Figure 5.

HRV predictions using multisignal inputs and circadian rhythms are more heterogeneous than for HR

(A) The amount of variance of the HR signal explained by circadian rhythms (light orange) and a combined model with circadian rhythms and physical activity (dark orange). Error bars represent the 5th and 95th percentiles of the posterior distribution.

(B) The amount of variance of the HRV (RMSSD⁻¹) signal explained by circadian rhythms (light orange), a combined model with circadian rhythms and physical activity (medium orange), and a combined model with circadian rhythms, physical activity, and HR (dark orange).

(C and D) Examples comparing HR data with model predictions for two participants. Red, HR data; black, baseline circadian rhythm; green, physical activity (shown on normalized scale where 1 represents the maximum value); orange, model prediction with circadian rhythm and integrating activity.

(E and F) Examples comparing HRV data with model predictions for two participants. Purple, HRV data; black, baseline circadian rhythm; green, physical activity (shown on normalized scale where 1 represents the maximum value); orange, model prediction using circadian rhythms, physical activity, and HR. Participant order is the same in (A) and (B).

See also Figure S11.