A total of 250 study participants were recruited from Korea University and Datamaker from March 2023 to November 2023 through advertisements on internet communities and campus noticeboards. The participants included both individuals with and without insomnia aged 19-70 years with the following inclusion criteria: more than 3 days a week of subjective insomnia symptoms during the past 3 months and an ISI score greater than 15 for the insomnia group, and fewer than 3 days a week of subjective insomnia symptoms during the past 3 months and an ISI score less than 8 for the no insomnia group. The exclusion criteria for the insomnia group were the following: history of intellectual disability or sign of organic brain injury; diagnosis of schizophrenia spectrum disorder; currently treated for a sleep disorder, such as narcolepsy, sleep-related breathing disorder, and parasomnia; and a smartphone nonuser. In addition to these, the exclusion criteria for the no insomnia group included the diagnosis of major mood disorder and anxiety disorder, shift worker, and current user of anxiolytic-sedatives or sleep-related functional foods.

The study participants provided informed consent before the study enrollment and completed the baseline assessments including demographic information; current and past medical and psychiatric history for all participants; and current medications, specifically sleep-related medication and functional foods for the insomnia group. In addition, the study participants were provided with wearable devices (Fitbit Inspire 3, Fitbit Inc) for the collection of step counts, heart rate, and sleep information. The participants also used a smartphone app named SOMDAY (Lumanlab Inc) as an ecological momentary assessment tool. They downloaded the smartphone app and recorded information on their daily alcohol consumption, smoking, caffeine intake, food consumption, and exercise.

First, a normality test of the 4 groups classified by the ISI score (no insomnia group, mild insomnia group, moderate insomnia group, and severe insomnia group) was performed. A Kruskal-Wallis test and a paired Wilcoxon’s rank sum test were performed when normal distribution could not be assumed. By the Bonferroni method, P<.0083 was considered statistically significant. When the normal distribution could be assumed, a one-way analysis of variance was performed for continuous data, and the Tukey honest significant different test was used for post hoc analysis. A chi-square test was used for the categorical data.

All study procedures were approved by the Institutional Review Board of Korea University Anam Hospital (2022AN0587), Seoul, Korea. Written informed consent was obtained from all the study participants at the beginning of the study. A compensation of 100,000 KRW (US $68.36) was provided for the completion of the 4-week study. The collected data were anonymized.

This study aimed to examine how the severity of insomnia, as measured by the ISI, reflects objective sleep quality and correlates with other sleep-related subjective and objective parameters, including mood state, substance use, circadian rhythm, sleep-related beliefs, and neurological symptoms. A distinctive feature and strength of this study is the use of digital phenotyping, leveraging wearable devices and smartphone apps to collect real-world, continuous data on participants’ sleep patterns and related behaviors. This digital phenotyping approach allows for a more comprehensive understanding of insomnia, capturing both physiological and behavioral aspects in a naturalistic environment.

The results indicated that there were no significant differences in general sleep structure, such as total sleep time and sleep stages, as measured by wearable devices, between individuals with no insomnia and those with varying levels of insomnia severity. This lack of significant differences might be due to the objective sleep parameters being collected using wearable devices and that subjective insomnia severity does not necessarily align with objectively measured sleep quality, emphasizing the importance of factors beyond traditional sleep metrics [31].

First, the objective sleep parameters in this study, including total sleep time, sleep stages, and circadian rhythm metrics, were measured using wearable devices rather than polysomnography, which may influence the interpretation of discrepancies observed between subjective and objective sleep assessments [11]. Digital phenotyping, through the use of wearable devices, provided continuous, real-world data that facilitated the assessment of sleep parameters without the constraints of sleep laboratories, which might not reflect natural sleep patterns [32]. This real-world approach may better capture the complexities of insomnia, which are often influenced by environmental and behavioral factors. The ability of digital phenotyping to provide both objective and subjective data highlights its potential for improving our understanding of insomnia’s multifaceted nature [33].

Second, a key finding was that participants in the insomnia groups had higher scores in measures of stress, dysfunctional beliefs about sleep, and symptoms of RLS compared with the no insomnia group. This highlights the potential role of psychological and physiological factors in shaping perceptions of insomnia severity. Stress and cognitive factors, such as dysfunctional beliefs about sleep, have been linked to increased sleep difficulties, which can exacerbate subjective complaints even in the absence of objective sleep disruptions [34,35]. Therefore, this finding underscores the necessity for treatment strategies that address these psychological elements, such as cognitive behavioral therapy for insomnia [36,37].

Furthermore, the participants in the insomnia groups had higher scores in measures of smartphone overuse, biological rhythm disturbance, anxiety as a state, depression, and generalized anxiety than the no insomnia group. Interestingly, the scores of such measures were significantly different among the insomnia groups as well; the higher the severity of subjective insomnia, the higher the scores of the measures. Another notable finding of this study is that such psychological and behavioral factors affect the perceived, subjective insomnia severity without making significant differences in the activity-based objective sleep parameters. Excessive smartphone use is consistently associated with poor sleep quality, depression, and anxiety [38-41]. Likewise, depression and anxiety are well-known risk factors for poor sleep quality. Depression, anxiety, and excessive smartphone use can lead to biological rhythm disturbances and vice versa. The intricate relationship among psychological distress, behaviors such as smartphone overuse, and biological rhythm disturbance remains to be revealed in future research.

