This week’s AI in Healthcare update covers four key developments: an NEJM AI randomized trial finding AI literacy training did not prevent automation bias from intentionally erroneous LLM diagnostic suggestions; FDA clearance and a CMS reimbursement pathway for Bunkerhill’s AI tools quantifying coronary and aortic valve calcium on routine contrast, non-gated chest CT (with limited independent peer-reviewed validation noted); a 48-trial Bayesian network meta-analysis (34,106 participants) reporting five AI colonoscopy systems improved adenoma detection rate but not advanced adenomas or sessile serrated lesions; and a Nature Medicine perspective urging higher evidentiary standards and patient-centered outcomes before claiming AI improves healthcare.

00:00 Weekly AI Healthcare Briefing

00:48 AI Literacy vs Automation Bias

02:22 FDA Cleared Calcium Detection

03:55 AI Colonoscopy Meta Analysis

05:54 Raising the Evidence Bar

07:23 Wrap Up and Next Week

In this week's episode of AI in Healthcare, your concise update for healthcare professionals on artificial intelligence in clinical medicine, we examine four developments from the week of April 20–27, 2026.

A New England Journal of Medicine Perspective on Utah's AI-assisted prescription-renewal sandbox pilot — a state-regulated program with pharmacist-mediated escalation, distinct from the FDA pathway for software as a medical device — and the corresponding American Hospital Association governance panel on April 20.

A multicenter Korean validation study in JMIR Medical Informatics introducing patient-wise recalibration to mitigate model drift in AI electrocardiography for left ventricular systolic dysfunction (reported AUC 0.956 internal, 0.940 external on follow-up pairs).

A randomized controlled trial in JMIR Mental Health in which both a structured generative AI therapy chatbot and plain GPT-4o produced significant PHQ-9 reductions versus control, with no significant difference between active arms (n = 147).

A methodological comparison in JMIR in which XGBoost (micro-F1 0.815) outperformed a LoRA-fine-tuned LLaMA-3 (0.780) on ASA Physical Status classification.

Evidence-based, reference-linked, ~5 minutes. For healthcare professionals only.

00:00 Weekly Headlines

00:31 Utah Prescribing Sandbox

02:09 Governance Takeaways

02:38 Drift Mitigation Study

04:06 GenAI Depression Trial

05:35 LLM vs XGBoost Methods

06:48 Wrap Up and References

REFERENCES

1. Utah Prescription-Renewal Pilot. NEJM Perspective, April 2026. DOI: 10.1056/NEJMp2601148
2. Utah Department of Commerce / Doctronic announcement, January 2026: commerce.utah.gov
3. AHA Panel — AI in Health Care: Navigating Policy, Regulation, and the Road Ahead. April 20, 2026: aha.org
4. Lee S, Son J-W, Kim S-A, et al. Deep Learning Model Using Transfer Learning for Detecting Left Ventricular Systolic Dysfunction. JMIR Med Inform. April 24, 2026. DOI: 10.2196/83127
5. Kuta B, Novak L, Zidkova R, et al. Effectiveness of a Fully Automated Mobile Therapeutic Versus a General Chatbot in Reducing Depression and Anxiety. JMIR Ment Health. April 22, 2026. DOI: 10.2196/82642
6. Chen M-C, Ruan S-J, Wu J-H, Chen P-F. Classifying ASA Physical Status With a Low-Rank-Adapted Large Language Model. J Med Internet Res. April 21, 2026. DOI: 10.2196/89540

Disclaimer: For healthcare professionals only. Not medical advice. Opinions expressed do not represent any institution.

#AIinHealthcare #ClinicalAI #DigitalHealth #FDA #AIRegulation #AIECG #GenerativeAI #LLM #NEJM #JMIR

In this episode of the AI in Healthcare podcast, we explore new research on using gradient-boosted models to predict coronary artery disease.

00:00 Introduction to AI in Healthcare Podcast

00:09 Gradient-Boosted Models in Coronary Artery Disease Prediction

00:47 Improving Referral Pathways with Transparent Models

01:18 Pragmatic Steps for Clinical Implementation

01:40 Key Takeaways and Recommendations

01:55 Conclusion

Williams MC, Guimaraes ARM, Jiang M, Kwieciński J, Weir-McCall JR, Adamson PD, et al. Machine learning to predict high-risk coronary artery disease on CT in the SCOT-HEART trial. Open Heart. 2025 Sep 1;12(2):e003162. doi:10.1136/openhrt-2025-003162. PMCID: PMC12406813. PMID: 40889953.

In this episode of the AI in Healthcare podcast, we break down a new study by Jeong and colleagues that explores the impact of AI assistance on radiologist reading times and workflow efficiency.

One‑line takeaway: AI assistance reduced reading time and improved throughput for bone‑age radiograph interpretation in a real‑world retrospective cohort.

Reference

The Impact of Artificial Intelligence on Radiologists’ Reading Time in Bone Age Radiograph Assessment Citation: Jeong S, Han K, Kang Y, et al. Journal of Imaging Informatics in Medicine. 2025 Aug;38(4):1915‑1923.

https://pubmed.ncbi.nlm.nih.gov/39528879/

Join us on the AI in Healthcare podcast as we discuss into the groundbreaking development of VentAI, a reinforcement learning algorithm designed to recommend optimal ventilator settings for ICU patients.

00:00 Introduction to AI in Healthcare

00:06 VentAI: Revolutionizing Ventilator Settings

00:20 Study and Validation of VentAI

00:54 Performance and Comparison with Clinician Care

01:37 The Role of AI in Clinical Decision Making

01:56 Future of AI in Intensive Care

02:02 Conclusion and Follow Us