What does a Pearson correlation of r = 0.923 mean? How does that compare to other brain age models?

The Pearson correlation (r) measures how closely the model’s predicted brain age tracks chronological age across a population. An r of 0 would mean the model has no relationship to age at all. An r of 1.0 would mean the model predicts chronological age perfectly — but just like with MAE, perfect prediction is not the goal. A model that perfectly predicts chronological age is just a birthday calculator. The value of a brain age clock is in the deviation: the gap between your predicted brain age and your chronological age is the meaningful signal.

BrainYears™ achieves r = 0.923. The square of that value (r² = 0.85) means the model explains approximately 85% of the variance in chronological age from functional brain data alone. The remaining 15% reflects real biological differences between people of the same chronological age — that is where the actionable information lives.

For context: the gold standard for brain age clocks has been structural MRI. MRI-based models achieve r values of 0.90–0.95. Published EEG-based brain age models typically report r values in the 0.70–0.85 range. BrainYears™ at r = 0.923 matches MRI-level accuracy from functional signals — the subtler, upstream changes in how the brain processes information.

The two accuracy metrics work together: r = 0.923 tells you the model reliably separates younger-functioning brains from older-functioning brains. MAE = 4.4 years tells you how close the individual predictions are on average. And within-subject stability of +0.07 years tells you that when you track the same person over time, the trajectory is precise.