AI Brain Scan Score Tracks Alzheimer's Progression Continuously

Alzheimer's disease doesn't hit like a light switch. It creeps in slowly, and people live on a spectrum — from normal aging to full dementia. But most current AI models look at one brain scan in time and output a label: 'healthy' or 'Alzheimer's,' maybe a 'mild cognitive impairment.' That misses the nuance.

A team of researchers has now built a system called DCP (short for Disease Continuum Positioning) that uses a type of brain scan called diffusion tensor imaging, which tracks how water moves through brain tissue, revealing the health of nerve pathways. The AI doesn't just snap a label on you. Instead, it watches how your brain changes over multiple scans and assigns a single number — the Disease Continuum Score — that reflects exactly where you sit on the Alzheimer's path. Because it also calculates uncertainty, doctors know how confident to be in that number.

The system was trained on data from the Alzheimer's Disease Neuroimaging Initiative, a large public database. It beat standard methods at predicting who would progress and how fast. Importantly, the score matched clinical tests and kept consistent across time, meaning it could be used to track how someone responds to treatment or a new drug.

What's promising is the shift in thinking: instead of giving a crude category, this gives a ruler. It's like moving from 'your fever is high' to 'your temperature is 102.3°F and rising.' That kind of precision could help people get interventions earlier, when they might work better — and give researchers a sharper tool for testing therapies.