The medical technology sector witnessed significant advancements in artificial intelligence (AI) during 2024. Major companies like GE Healthcare, Medtronic, and Dexcom introduced innovative AI features, while others expanded their AI capabilities through mergers and acquisitions. The Food and Drug Administration (FDA) provided new guidelines for AI-enabled devices, but uncertainties remain regarding generative AI and continuous adaptation tools. Despite the hype, challenges such as lack of insurance reimbursement and regulatory ambiguity continue to hinder widespread adoption. Experts predict that 2025 will see more clarity on AI regulations, increased focus on foundational models, and better evaluation methods for hospitals.
Regulatory Clarity and Future Directions
New guidance from the FDA aims to provide greater transparency for developers of AI-powered medical devices. Recent updates include a finalized framework for pre-specified modifications after market release and draft guidelines on submission requirements. These changes encourage companies to adopt predetermined change control plans (PCCPs), which can streamline post-market adjustments. However, the Trump administration's stance on AI regulation remains uncertain, potentially impacting future policies.
The FDA’s recent actions, including finalizing PCCPs and issuing draft guidelines, signal a move towards clearer pathways for AI device submissions. Attorneys anticipate that developers will increasingly utilize these frameworks to enhance efficiency. While the Biden-era executive order on AI was rescinded, HHS has released a strategic plan for healthcare AI oversight. Under the Trump administration, experts expect continued focus on AI and machine learning, alongside growing concerns about state and national privacy laws affecting AI adoption.
Payment Challenges and Emerging Focus Areas
Despite the potential of AI to revolutionize healthcare, payment challenges persist. Insurance coverage for AI features remains limited, with Medicare lacking specific reimbursement mechanisms. Companies are exploring workarounds, such as leveraging New Technology Add-on Payments. Hospitals are becoming more cautious, scrutinizing AI tools for economic return on investment. This shift reflects a growing demand for tangible benefits over initial excitement.
As AI applications expand beyond radiology into pathology, ophthalmology, and cardiology, there is increasing interest in foundational models and administrative tasks. Large language models are being used for clinical note generation, though their regulatory status remains unclear. Vision language models, which analyze images and draft reports, are under development but not yet authorized by the FDA. Hospitals need robust evaluation processes to ensure AI tools perform well on local data and address potential biases. Initiatives like model cards and assurance labs aim to provide upfront information and objective evaluations, enhancing informed procurement decisions. Monitoring AI performance over time is crucial, requiring strong partnerships between vendors and healthcare systems.