What happened
Researchers built a machine learning system that can classify ADHD from brain MRI scans with 80% accuracy while also showing which brain regions it's looking at — a shift from earlier AI systems that were complete black boxes. This matters because doctors are more likely to trust and use AI tools in real clinical practice if they can understand what the system is actually seeing.
Why it matters
Most AI tools for medical diagnosis work like locked boxes — they give you an answer but won't explain their reasoning, which makes hospitals reluctant to adopt them. This system adds a transparency layer that identifies which parts of the brain the AI considered important, potentially bridging the gap between what machines can detect and what clinicians can verify.