Navigate the healthcare industry
Healthcare Brew covers pharmaceutical developments, health startups, the latest tech, and how it impacts hospitals and providers to keep administrators and providers informed.
New year, new rules—that’s the saying, right?
The FDA sure hopes so.
On January 6, the agency shared draft guidelines about what role artificial intelligence (AI) should play in the approval of medical technology, drugs, and other biological products. It’s an increasingly important issue: AI in healthcare was a nearly $19 billion industry in 2023 and is predicted to be worth over $300 billion in 2032, according to market research firm Global Market Insights.
“With the appropriate safeguards in place, artificial intelligence has transformative potential to advance clinical research and accelerate medical product development to improve patient care,” FDA Commissioner Robert Califf said in a press release.
The FDA is seeking public comment on this draft guidance.
Drug discourse. The FDA proposed these suggestions because an increasing number of companies are including AI-generated components—like predictions or analyses—in their submissions to the agency for drug approvals. Tala Fakhouri, the FDA’s associate director for data science and AI policy, told Stat in an October interview that the agency has seen more than 500 submissions since 2016.
The agency wants to make sure this AI data is accurate.
In a draft guidance document, the agency laid out a seven-step risk-based credibility assessment framework. It included recommendations like asking applicants to explain what data they trained their models on, how their models work, and why they chose their approach.
Let’s talk tech. The FDA also shared on January 6 draft guidance for AI-enabled medical devices.
Troy Tazbaz, director of the digital health center of excellence within the FDA’s devices and radiological health center, said in the agency’s statement that it has already authorized more than 1,000 AI-enabled devices.
This new draft guidance includes what information manufacturers should put on their application for FDA approval—like, for instance, a plan for how the device will be monitored after it hits the market.
The technology draft guidance also has some recommendations to help manufacturers avoid creating bias in their AI systems.
Machine learning algorithms are trained with existing healthcare data. But people of color have been marginalized by the medical system, and the data reflects these inequities, outlets like NPR have reported.