Radar Médico – IA na Medicina: regulação, ética e o que o médico precisa saber

·21 min·2 clips

How is Brazil's new 2026 resolution trying to regulate AI tools that diagnose patients?

The episode opens by framing AI's rapid integration into medicine, from image diagnosis to clinical decision support, against a backdrop of evolving regulation. It notes that by June 2024, the FDA had authorized 950 AI-based medical devices, with 723 in radiology. The host explains that many of these entered the market via a simplified 510K approval process, which often doesn't require independent clinical evidence. Specific risks are outlined, including privacy violations, diagnostic errors due to biased training data, and the 'black box' problem where clinicians can't trace an AI's reasoning. Ethical challenges around patient consent and professional responsibility are highlighted. The regulatory landscape is explored: the FDA is developing flexible guidelines, the EU's 2024 AI Act imposes strict rules on high-risk health tools, and Brazil's Federal Medical Council published Resolution CFM No. 2454 in 2026. This Brazilian resolution classifies AI systems by risk level, prohibits automated diagnosis communication, and requires transparency with patients. The episode debates arguments for and against stringent regulation, with proponents citing safety and opponents warning of stifled innovation. International perspectives from the WHO, FDA, and EMA are compared. Concrete applications are listed, such as AI for mammography analysis, electrocardiogram interpretation, and virtual triage chatbots. The conclusion emphasizes the need for doctors to critically evaluate AI tools, demand transparency about training data, and consider the ethical track record of developers. The episode ends by stressing the balance between harnessing AI's potential and ensuring patient safety.

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from Afya Podcasts