CARE as a wearable derived feature linking circadian amplitude to human cognitive functions

Abstract

Circadian rhythms are crucial for regulating physiological and behavioral processes. Pineal hormone melatonin is often used to measure circadian amplitude but its collection is costly and time-consuming. Wearable activity data are promising alternative, but the most commonly used measure, relative amplitude, is subject to behavioral masking. In this study, we firstly derive a feature named circadian activity rhythm energy (CARE) to better characterize circadian amplitude and validate CARE by correlating it with melatonin amplitude (Pearson's r = 0.46, P = 0.007) among 33 healthy participants. Then we investigate its association with cognitive functions in an adolescent dataset (Chinese SCHEDULE-A, n = 1703) and an adult dataset (UK Biobank, n = 92,202), and find that CARE is significantly associated with Global Executive Composite (β = 30.86, P = 0.016) in adolescents, and reasoning ability, short-term memory, and prospective memory (OR = 0.01, 3.42, and 11.47 respectively, all P < 0.001) in adults. Finally, we identify one genetic locus with 126 CARE-associated SNPs using the genome-wide association study, of which 109 variants are used as instrumental variables in the Mendelian Randomization analysis, and the results show a significant causal effect of CARE on reasoning ability, short-term memory, and prospective memory (β = -59.91, 7.94, and 16.85 respectively, all P < 0.0001). The present study suggests that CARE is an effective wearable-based metric of circadian amplitude with a strong genetic basis and clinical significance, and its adoption can facilitate future circadian studies and potential intervention strategies to improve circadian rhythms and cognitive functions.

Fig. 1. Workflow chart of the study.

CARE circadian activity rhythm energy. *We also selected the significant loci that were associated with the relative amplitude. However, only three single-nucleotide polymorphisms (SNPs) were found, the number of which was too small to conduct further causal analysis.

Fig. 2. Pipeline to derive the measure circadian activity rhythm energy (CARE) from accelerometer data.

a Actogram to show 7 days of wrist accelerometer activity of a single participant. b A visual representation of the raw activity signal and the singular spectrum analysis (SSA) decomposed signals, including the base signal (i.e., the first sub-signal indicating the non-periodic trend with the largest energy among all sub-signals), the 24-h signal, and the behavioral noise signal (i.e., sum of the sub-signals with periods <24 h). We can obtain the raw activity signal by adding these three signals together. c The graphical display of SSA algorithm. Specifically, the activity time series ${\mathrm{X}}_N$ of length N could be decomposed using SSA as follows. First, we chose an appropriate window length L such that $2\le L\le \frac{N}{2}$. Then, ${\mathrm{X}}_N$ was transferred into a trajectory matrix with K lagged vectors of ${\mathrm{X}}_L$ as given by T, where K = N – L + 1. The trajectory matrix T was decomposed by singular value decomposition. By grouping the eigentriples and averaging the elements of reconstructed trajectory matrix along anti diagonals, we could get filtered time series represented by ${\mathrm{G}}_N=\left\{g_1,g_2,\dots ,g_N\right\}$.

Fig. 3. Correlations between wearable-derived CARE, relative amplitude, and relative energy of behavioral noise with melatonin amplitude.

CARE circadian activity rhythm energy. Scatter plots between CARE, relative amplitude, and relative energy of behavioral noise with melatonin amplitude in the melatonin study are shown in the bottom-left corner, and the correlation coefficients between these variables are shown in the upper-right corner. The black lines indicate the linear regression fits, and the shaded areas represent the confidence intervals of the fitted mean values. The correlation coefficients annotated with an asterisk (*) indicate that the correlation is significant at P < 0.05 level, while two asterisks (**) indicate that the correlation is significant at P < 0.01 level.

Fig. 4. Manhattan and QQ plots for CARE in genome-wide association study in the adult sample (n = 85,361).

a The Manhattan plot shows association test (−log₁₀ P value on the y-axis against physical autosomal location on the x-axis). The red line represents the genome-wide significant locus (P < 5×10⁻⁸). Heritability estimates were calculated using LDSC tool. b The QQ plot identifies a slight inflation (λ_GC = 1.10) in the test statistic.