Figure 4.

Plots of predicted age vs. chronological age are shown for the most predictive aging clock identified. The most accurate aging clock was identified by LASSO modeling of all 2978 proteins available for measurement in the plasma proteomic dataset derived from 3301 human participants (age range of 18–76 years). This clock used 491 SOMAmers, had a Pearson correlation of 0.98 in the learning set (a), a median absolute error of 1.84 years in the learning set (a), a Pearson correlation of 0.96 in the test set (b), and a median absolute error of 2.44 years in the test set (b). 2178 subjects were utilized for the learning set (a) and 1123 subjects were utilized for the test set (b). MAE = median absolute error