🧠 Clinical Adoption of Medical AI Remains Modest, Early Trends Show Promise

A new NEJM‑AI analysis explores how widely AI tools are being used in real-world U.S. clinical settings, using insurance claim data to provide actionable insights for healthcare professionals.

🔍 Key Findings

Slow adoption despite approvals: Over 1,000 AI-enabled medical devices have received FDA clearance, yet usage remains limited. Only 16 procedures corresponding to AI-specific CPT codes were billed in 2023.
Concentration of use: Most utilization centers on a few high-impact AI tools, especially in coronary artery disease diagnostics, diabetic retinopathy, and radiology.
An implementation gap persists: Despite explosive research output and regulatory approval, there’s a major disparity between AI innovation and real-world clinical use—sometimes called the “AI chasm.”

🛠️ Barriers to Broader Clinical Use

Billing & reimbursement hurdles: With only 16 AI-specific CPT codes issued, few systems are financially incentivized to integrate AI workflows.
Technical and operational friction: Cost of deployment, IT interoperability, and workflow integration are persistent obstacles in everyday care.
Evidence gaps: Many AI systems lack robust prospective, randomized validation — rather, approvals often rely on retrospective or limited datasets.

📈 Increasing Uptake in Key Domains

AI tools are most used in:
- Coronary artery disease (e.g., FFR-CT analysis)
- Diabetic retinopathy screening
- Radiology workflows like lung nodule detection and stroke triage
These represent early wins where AI targets clear diagnostic bottlenecks.

🧩 Bridging the AI Chasm: What’s Next?

Dynamic AI deployment: Transition from static tools to adaptive systems with continuous learning and real-time outcome tracking.
Incentivize billing integration: Expand CPT codes and structured reimbursement to support AI-enhanced care.
Invest in validation: Prioritize prospective trials and real-world evidence post-approval, ensuring clinical rigor.
Streamline systems integration: Simplify technical implementation through vendor partnerships and EHR interoperability.

💡 Clinical Takeaways

Insight	Impact on Practice
Few tools in use	AI potential is high—but only a small selection is routinely used.
Billing drives adoption	Financial incentives accelerate uptake—more CPT codes could broaden access.
Evidence still needed	Clinicians should look for AI tools with strong prospective validation.
Adaptability is key	AI that evolves with data and use-cases will be more sustainable and trustworthy.

Bottom line:

FDA clearance is just the start. To move from promise to practice, AI tools must be validated prospectively, integrated seamlessly into workflow, and supported by reimbursement models. Clinicians should track AI that delivers tangible impact in diagnostics and outcomes—and advocate for dynamic, continually learning systems that bridge today’s implementation gap.

📖 To read the full article in NEJM AI, click here:

Characterizing the Clinical Adoption of Medical AI Devices Through U.S. Insurance Claims

NEJM AI, June 2024