Authorization of prognostic AI medical devices

2026-02-06 · Source: Nature Machine Intelligence · Field: Health & Wellbeing — Medical Devices & Health Technology, Healthcare Systems & Policy · Depth: Intermediate, quick

Summary

Less than 2% of artificial intelligence (AI) medical devices authorized by the US Food and Drug Agency (FDA) are prognostic, meaning they predict future health outcomes. These prognostic AI devices exhibit prediction horizons ranging from minutes to several years. As the prevalence of prognostic AI devices is expected to grow, the article emphasizes the critical need to address the associated regulatory and ethical challenges. The content references figures illustrating authorized prognostic AI/ML-enabled medical devices by specialty and year, and their respective prediction horizons, highlighting the current landscape and future implications for this emerging category of medical technology.

Key takeaway

For regulatory bodies and healthcare technology developers, the low authorization rate of prognostic AI medical devices by the FDA signals an underdeveloped regulatory framework. You should prioritize developing clear guidelines for the authorization, validation, and post-market surveillance of these devices to ensure patient safety and ethical deployment as their numbers increase.

Key insights

Prognostic AI medical devices, currently under 2% of FDA authorizations, present unique regulatory and ethical challenges.

Principles

• Prognostic AI devices predict future health outcomes.
• Prediction horizons vary from minutes to years.

Topics

• Prognostic AI
• Medical Devices
• FDA Regulation
• AI Ethics
• Machine Learning

Best for: CTO, VP of Engineering/Data, Director of AI/ML, AI Ethicist, Policy Maker, AI Product Manager

Related on AIssential

• AI-Powered Tools in the Operating Room Is Probably a Bad Idea — Integrating AI into medical devices can introduce significant reliability and safety challenges, as evidenced by increased malfunction reports.
• Dozens of AI disease-prediction models were trained on dubious data — AI models for disease prediction are being trained on potentially fabricated public datasets, risking patient harm.
• Something’s off with Midjourney’s pivot to body scanners — Midjourney's water-vat ultrasound scanner for "wellness" faces expert skepticism due to unproven claims and technical hurdles.
• A randomized controlled trial of artificial intelligence-based analytics for clinical deterioration — Passive AI-based predictive analytics displays did not significantly improve patient outcomes in a clinical trial.
• Expanding our AI and Healthcare Portfolio — AI deployment in healthcare, driven by profit, poses significant risks to patient safety, worker dignity, and democratic oversight.
• Healthcare AI Has a Reliability Problem Nobody Talks About — Healthcare AI needs to assess prediction and explanation stability, not just accuracy, to ensure clinical trustworthiness.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Nature Machine Intelligence.