The predictive value of supervised machine learning models for insomnia symptoms through smartphone usage behavior

Introduction

Digital phenotyping can be an innovative and unobtrusive way to improve the detection of insomnia. This study explores the correlations between smartphone usage features (SUF) and insomnia symptoms and their predictive value for detecting insomnia symptoms.

Methods

In an observational study of a German convenience sample, the Insomnia Severity Index (ISI) and smartphone usage data (e.g., time the screen was active, longest time the screen was inactive in the night) for the previous 7 days were obtained. SUF (e.g., min, mean) were calculated from the smartphone usage data. Correlation analyses between the ISI and SUF were conducted. For the specification of the machine learning models (ML), 80 % of the data was allocated to training, 20 % to testing, and five-fold cross-validation was used. Six algorithms (support vector machine, XGBoost, Random Forest, k-Nearest-Neighbor, Naive Bayes, and Logistic Regressions) were specified to predict ISI scores ≥15.

Results

752 participants (51.1 % female, mean ISI = 10.23, mean age = 41.92) were included in the analyses. Small correlations between some of the SUF and insomnia symptoms were found. In the ML models, sensitivity was low, ranging from 0.05 to 0.27 in the testing subsample. Random Forest and Naive Bayes were the best-performing algorithms. Yet, their AUCs (0.57, 0.58 respectively) in the testing subsample indicated a low discrimination capacity.

Conclusions

Given the small magnitude of the correlations and low discrimination capacity of the ML models, SUFs, as measured in this study, do not appear to be sufficient for detecting insomnia symptoms. Further research is necessary to explore whether examining intra-individual variations and subpopulations or employing alternative smartphone sensors yields more promising outcomes.

Social anxiety and attentional bias to negative emotional information: the relationship and intervention

BACKGROUND

According to the cognitive behavioral model of social anxiety, attentional bias to negative emotional information causes and maintains anxiety. The goal of attentional bias modification (ABM) is to reduce anxiety by reducing attention bias to negative emotional information.

METHOD

We used questionnaires and experiments to explore the improvement effect of ABM training on social anxiety in college students. In Study 1, we used dot-probe tasks to investigate the attentional bias to negative emotional information and the relationship with social anxiety severity in college students. In Study 2, college students with high social anxiety were divided into two groups: attentional bias modification training task group (ABM) and attention control condition task group (ACC). The ABM group received a continuous intervention for 10 days to observe changes in social anxiety levels and attentional bias scores in the pretest and posttest stages.

RESULTS

The results showed that the correlation of attentional bias to negative emotional information and social anxiety severity was significant. Meanwhile, the high social anxiety participants responded more quickly to negative emotional information. After the intervention, social anxiety levels and attentional bias scores of the training group were significantly reduced.

CONCLUSIONS

The results showed that attentional bias modification training can reduce attentional bias to negative emotional information in college students with social anxiety and effectively improve their social anxiety.

Problematic social media use in childhood and adolescence

At the time of writing, about 4.59 billion people use social media with many adolescents using their social media accounts across a myriad of applications and platforms. According to recent statistics, in 2022 individuals spent an average of 151 minutes on social media each day, illustrating the global relevance of social media (Dixon, 2022a,b). One of the pressing questions, internationally, is whether social media use is harmful and/or addictive. This question is of particular importance because many teenagers - and younger adolescents - spend considerable time on these platforms, which have increasingly become an integral part of their lives. Moreover, considering lifespan development, adolescents may be particularly vulnerable to specific features and advertisements shown to them on social media platforms. Growing prevalence of poor mental health in young people has led to recent recommendations in the United States to routinely screen for anxiety in 8-18 year olds, and for depression and suicide risk for adolescents between 12-18 years of age (US Preventive Services Task Force et al., 2022 a,b) - the conditions often accompanying problematic social media use. The present work not only provides insights into the current state of the literature but provides also recommendations.

Attentional bias to threat is modulated by stimulus content: an fNIRS study

People are evolutionarily predisposed to associate threat relevant stimuli with fear or aversiveness and show an attentional bias toward threat. Attentional bias modification (ABM) has been shown to reduce threat biases, while quantitative reviews assessing the effectiveness of bias modification yielded inconsistent results. The current study examined the relationship between the training effect of attentional bias to threat and the type of threatening stimuli. Twenty-two participants performed a modified dot-probe task while undergoing functional near-infrared spectroscopy (fNIRS) imaging. Results indicated that there was a strong pattern of attentional avoidance among individuals in an animal but not human threat condition. Furthermore, findings from fNIRS confirmed that the influence from type of threatening stimulus would be modulated by cortical activation patterns, especially in the ventrolateral prefrontal cortices (vlPFC) and angular gyrus. Overall, these results suggest that stimulus-specific may play a major role in personalization of specific psychological interventions.

