Background and Rationale
Exercise is described to be a primary contributor to longevity, both in terms of extending lifespan as well as health span. Over the past 50 years, numerous large-cohort studies across age groups have consistently demonstrated that exercise (in various forms) helps increase lifespan, maintain flexibility and muscle strength, and improves resilience from illness1,2,3. In terms of sleep and repair, studies report that regular exercise helps combat insomnia by reducing the amount of time it takes to fall asleep and increasing general sleep volume4. A recent observational study demonstrated sleep profiles across several countries with strong regional and gender effects5. However, the contribution of exercise and fitness to the sleep profiles was not examined. The circadian phase alignment theory suggests that exercise earlier in the day should be linked to more health benefits than exercise closer to sleep time. However, some studies report that early evening exercise may offer a beneficial strategy for enhancing sleep6, while others demonstrate that evening exercise increases rapid eye movement latency and slow-wave sleep while decreasing stage 1 sleep7. Wearable technology with long-term data collection offers an ideal method to test such hypotheses across a wide range of people.
In this brief study, we looked at the impact of cardiovascular training exercise on sleep profiles of men and women across four geographic regions of different sizes (Europe, Gulf Cooperation Council Region, India, and the New York metropolitan area) to assess sleep profiles in habitual cardiovascular exercise practitioners over a 30 day period. The aim was to extract common insights for our fitness-oriented, healthy, yet demographically heterogeneous Ring AIR user base.
Methods
We undertook a retrospective analysis of individual-logged workout data collected from Ultrahuman Ring AIR users in the Europe, Gulf Cooperation Council Region, India, and the New York metropolitan areas between April 16th and May 16th, 2024. A filtering criteria was applied to select permissible workout and sleep sessions, which included: 1) users aged 20-40 years, 2) who had logged cardiovascular workout sessions (indoor/outdoor walking, cycling and running) on the Ultrahuman app, 3) wore the ring during sleep on the same day, and 4) slept between 8 p.m. and 11.59 p.m. local time the same day.
This generated yoked exercise-sleep data sets by ensuring data inclusion for only those users who had sleep data from the same nights as their respective workout logs in the day. Users were categorised into before - 1p.m. (AM) and after - 1p.m. (PM) cohorts. The AM and PM cohorts were further split into male and female groups, resulting in 4 subgroups (AM male, AM female, PM male and PM female). We excluded users who logged more than one workout or changed time zones within a given day. The global dataset analysed consisted of 1938 users (N=1202 male and N=736 female) with 6678 eligible workout and sleep sessions over a 30 day period (n=3134 AM and n=3544 PM sessions). We filtered this dataset further into the aforementioned four regions, resulting in a final combined population of 1311 users (N=877 male and N=434 female) with 4365 eligible workout sessions (n=2062 AM and n=2303 PM sessions).
We evaluated the following sleep metrics for both genders in the AM and PM cohorts: Total Sleep Time (TST), Sleep Efficiency (SE), Awake Time (time spent awake during a sleep session - AT) and step count in the day (steps) Data analyses and aggregation is covered under terms of use of the Ultrahuman Ring AIR and application. Since this is a retrospective observational analysis, data analysts had no contact with any users and worked with de-identified data. Data was statistically analysed in-house using Python-MATLAB modules, and the Mann-Whitney U test was employed for comparison within and between groups.
Result
Sleep duration and efficiency are uniformly superior across all geographic and gender groups:
We observed a set of individuals across different geographies who habitually logged a cardiovascular (aerobic) exercise session in a day for 30 days. These exercises were logged on the Ultrahuman application and not imported from other wearables that they may use. We found that the median number of steps logged across all users was 8902, with a range of 8363-10906 steps varying across regions (Table 1).
Across regions we observed that all Ring AIR users had remarkably similar sleep profiles that were independent of gender and time of exercise. Night time sleep duration ranged between 6.12-8.87 hours (h) with a median of 7.45 h. Focussing on the differences between regions, users in India and GCC countries seemed to report a marginally lower sleep duration of ~9 minutes which may have a residual effect at the end of Ramadan period. Sleep efficiency (SE) as measured by the Ultrahuman app, is the time spent asleep within one sleep session. SE for all users ranged between 78-92% with a median of 88%. By extension the amount of time spent awake across users was between 6.12-10.44% in various regions. Highest wakefulness was found in the GCC and India, whereas European Ring AIR users reported the lowest awake times (Table 1).
Table 1. Median values (Q1-Q3) for TST, SE, AT and steps in the global cohort and four target regions during the 30 day study period. Red and green colour depict lower or higher values from corresponding global median respectively.
