Loading Curie Briefs...
Getting the latest healthcare news for you
Getting the latest healthcare news for you
A single blood test may predict your disease risk years before symptoms appear. Researchers developed machine-learning "cellular aging clocks" that map over 7,000 plasma proteins to more than 40 cell types, revealing which cells are aging fastest. The tool predicted risks for Alzheimer's, ALS, lung cancer, type 2 diabetes, and overall mortality — in some cases up to 15 years in advance.
What if a single blood test could tell you which of your cells are aging too fast — and what diseases that might bring? That's the promise of a new machine-learning framework reviewed in Cell Reports Medicine, which builds "cellular aging clocks" from plasma proteins found in a routine blood draw.
The system maps more than 7,000 circulating proteins to over 40 distinct cell types using Human Protein Atlas data, then uses ML models to calculate a biological age for each cell type. Validated across nearly 60,000 individuals from multiple large cohorts, the framework revealed that biological aging is far from uniform — different cell populations age at dramatically different rates within the same person.
The disease predictions were striking. Extreme astrocyte aging was the single strongest predictor of Alzheimer's disease — outperforming even APOE4 carrier status and polygenic risk scores.
By the numbers:
Why it matters: This framework could shift disease prevention from reactive to truly predictive — identifying high-risk individuals years before symptoms emerge. While broader validation in diverse populations and regulatory review are still needed, it marks a meaningful step toward precision medicine that goes beyond static genetic snapshots.
