Preprint

In a retrospective study across a 17‑facility academic health system evaluating 30,678 CTPA and 37,191 routine CTs, Aidoc’s FDA‑cleared AI models for pulmonary embolism detection showed sensitivity below FDA‑cleared benchmarks while maintaining specificity above cleared thresholds, with lower performance for peripheral and non‑acute emboli. [1]

Research & Evidence

Seven‑variable random forest predicts one‑year anemia after bariatric surgery (AUC 0.823).

Journal article

In a multicenter study of 511 derivation and 300 external validation patients, a seven‑variable random‑forest model for predicting one‑year anemia after bariatric surgery outperformed junior, middle‑level and senior clinicians. [2]

Operations & Workflow

Inova Health spends about 10% of time taking, 80% shaping, 10% building market‑differentiating capabilities.

Podcast from Healthcare AI Pioneers

The interview describes Inova Health’s workflow: about ten percent of the time is spent taking FDA‑approved imaging algorithms as‑is, about eighty percent shaping them through low‑code UI configuration to fit local needs, and about ten percent building market‑differentiating capabilities, with the ‘make’ portfolio starting to grow. [3]

Policy & Governance

Interview: Joint Commission’s June AI governance could shape health AI talks.

Podcast from KFF's The Business of Health with Chip Kahn

The interview describes that taking existing scaffolding—including the Joint Commission’s June AI governance for health—could help lead health AI discussions, building on earlier talks of a pause before health‑specific applications. [4]

Clinical Practice

Physician warns AI efficiency focus may raise workload but could restore time for patient‑centered care.

Dr. David Kirk, CMO at Regard, warns that AI’s efficiency focus can add regulatory burden and increase clinician workload, but argues that with team‑based oversight scaling autonomy to risk, AI could restore time for listening, understanding, reassuring, deciding, and caring for patients. [5]