Clinical Adoption of AI

The New England Journal of Medicine (NEJM) is planning to publish a new journal in 2024, NEJM AI, to identify and evaluate state-of-the-art applications of AI to clinical medicine. One of its early-release articles analyzed the usage and adoption patterns of over 500 FDA-approved medical AI devices in the United States. (FDA listed 692 AI-enabled medical devices on its website.) By tracking insurance claims using Current Procedural Terminology (CPT) codes specific to medical AI, it revealed that adoption is still in its early stages, concentrated in a few leading devices.

The study, examining 11 billion CPT claims from January 2018 to June 2023, finds significant usage only in AI devices for coronary artery disease and diabetic retinopathy diagnostics. It also highlights a disparity in AI device usage, being more prevalent in metropolitan, higher-income areas with academic medical centers. The findings emphasize the need for further exploration into the barriers and incentives for broader, equitable integration of AI in healthcare.

Comment 1: It should be noted that any observational study has selection biases, for examples, the study only examined 32 CPT codes; not all AI devices have CPT codes; not all AI devices were covered through CPT codes by payer (CMS uses NTAP for many devices; many don’t even go through insurance); many AI devices share the same CPT codes as traditional devices; IQVIA PharMetricsVR Plus claims dataset only covers commercially insured patients; etc.

Comment 2: The journey of an AI algorithm from development to widespread clinical adoption in healthcare is multifaceted and complex, involving several critical steps: development of the AI algorithm, regulatory approval, clinical adoption, payer reimbursement, and ongoing evaluation and improvement. It’s important to consider these complexities when developing AI technologies for healthcare.

US Congress urges CMS on AI-based claims denials in Medicare Advantage

In a related news, U.S. House Democrats have raised concerns with CMS about the use of AI in Medicare Advantage plans, particularly focusing on the high rates of claims denials. They suggest that AI might be causing more restrictive coverage than traditional Medicare. The lawmakers propose several oversight measures, including requiring MA plans to report prior authorization details and comparing AI-generated guidelines with actual decisions. They also call for analyzing data to identify denial trends and ensuring coverage guidelines are not overly restrictive. These concerns align with broader healthcare industry discussions on the use of AI, including the recent executive order to establish AI standards in healthcare.

AI could lead to a better understanding of the brain and related diseases

Shifting the gear to basic and exploratory research, we’d like to highlight a few articles from Nature.

• How AI could lead to a better understanding of the brain. AI (or artificial neural networks) was inspired by the understanding of how the brain works. It's intriguing to see how AI can, in turn, help decipher the complexities of the brain.

• AI that reads brain scans shows promise for finding Alzheimer’s genes. The article shared a few promising AI models to identify genomic or imaging biomarkers for Alzheimer’s disease.

• Contrastive Language-Image Pretraining (CLIP) model better predicts brain responses to real-world images. It demonstrates the surprising convergence of AI’s language encoding, AI’s image encoding, and the actual human brain’s voxel response and image representation.

• Functional genomics and systems biology in human neuroscience. The authors shared their perspectives on how systems-level network analyses can help advance neuroscience. (We will have a newsletter to focus on using AI to infer gene networks).

##########

If you find the newsletter helpful, please consider:

• 🔊 Sharing the newsletter with other people

• 👍 Upvoting on Product Hunt

• 📧 Sending any feedback, suggestions, and questions by directly replying to this email or writing reviews on Product Hunt

• 🙏 Supporting us with a cup of coffee.

Thanks, and see you next time!