Morning exercisers accumulate more steps throughout the day, but this comes with distinct regional and gender profiles:
We next looked to test whether there was a gender- or region-wise difference in the daily step count of users and whether timing of the cardiovascular exercise influenced the overall daily steps a user took. We found that in almost all regions except GCC, users who exercised earlier in the day (before 1pm. local time), reported higher daily step counts. This trend was found in both genders; however, the effect of timing was significantly different in males than females (Figure 1 vs 2). Comparing across regions, this early vs. late workout and daily step count trend was most pronounced in India for males and Europe for female Ring AIR users.
Figure 1: Graphical representation of median steps for the AM male and PM male groups in the global cohort and across the four regions. Statistical analysis comparing AM to PM groups using Mann-Whitney U tests. *** denotes p<0.001, ** denotes p<0.01, n.s. denotes no significant difference, N (India) = 265, N (Europe) = 463, N (GCC) = 44, N (NYC) = 105. N = combined AM and PM male users.
We expected a wide range of exercise sessions to be captured by our analysis filters. Indeed, we found the populations to have high levels of skew and relatively broad peaks (kurtosis) indicating a variation in activity levels in the individuals and presences of small clusters of users within the larger groups with comparable sleep patterns. The change of these sub-clusters according to the workout times provided an additional layer of information as described below.
In men, the entire cohort included a small set of workout sessions with extremely large volume (>20,000 steps at a stretch), possibly representing endurance activities like marathon running or a lifestyle/occupation that keeps them active. We found a similar pattern in the Indian user base, as shown by the large upward skew in the violin plot (Figure 1). European users may potentially represent two populations in the morning workout cohort with different clusters of steps. The step count distribution of late-day exercisers did not show much change across all regions except the GCC, where the users had more variation in movement in the evening.
There was comparatively lesser variation in female users across regions both in terms of step volume (skew) and in changes by exercise timing (Figure 2). Indian, European and GCC female users likely had different sub-clusters of individuals with different activity levels. The most dramatic differences between early- and later- day exercise logging within female exercisers were indicated in Europe, NYC and GCC clusters by changes in kurtosis. Finally, we noted that women who exercised later in the day seem to have a slightly increased awake percentages (0.07-2.75%) in their subsequent night time sleep, though the trends were not statistically significant.
Figure 2: Graphical representation of median steps for the AM female and PM female groups in the global cohort as well as across the four regions. Statistical analysis comparing AM to PM groups using Mann-Whitney U tests. ** denotes p<0.01, n.s. denotes no significant difference, N (India) = 63, N (Europe) = 199, N (GCC) = 35, N (NYC) = 137. N = combined AM and PM female users.
Conclusions, Limitations and Future Directions
The results from this initial investigation support previous reports that sleep volume and quality in healthy individuals who exercise regularly are enhanced. Ring AIR users reported an SE of 88/100, which puts them in the superior category of sleep quality and recovery profiles. Similarly, ~7.5h of sleep is squarely within the 7-9h sleep recommendation by the American Academy of Sleep Medicine8 guidelines that are accepted globally. Step counts indicate these individuals to be highly active averaging close to the American Heart Association's recommended daily steps, which has a strong correlation with lowered all cause mortality9. Collectively, the conclusion is that actively exercising men and women above a certain threshold consistently exhibit superior sleep profiles, and this trend is independent of geography or gender.
Circadian phase alignment posits that the more aligned eating, exercise, and sleep patterns are to diurnal rhythms, the greater the benefits towards increasing longevity. This dictates that morning exercise may lead to more overall daily movement than the same exercise done later in the day. We found that Ring AIR users who logged their cardio-workout in the morning moved more across the day as compared to those who exercised in the afternoon or early evening. We also found a wide range of step counts within the day, representing the general heterogeneity of the user base and how subsets may exist that need more nuanced examination. A further investigation into heart rate and heart rate variability is merited, which will be the focus of future studies.
These results require to be interpreted in the context of the limitations of real-world evidence gathered from a heterogeneous group of users in a specific geographic location, with a limited data stream collected from wearable devices only. We do not have data on work habits, stress profiles or metabolic state for further correlations. Furthermore, our filtering criteria surfaced geographic- groups of uneven sizes, hence only within group comparisons were possible. In addition, our analysis only pertains to cardiovascular exercise and it is possible that other forms of exercise would have different outcomes. Finally, we acknowledge the possibility that AM vs PM cohort individuals have differences in innate fitness that are being reflected in the outcomes measured. Nevertheless, we believe that this study provides an early primer into a more nuanced dissection of the relationship between sleep and exercise in an active adult with real-world lifestyles and habits.
Reach out to support@ultrahuman.com for commercial queries and science@ultrahuman.com for scientific queries.
