# ANaMed BioAge — Model Card

## rTL research model
- Target: rTL_0_100
- Algorithm: Ridge regression
- N valid target: 6719
- Features: 40
- Validation: GroupKFold by registration office
- R²: 0.0273
- MAE: 6.451
- RMSE: 8.111
- Status: research use only; not a diagnostic test

## Short-rTL research classifier
- Algorithm: Logistic regression
- Definition: lower 20% versus upper 20% rTL, fold-specific thresholds in validation
- GroupKFold AUC: 0.6316
- Balanced accuracy: 0.5932

## CoreClock
CoreClockScore_beta is implemented as a deterministic transparent calculation from three reconstructed components. It is not presented as an independent ML target.

## Data quality
- Dataset: 6,720 × 292
- QC flags: 116 across 114 records
- Office effect on rTL: η²=0.0491