Examining relationships between posttraumatic stress disorder severity and types of media/technology usage

OBJECTIVE

Psychopathology, cyberpsychology, and mass media theories link psychological symptoms, such as posttraumatic stress disorder (PTSD), to increased media and technology usage (MTU). Given limited research in this area, we uniquely examined if specific MTU facets were associated with PTSD symptom severity.

METHOD

A sample of 404 socioeconomically diverse adults (Mage = 35.89; 57.4% female) was recruited from Amazon's Mechanical Turk and completed the Life Events Checklist for DSM-5, Posttraumatic Stress Disorder Checklist for DSM-5, Media and Technology Usage and Attitudes Scale, and Patient Health Questionnaire-9.

RESULTS

Hierarchical multiple regression analyses indicated that, controlling for depression severity, greater frequency of TV viewing (p = .004) and media sharing (p = .040) and greater quantity of online friendships (p = .006) were associated with greater PTSD symptom severity.

CONCLUSION

Study findings suggest that the extent of MTU behaviors (especially extent of TV use, media sharing, and online friendships) are important to examine in trauma-exposed individuals with PTSD symptoms. Results generalize to trauma-exposed community members and are considered within the context of current MTU theories. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Predicting Big Five personality traits from smartphone data: A meta-analysis on the potential of digital phenotyping

OBJECTIVE

Since the first study linking recorded smartphone variables to self-reported personality in 2011, many additional studies have been published investigating this association. In the present meta-analyses, we aimed to understand how strongly personality can be predicted via smartphone data.

METHOD

Meta-analytical calculations were used to assess the association between smartphone data and Big Five traits. Because of the lack of independence of many included studies, analyses were performed using a multilevel approach.

RESULTS

Based on data collected from 21 distinct studies, extraversion showed the largest association with the digital footprints derived from smartphone data (r = .35), while remaining traits showed smaller associations (ranging from 0.23 to 0.25). For all traits except neuroticism, moderator analyses showed that prediction performance was improved when multiple features were combined together in a single predictive model. Additionally, the strength of the prediction of extraversion was improved when call and text log data were used to perform the prediction, as opposed to other types of smartphone data CONCLUSIONS: Our synthesis reveals small-to-moderate associations between smartphone activity data and Big Five traits. The opportunities, but also dangers of the digital phenotyping of personality traits based on traces of users' activity on a smartphone data are discussed.

On fear of missing out, social networks use disorder tendencies and meaning in life

Much research has been conducted on the association between social networks use disorder tendencies (SNUD; i.e. problematic social media use) and well-being. In more detail, a meta-analysis and further research suggest an inverse association between higher SNUD tendencies and lower well-being. In existing studies on social media use and well-being, this association has often been investigated by asking participants about their life satisfaction and/or emotional well-being. A path that has only recently been used to shed further light on SNUD tendencies is the assessment of meaning in life. Against this background, the present study aimed to investigate both the search for and presence of meaning in life in the context of SNUD tendencies in 955 participants. To the best of our knowledge and beyond what has been mentioned, this study is the first to investigate the links between meaning in life variables and state/trait fear of missing out (FoMO), with FoMO being a critical variable known to be linked to the overuse of digital technologies. In a structural equation model, we observed that associations between FoMO and meaning in life variables was mediated by SNUD tendencies. Although the present study is of cross-cultural nature and cannot establish causality, it might be the case that proneness to FoMO might trigger SNUD tendencies and as a consequence, results in more search for meaning in life, while hindering presence of meaning in life.

