Wearable-derived cardiovascular fitness age and its lifestyle correlates in 442 adults

Figure 1. Distribution of the Cardio Age gap, stratified by gender. Density histograms and Gaussian kernel-density estimates of the Cardio Age gap (mean Cardio Age minus chronological age, averaged over all available days within the 12-month window) for Female (N=289) and Male (N=144) participants; the Other group (N=9) is omitted from the density overlay because of small sample size. Vertical dashed lines mark within-gender medians (Female −2.55 years, Male −0.27 years). The male distribution is shifted toward less favorable Cardio Age gap values (Mann–Whitney p<0.001; Cohen’s d=+1.21; rank-biserial r=+0.573). Negative values indicate a cardiovascular system estimated to be younger than chronological age. Youngest and Oldest Hearts groups are defined elsewhere by the extreme deciles of the Cardio Age gap distribution, not by chronological age.

Figure 2. Sleep metrics in extreme Cardio Age gap groups (12-month window). Grouped bars comparing the Youngest Hearts (10th percentile, N=45) and Oldest Hearts (90th percentile, N=45) on independent sleep metrics. All comparisons by Mann-Whitney U tests. Sample sizes vary by metric due to data availability (sleep metrics: N=42 vs. 39). ∗∗p<0.01; ∗p<0.05; NS, not significant. Youngest and Oldest Hearts groups are defined by the extreme deciles of the Cardio Age gap distribution, not by chronological age.

Figure 3. Spearman rank correlations between Cardio Age gap and wearable-derived metrics (12-month window). Horizontal bars show correlation coefficients. Metrics are grouped into three tiers: constituent (direct algorithm inputs, ‡), indirect constituent (shares algorithmic components, †), and independent lifestyle correlates. Error bars represent bootstrapped 95% confidence intervals (1,000 iterations). ∗∗∗p<0.001, ∗∗p<0.01, ∗p<0.05. NS, not significant. Spearman correlations are computed across the full cohort, not within the extreme groups; the Youngest and Oldest Hearts labels elsewhere refer to the extreme deciles of the Cardio Age gap distribution, not chronological age.

Figure 4. Body mass index dose-response relationship with Cardio Age gap (12-month window). Mean CA gap (±95% CI) by BMI† category: underweight (<18.5 kg/m2, N=17), normal weight (18.5–24.9, N=207), overweight (25.0–29.9, N=126), and obese (≥30.0, N=92). †BMI is not a direct algorithm input but is a known physiological determinant of VO2 max (the algorithm’s primary input); this association is therefore partly mediated through the algorithm.

Figure 5. Twelve-month Cardio Age trajectories. Change from baseline (first 30-day mean) for sustained large improvers (delta <−2.0 years, N=52) and sustained large worseners (delta >+2.0 years, N=59). Lines represent group medians; shaded bands represent interquartile ranges. Sustained improvers showed a mean CA reduction of 3.24 years; sustained worseners showed a mean increase of 3.49 years over the 12-month period. Improver/worsener classification is based on within-person change in Cardio Age over the 12-month window and is independent of the Youngest/Oldest Hearts extreme-decile labels used elsewhere.