Clinical implications of these findings include the importance of using both subjective and objective assessments in the diagnosis and treatment of insomnia. For example, cognitive behavioral therapy for insomnia, which has been shown to be effective in modifying dysfunctional sleep beliefs [42], may be particularly beneficial for patients whose subjective insomnia symptoms do not match objective findings. In addition, digital phenotyping could play a significant role in treatment monitoring and tailoring interventions based on individual patient needs. This capability is particularly advantageous when compared with traditional polysomnography, as wearable devices provide ongoing monitoring that can be used to assess changes in sleep patterns over time in a nonintrusive manner.

However, this study has several limitations. The reliance on wearable devices, while offering a practical method for assessing sleep in a naturalistic setting, may not provide the same level of accuracy as polysomnography, the gold standard in sleep assessment [43]. The American Academy of Sleep Medicine stated that “consumer sleep technologies cannot be used for the diagnosis and treatment of sleep disorders at this time” due to the lack of validation against polysomnography and US Food and Drug Administration clearance [44]. For example, sleep onset latency is found to be underestimated by Fitbit, leading to the possible overestimation of total sleep time. However, Fitbit overall showed high sensitivity (0.95-0.96) and relatively low specificity (0.58-0.69) for sleep detection in a meta-analysis of the accuracy of Fitbit and polysomnography, with no significant differences in wake time after sleep onset, total sleep time, and sleep efficiency [45]. Thus, Fitbit is considered a useful alternative to collecting objective sleep parameters and is widely used for research, including the All of Us research program and the Adolescent Brain Cognitive Development study of the National Institutes of Health [46,47]. Future studies that further integrate both polysomnographic and wearable device data to provide a more comprehensive view of sleep and to validate the findings from wearable technologies are warranted to overcome the current limitations of wearable devices in measuring sleep parameters. In addition, the sleep parameters collected across the no insomnia group and insomnia groups in this study were through the uniformly applied Fitbit and its algorithm. Even though the measurement reliability of sleep parameters obtained from Fitbit is yet to be sufficiently validated, there were no device-related deviations when comparing sleep parameters across the groups. Since evaluating differences across the groups was the main interest, rather than examining the values of sleep parameters, comparing the groups using uniformly measured parameters serves its purpose.

Furthermore, the study sample, predominantly composed of volunteers recruited from specific settings, may limit the generalizability of the findings to the broader population. The two-site recruitment may lead to a potential bias, and a younger population is prone to participate in the study due to their accessibility to and familiarity with the digital technology in this naturalistic study design. Given these characteristics and limitations of the study sample, caution is required when interpreting the results. As an important contributor to and possible consequence of insomnia, naps were collected as self-reported sleep behavior recorded by the study participants using the SOMDAY app. In this study, nap was measured as a categorical variable, and it would have been more informative if objective, continuous naptime data had been included.

Future research should clearly delineate the data sources, including wearable devices and polysomnography, to avoid potential misinterpretations of the sleep metrics used. In addition, future research should explore the relationship between psychological factors, such as anxiety and stress, and their impact on both subjective and objective sleep parameters using digital phenotyping methods. Studies that include diverse populations and combine multiple methods of sleep assessment will be crucial for understanding the full spectrum of insomnia. Moreover, further exploration of the role of circadian rhythms in subjective sleep complaints could help to develop chronotherapy-based interventions that may be effective for certain subsets of patients with insomnia. Digital phenotyping could also be used to assess circadian disruptions in real time and provide personalized interventions.

In summary, this study highlights the complex interplay between subjective insomnia severity and objective sleep parameters and underscores the strengths of using digital phenotyping to assess these relationships in real-world settings. The lack of significant differences in objective sleep across insomnia severities suggests that insomnia is influenced by factors beyond sleep quantity and quality, particularly psychological elements such as stress, dysfunctional beliefs, depression, and anxiety. These findings advocate for a comprehensive treatment approach that includes both cognitive and physiological aspects of sleep disturbances, facilitated by the capabilities of digital phenotyping.

In conclusion, this study underscores the complex interplay between subjective insomnia severity and objective sleep parameters, highlighting the value of digital phenotyping in understanding these relationships in real-world settings. The findings reveal that subjective perceptions of insomnia are often not aligned with objective sleep metrics, emphasizing the role of psychological factors, such as stress, dysfunctional beliefs, depression, and anxiety, in shaping insomnia severity. This discrepancy points to the need for comprehensive treatment strategies that integrate both cognitive and physiological aspects of sleep disturbances.

Digital phenotyping, using wearable devices and smartphone apps, offers a unique advantage in capturing continuous, naturalistic sleep data, providing insights that are not possible through traditional sleep laboratory settings alone. Future research and clinical interventions should continue to leverage these technologies to develop personalized and effective treatment approaches for individuals with insomnia, addressing both their subjective experiences and objective sleep health.