Emotion dysregulation factors associated with problematic smartphone use severity: The mediating role of fear of missing out

Based on current theoretical frameworks, there has been increasing research examining psychopathology leading to problematic smartphone use (PSU). However, less is known about the affective and cognitive processes linked to PSU. The present study aimed at analyzing the fear of missing out (FoMO) as a mediator in the association between emotion dysregulation and PSU severity. Participants were 343 U.S. undergraduate students (64.7  % female, Mage = 19.3, SD = 2.51) who completed online measures of emotion dysregulation, FoMO and PSU. A fully latent structural equation model was analyzed. Results indicate greater impulse control dysregulation was associated with heightened PSU via increased FoMO. Our findings present evidence suggesting emotion dysregulation and FoMO as affective and cognitive mechanisms associated with PSU, with FoMO serving a mediating role between impulse control and PSU severity. Clinical implications are discussed.

Biased processing of game-related information in problematic mobile gaming users

Background and aims

The present study investigated processing bias for game-related cues in problematic mobile gamers (PMGs) under or above the threshold of conscious awareness.

Methods

In Experiment 1, all participants (20 PMGs and 23 casual players (CPs)) finished a masked visual probe task during a brief (17ms) masked exposure condition. In Experiment 2, an unmasked visual probe task was conducted by an additional forty participants (20 PMGs and 20 CPs) at two exposure durations (200 and 500ms).

Results

Results showed that PMGs, but not CPs, had an attentional bias for game-related cues which had been presented with two exposure durations (17 and 200ms).

Discussion and conclusion

In conclusion, the present study provides evidence that bias in PMGs could be observed both preconsciously and consciously. The results are discussed with reference to incentive sensitization theory and automatic action schema theory.

Is Nomophobia Problematic or Functional? A Perspective from Bifactor Structure

With the extensive use of mobile phones globally, some people engage in excessive or problematic phone use behaviors. However, little is known regarding the latent structure of problematic mobile phone use. The current study employed the Chinese versions of the Nomophobia Questionnaire, Mobile Phone Addiction Tendency Scale, and Depression-Anxiety-Stress Scale-21 to explore the latent psychological structure of problematic mobile phone use and nomophobia and their associations with mental health symptoms. Results showed that a bifactor latent model best fit nomophobia, which contained a general factor and four unique factors involving the fear of being unable to access information, losing convenience, losing contact, and losing one's Internet connection. Results also showed significant correlations among latent factors of nomophobia, problematic mobile phone use, and mental health symptoms. Through these findings, we can conclude that two problematic mobile phone use behaviors share a common factor concerning excessive use, and nomophobia has independent unique factors concerning usable function. This study clarifies the structure of problematic mobile phone use, and it implies that we can distinguish problematic mobile phone use from functional use; further investigation of problematic mobile phone use is warranted.

Early self-reported post-traumatic stress symptoms after trauma exposure and associations with diagnosis of post-traumatic stress disorder at 3 months: latent profile analysis

BACKGROUND

Trauma exposure can cause post-traumatic stress symptoms (PTSS), and persistently experiencing PTSS may lead to the development of post-traumatic stress disorder (PTSD). Research has shown that PTSS that emerged within days of trauma was a robust predictor of PTSD development.

AIMS

To investigate patterns of early stress responses to trauma and their associations with development of PTSD.

METHOD

We recruited 247 civilian trauma survivors from a local hospital emergency department. The PTSD Checklist for DSM-5 (PCL-5) and Acute Stress Disorder Scale (ASDS) were completed within 2 weeks after the traumatic event. Additionally, 3 months post-trauma 146 of these participants completed a PTSD diagnostic interview using the Clinician Administered PTSD Scale for DSM-5.

RESULTS

We first used latent profile analysis on four symptom clusters of the PCL-5 and the dissociation symptom cluster of the ASDS and determined that a four-profile model ('severe symptoms', 'moderate symptoms', 'mild symptoms', 'minimal symptoms') was optimal based on multiple fit indices. Gender was found to be predictive of profile membership. We then found a significant association between subgroup membership and PTSD diagnosis (χ2(3) = 11.85, P < 0.01, Cramer's V = 0.263). Post hoc analysis revealed that this association was driven by participants in the 'severe symptoms' profile, who had a greater likelihood of developing PTSD.

CONCLUSIONS

These findings fill the knowledge gap of identifying possible subgroups of individuals based on their PTSS severity during the early post-trauma period and investigating the relationship between subgroup membership and PTSD development, which have important implications for clinical practice.

Longitudinal PTSD symptom trajectories: Relative contributions of state anxiety, depression, and emotion dysregulation

BACKGROUND

Prospective research on the development and trajectory of PTSD symptoms after a traumatic event is crucial for assessment and early intervention. Further, examining predictors of PTSD pathology provides a better conceptualization of the temporal course of PTSD in trauma victims.

