Journal of Medical Internet Research - International Scientific Journal for Medical Research, Information and Communication on the Internet #Preprint #PeerReviewMe: Warning: This is a unreviewed preprint. Readers are warned that the document has not been peer-reviewed by expert/patient reviewers or an academic editor, may contain misleading claims, and is likely to undergo changes before final publication, if accepted, or may have been rejected/withdrawn. Readers with interest and expertise are encouraged to sign up as peer-reviewer, if the paper is within an open peer-review period. Please cite this preprint only for review purposes or for grant applications and CVs (if you are the author). Background: Virtually all organisms on Earth have their own circadian rhythm, and humans are no exception. Circadian rhythms are associated with various human states, especially mood disorders and disturbance of circadian rhythm is known to be very closely related. Attempts have also been made to derive clinical implications associated with mood disorders using the acquiring vast amounts of the digital log as digital technologies develop and using computational analysis techniques. Objective: The present study was conducted to evaluate the mood state/episode, activity, sleep, light exposure, and heart rate during a period of about two years by acquiring various digital log data through wearable devices and smartphone applications as well as conventional clinical assessments. We investigated a mood prediction algorithm developed with machine learning using passive data phenotypes based on circadian rhythms. Methods: We performed a prospective observational cohort study on fifty-five patients with mood disorders (major depressive disorder, bipolar disorder type 1 and 2; MDD, BD I, and BD II, respectively) for two years. A smartphone application for self-recording daily mood scores and detecting light exposure (using the installed sensor) were provided. From daily worn activity trackers, digital log data of activity, sleep, and heart rate were collected. Passive digital phenotypes were processed into 130 features based on circadian rhythms, and a mood prediction algorithm was developed by random forest. Results: The mood state prediction accuracies for next three days in all patients, MDD, BD I, and BD II were 65, 65, 64, and 65%, with 0.7, 0.69, 0.67, and 0.67 area under the curve (AUC) values, respectively. The accuracies of all patients for no episode (NE), depressive episode (DE), manic episode (ME) and hypomanic episode (HME) were 85.3, 87, 94, and 91.2%, with 0.87, 0.87, 0.958, and 0.912 of AUCs, respectively. The prediction accuracy in BD II patients was distinctively balanced high showing 82.6, 74.4, and 87.5% of the accuracies (with generally good sensitivity and specificity), with 0.919, 0.868, and 0.949 of the AUCs for NE, DE, and HME, respectively. Conclusions: Based on the theoretical basis of chronobiology, this study proposed a good model of future research by developing a mood prediction algorithm using machine learning by processing and reclassifying digital log data. In addition to academic value, it is expected that this study will be of practical help to improve the prognosis of patients with mood disorder by making it possible to apply actual clinical application due to the rapid expansion of digital technology. Clinical Trial: ClinicalTrials.gov: NCT03088657

Disturbances in circadian rhythms have been suggested as a possible cause of bipolar disorder (BD). Included in this study were 31 mood episodes of 26 BD patients, and 18 controls. Circadian rhythms of BD were evaluated at admission, at 2-week intervals during hospitalization, and at discharge. All participants wore wrist actigraphs during the studies. Saliva and buccal cells were obtained at 8:00, 11:00, 15:00, 19:00, and 23:00 for two consecutive days. Collected saliva and buccal cells were used for analysis of the cortisol and gene circadian rhythm, respectively. Circadian rhythms had different phases during acute mood episodes of BD compared to recovered states. In 23 acute manic episodes, circadian phases were ~7hour advanced (equivalent to ~17hour delayed). Phases of 21 out of these 23 cases returned to normal by ~7hour delay along with treatment, but two out of 23 cases returned to normal by ~17hour advance. In three cases of mixed manic episodes, the phases were ~6-7hour delayed. For five cases of depressive episodes, circadian rhythms phases were ~4-5hour delayed. After treatment, circadian phases resembled those of healthy controls. Circadian misalignment due to circadian rhythm phase shifts might be a pathophysiological mechanism of BD.

