Daily Stress and Heart Rate Variability Among Mindfulness Meditation Practitioners: mHealth Observational Study

Figure 1. Graphical Summary of our study design. Data were obtained from 90 participants divided into 3 groups—19 meditation practitioners (meditation group), 32 recreational runners (running group), and 39 individuals with neither regular meditation nor exercise routines (control group). Over a 3-week period, participants continuously wore Garmin smartwatches to monitor heart rate variability. In parallel, a smartphone-based ESM was used to assess momentary stress and activity in daily life. At the end of the study, participants completed a web-based questionnaire on subjective stress and related psychological variables. For the meditation group, daily practice times were logged via a smartphone app to enable fine-grained analysis of heart rate variability dynamics during and after meditation. ESM: experience sampling method; HRV: heart rate variability.

Figure 2. Comparative analysis of the 5 questionnaire scores. The horizontal line within each violin plot denotes the median, while the vertical line or shaded region represents the IQR, encompassing values from the 25th to 75th percentiles. Points represent scores from each participant in the control, meditation, and runner groups. (A) High International Physical Activity Questionnaire scores indicate high physical activity levels, (B) high Mindful Attention Awareness Scale scores indicate high mindfulness levels, (C) low Perceived Stress Scale scores indicate low perceived stress levels, (D) high World Health Organization Well-Being Index scores indicate high psychological well-being levels, and (E) low PSQI scores indicate better sleep quality. Significant differences were assessed using Kruskal-Wallis tests followed by Wilcoxon test. IPAQ: International Physical Activity Questionnaire; K-W: Kruskal-Wallis; MAAS: Mindful Attention Awareness Scale; n.s.: not significant (P>.05); PSQI: Pittsburgh Sleep Quality Index; PSS: Perceived Stress Scale; WHO-5: World Health Organization Well-being Index. *P<.05, **P<.01, ***P<.001 in the Wilcoxon test.

Figure 3. Comparative analysis of the heart rate variability indexes. For each participant, the median values of various heart rate variability indices were calculated from smartwatch-derived beat-to-beat intervals, and their distributions were compared across 3 groups—control, meditation, and runner. The indices include (A) 5-min root-mean-square of successive differences, (B) 24-hour standard deviation of normal-to-normal interval, and (C) root-mean-square of successive differences stratified by 4 physical activity states (sleep, rest, walk, and run), categorized using smartwatch-derived step counts and self-reported sleep times. Figure presentation and statistical tests were conducted in the same manner as in Figure 2. Significant differences were assessed using Kruskal-Wallis tests followed by Wilcoxon test. K-W: Kruskal-Wallis; n.s.: not significant (P>.05). *P<.05, **P<.01, ***P<.001 in the Wilcoxon test.

Figure 4. Associations between stress levels, daily activities, and heart rate variability. (A) Stress level distribution by activity type. The proportion of self-reported stress levels across 7 daily activity states, derived from 4557 experience sampling method responses. The vertical axis denotes the proportions of 4-staged stress levels, while the horizontal axis represents activity types. Horizontal black lines across the bars indicate the overall middle-high stress proportion. (B) root-mean-square of successive differences (RMSSD) density plots by stress levels. Posterior density plots of averaged RMSSD under no-low stress and middle-high stress conditions. RMSSD was calculated using 5-minute windows centered on reporting times. A hierarchical Bayesian model (Multimedia Appendix 12) was used to account for variability in response frequency and baseline heart rate variability across participants. Posterior densities of αₘ were estimated using Markov Chain Monte Carlo, with shaded regions denoting 95% credible intervals and vertical dotted lines indicating posterior medians. C: control; M: meditation; R: running.

Figure 5. Variation of heart rate variability levels during daily meditation practices. (A) Posterior density plots illustrating the average root-mean-square of successive differences (RMSSD) across different phases of meditation practice. RMSSD data were categorized as follows: the 10-minute before meditation phase, the during meditation phase, and the after meditation phases with 10-minute intervals up to 60 minutes post meditation. RMSSD was computed using start time-aligned 5-minute windows for the before and during phases, while end time-aligned windows were used for the after phases. The data were analyzed and visualized as in Figure 4B. (B) Start-aligned time-series RMSSD profiles across meditation practices of varying durations. Assuming the start of meditation as time 0, RMSSD was calculated using start time-aligned windows spanning from −20 to +55 minutes. Profiles were grouped by meditation duration in 10-minute intervals, and Markov Chain Monte Carlo–based hierarchical Bayesian analysis (Multimedia Appendix 13) was used to estimate posterior medians and 95% credible intervals of αₜᵏ, represented by solid and dotted lines, respectively. Horizontal time scales were aligned with…