Researchers have developed a new AI system that analyzes medical images by mimicking the way trained radiologists view and interpret scans. The approach leverages real-time eye movement data from radiologists to guide the AI in focusing on the most clinically relevant areas of X-rays, MRIs, and CT scans. This method not only improves the accuracy of AI diagnoses but also enhances transparency, making the AI’s decision-making process more understandable and trustworthy. By tracking where radiologists look and how long they dwell on specific regions of an image, researchers captured critical insights into their visual attention patterns. These patterns were then used to train the AI to prioritize similar regions when analyzing new images. The result is an AI model that doesn’t just detect abnormalities but does so with a focus on areas that matter most to human experts. The study highlights the importance of incorporating clinical expertise into AI development. Rather than relying solely on large datasets, this approach embeds human judgment and diagnostic intuition directly into the model’s learning process. This leads to more reliable AI tools that align with real-world medical practice. Early results show that the AI system outperforms traditional models in identifying conditions such as lung nodules, fractures, and tumors. More importantly, it provides visual explanations—highlighting the regions it focused on—giving radiologists confidence in its findings and enabling better collaboration between humans and machines. The researchers believe this method could become a standard in developing AI for healthcare, ensuring that these tools don’t just perform well on paper but also reflect the nuanced decision-making of experienced clinicians. As AI continues to integrate into medical workflows, such human-in-the-loop approaches may be key to building systems that are not only accurate but also trustworthy and clinically meaningful.