Background: Artificial intelligence (AI) social chatbots represent a major advancement in merging technology with mental health, offering benefits through natural and emotional communication. Unlike task-oriented chatbots, social chatbots build relationships and provide social support, which can positively impact mental health outcomes like loneliness and social anxiety. However, the specific effects and mechanisms through which these chatbots influence mental health remain underexplored. Objective: This study explores the mental health potential of AI social chatbots, focusing on their impact on loneliness and social anxiety among university students. The study seeks to (i) assess the impact of engaging with an AI social chatbot in South Korea, "Luda Lee," on these mental health outcomes over a 4-week period and (ii) analyze user experiences to identify perceived strengths and weaknesses, as well as the applicability of social chatbots in therapeutic contexts. Methods: A single-group pre-post study was conducted with university students who interacted with the chatbot for 4 weeks. Measures included loneliness, social anxiety, and mood-related symptoms such as depression, assessed at baseline, week 2, and week 4. Quantitative measures were analyzed using analysis of variance and stepwise linear regression to identify the factors affecting change. Thematic analysis was used to analyze user experiences and assess the perceived benefits and challenges of chatbots. Results: A total of 176 participants (88 males, average age=22.6 (SD 2.92)) took part in the study. Baseline measures indicated slightly elevated levels of loneliness (UCLA Loneliness Scale, mean 27.97, SD (11.07)) and social anxiety (Liebowitz Social Anxiety Scale, mean 25.3, SD (14.19)) compared to typical university students. Significant reductions were observed as loneliness decreasing by week 2 (t175=2.55, P=.02) and social anxiety decreasing by week 4 (t175=2.67, P=.01). Stepwise linear regression identified baseline loneliness (β=0.78, 95% CI 0.67 to 0.89), self-disclosure (β=–0.65, 95% CI –1.07 to –0.23) and resilience (β=0.07, 95% CI 0.01 to 0.13) as significant predictors of week 4 loneliness (R2=0.64). Baseline social anxiety (β=0.92, 95% CI 0.81 to 1.03) significantly predicted week 4 anxiety (R2=0.65). These findings indicate higher baseline loneliness, lower self-disclosure to the chatbot, and higher resilience significantly predicted higher loneliness at week 4. Additionally, higher baseline social anxiety significantly predicted higher social anxiety at week 4. Qualitative analysis highlighted the chatbot's empathy and support as features for reliability, though issues such as inconsistent responses and excessive enthusiasm occasionally disrupted user immersion. Conclusions: Social chatbots may have the potential to mitigate feelings of loneliness and social anxiety, indicating their possible utility as complementary resources in mental health interventions. User insights emphasize the importance of empathy, accessibility, and structured conversations in achieving therapeutic goals. Trial Registration: Clinical Research Information Service (CRIS) KCT0009288; https://tinyurl.com/hxrznt3t

The purpose of this study was to evaluate the applicability of data obtained from a wearable activity tracker (Fitbit Charge HR) to medical research. This was performed by comparing the wearable activity tracker (Fitbit Charge HR) with actigraphy (Actiwatch 2) for sleep evaluation and circadian rest-activity rhythm measurement. Sixteen healthy young adults (female participants, 62.5%; mean age, 22.8 years) wore the Fitbit Charge HR and the Actiwatch 2 on the same wrist; a sleep log was recorded over a 14-day period. We compared the sleep variables and circadian rest-activity rhythm measures with Wilcoxon signed-rank tests and Spearman's correlations. The periods and acrophases of the circadian rest-activity rhythms and the sleep start times did not differ and correlated significantly between the Fitbit Charge HR and the Actiwatch 2. The Fitbit Charge HR tended to overestimate the sleep durations compared with the Actiwatch 2. However, the sleep durations showed high correlation between the two devices for all days. We found that the Fitbit Charge HR showed high accuracy in sleep evaluation and circadian rest-activity rhythm measurement when compared with actigraphy for healthy young adults. The results suggest that the Fitbit Charge HR could be applicable on medical research as an alternative tool to actigraphy for sleep evaluation and measurement of the circadian rest-activity rhythm.

Wearable devices enable passive collection of sleep, heart rate, and step-count data, offering potential for mood episode prediction in mood disorder patients. However, current models often require various data types, limiting real-world application. Here, we develop models that predict future episodes using only sleep-wake data, easily gathered through smartphones and wearables when trained on an individual’s sleep-wake history and past mood episodes. Using mathematical modeling to longitudinal data from 168 patients (587 days average clinical follow-up, 267 days wearable data), we derived 36 sleep and circadian rhythm features. These features enabled accurate next-day predictions for depressive, manic, and hypomanic episodes (AUCs: 0.80, 0.98, 0.95). Notably, daily circadian phase shifts were the most significant predictors: delays linked to depressive episodes, advances to manic episodes. This prospective observational cohort study (ClinicalTrials.gov: NCT03088657, 2017-3-23) shows sleep-wake data, combined with prior mood episode history, can effectively predict mood episodes, enhancing mood disorder management.

Exposure to artificial light at night (ALAN) has become increasing common, especially in developed countries. We investigated the effect of dALAN exposure during sleep in healthy young male subjects. A total of 30 healthy young male volunteers from 21 to 29 years old were recruited for the study. They were randomly divided into two groups depending on light intensity (Group A: 5 lux and Group B: 10 lux). After a quality control process, 23 healthy subjects were included in the study (Group A: 11 subjects, Group B: 12 subjects). Subjects underwent an NPSG session with no light (Night 1) followed by an NPSG session randomly assigned to two different dim light conditions (5 or 10 lux, dom λ: 501.4 nm) for a whole night (Night 2). We found significant sleep structural differences between Nights 1 and 2, but no difference between Groups A and B. Exposure to dALAN during sleep was significantly associated with increased wake time after sleep onset (WASO; F = 7.273, p = 0.014), increased Stage N1 (F = 4.524, p = 0.045), decreased Stage N2 (F = 9.49, p = 0.006), increased Stage R (F = 6.698, p = 0.017) and non-significantly decreased REM density (F = 4.102, p = 0.056). We found that dALAN during sleep affects sleep structure. Exposure to dALAN during sleep increases the frequency of arousals, amount of shallow sleep and amount of REM sleep. This suggests adverse effects of dALAN during sleep on sleep quality and suggests the need to avoid exposure to dALAN during sleep.