Table 2.

Association of the biological age- and all-cause mortality -predicting models with all-cause mortality.

HRA Covariates	Age, Gender, Smoking, Diabetes, Hypertension	Age, Gender, Smoking, Diabetes, Hypertension, Activity
Model	Hazard Ratio, 95% CI, p-value	Hazard Ratio, 95% CI, p-value
PCA Age	1.45, CI[1.35, 1.55] (4.4 yr., p < 10−10)	1.07, CI[0.99, 1.15] (0.8 yr., p = 0.08)
REG Age	1.31, CI[1.21, 1.42] (3.2 yr., p < 10−10)	1.01, CI[0.93, 1.08] (0.0 yr., p = 1.0)
CNN Age	1.05, CI[0.96, 1.15] (0.6 yr., p = 0.3)	0.94, CI[0.86, 1.03] (−0.7 yr., p = 0.2) (wrong sign)
Hazard Weighted regr	1.27, CI[1.21, 1.32] (2.8 yr., p < 10−10)	1.08, CI[1.01, 1.16] (0.9 yr., p = 0.03)
Hazard CNN	1.53, CI[1.42, 1.65] (5.0 yr., p < 10−10)	1.15, CI[1.07, 1.24] (1.6 yr., p = 3 × 10−4)

Two class of models were evaluated: one including the HRA parameters: age, gender, diabetes, smoking and hypertension as covariates (the left column) and the other including additionally the negative logarithm of the average daily physical activity (the biological age proxy, the right column). The corresponding 95% hazard ratio intervals are given along with the significance p-value and the effect levels, expressed in years of life lost (see the text for details).