Table 3. Cox elastic net regression model.
| Covariate (X) | Abbreviation | Coefficients (β) |
| DNAm adrenomedullin | DNAmADM | 0.00609 |
| DNAm beta-2-microglobulin | DNAmB2M | 2.79E-07 |
| DNAm cystatin-C | DNAmCystatin C | 4.08E-06 |
| DNAm growth differentiation factor 15 | DNAmGDF-15 | 0.00035 |
| DNAm leptin | DNAmLeptin | -2.03E-05 |
| DNAm log C-reactive protein | DNAmlogCRP | 1.90266 |
| DNAm log hemoglobin A1C | DNAmlogA1C | 0.40359 |
| DNAm plasminogen activator inhibitor 1 | DNAmPAI-1 | 0.02941 |
| DNAm tissue inhibitor metalloproteinases 1 | DNAmTIMP-1 | 3.67E-06 |
| DNAm pack-years | DNAmPACKYRS | 0.00014 |
| Chronological age | Age | 0.02676 |
| Female | Female | -0.14212 |
The table lists the finalized covariates in the final Cox regression model with elastic net penalty and their coeffects. A linear combination of XTβ yields an estimate of logarithm of proportional hazard, which is the raw value of DNAmGrimAge2 before calibration. The finalized DNAm GrimAge2 is based on transforming the raw variable into a distribution in units of year The columns report the name of the covariate (e.g. DNAm based biomarker), its abbreviation and coefficient under the final Cox regression model with tuning parameter λ determined by 10-fold cross validation.