METHODS

The present study examined PTSD symptom severity in individuals presenting to the emergency department (ED) following a traumatic event. Participants (N = 147) were assessed at four timepoints: 2-weeks, 3-months, between 6 and 9 months, and 12-months after ED admission. Growth curve modeling was conducted to examine changes in PTSD symptom severity over time. Age, sex, state anxiety, trait anxiety, emotion dysregulation, depression, and trauma type (motor vehicle accident [MVA] and assault), and PTSD diagnosis were included as covariates in the model.

RESULTS

Results demonstrated that baseline PTSD symptom severity was positively associated with severity of depression and state (but not trait) anxiety, emotion dysregulation, and PTSD diagnosis. Results also revealed significant associations with PTSD symptom changes over time; greater state anxiety and depression symptoms at baseline were associated with steeper declines in PTSD symptoms over time.

LIMITATIONS

Data were collected at only four timepoints over the course of 12-months. Results may be different with more measurement points over longer periods and inclusion of pre-, peri- and post-trauma risk factors.

CONCLUSIONS

Results illustrate the relevance of assessing state anxiety, depression, and emotion dysregulation in following trauma victims for trauma-related psychopathology over the course of time to alleviate the negative impact of the same.

The relationship between adolescent emotion dysregulation and problematic technology use: Systematic review of the empirical literature

BACKGROUND AND AIMS

We aimed to systematically identify, evaluate and summarize the research on adolescent emotion dysregulation and problematic technology use. We critically appraise strengths and limitations and provide recommendations for future research.

METHODS

We followed the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) and conducted a systematic review of published original reports on adolescent emotion dysregulation and problematic technology use published until March 1, 2022. A thorough search preceded the selection of studies matching prespecified criteria. Strengths and limitations of selected studies, regarding design and reporting, were identified based on current best practices.

RESULTS

39 studies met inclusion criteria. All of these studies provided on the relationship between adolescent emotion dysregulation and problematic technology use severity based on self-report data.

DISCUSSION

There was a positive correlation between adolescent emotion dysregulation and the severity of problematic technology use. Beyond this, other variables (such as anxiety, depression, self-esteem, etc.) were also closely related to emotion dysregulation and problematic technology use. Such studies are of importance to better understand cause-effect relations regarding both variables.

Hyperinsulinemia Associated Depression

Hyperinsulinemia promotes fat accumulation, causing obesity. Being an inflammatory state, obesity can induce further inflammation and is a risk factor for HPA (hypothalamic pituitary axis) dysregulation through hypercortisolism-related hyperglycemia. In another hypothesis, the sympathetic nervous system (SNS) plays a significant role in the regulation of hormone secretion from the pancreas such as an increase in catecholamines and glucagon as well as a decrease in plasma insulin levels, a disruption on SNS activity increases insulin levels, and induces glycogenolysis in the liver and lipolysis in adipose tissue during hypoglycemia. Hyperglycemia-hyperinsulinemia exacerbates inflammation and increases the oxidative stress along with regulating the levels of norepinephrine in the brain sympathetic system. Increased inflammatory cytokines have also been shown to disrupt neurotransmitter metabolism and synaptic plasticity which play a role in the development of depression via inhibiting serotonin, dopamine, melatonin, and glutamate signaling. An increased level of plasma insulin over time in the absence of exercising causes accumulation of lipid droplets in hepatocytes and striated muscles thus preventing the movement of glucose transporters shown to result in an increase in insulin resistance due to obesity and further culminates into depression. Further hyperinsulinemia-hyperglycemia condition arising due to exogenous insulin supplementation for diabetes management may also lead to physiological hyperinsulinemia associated depression. Triple therapy with SSRI, bupropion, and cognitive behavioral therapy aids in improving glycemic control, lowering fasting blood glucose, decreasing the chances of relapse, as well as decreasing cortisol levels to improve cognition and the underlying depression. Restoring the gut microbiota has also been shown to restore insulin sensitivity and reduce anxiety and depression symptoms in patients.

Exploring subtypes and correlates of internet gaming disorder severity among adolescents during COVID-19 in China: A latent class analysis

The WHO recently included Gaming Disorder as a psychiatric diagnosis. Whether there are distinct groups of adolescents who differ based on severity of gaming disorder and their relationships with other mental health and addictive behavior outcomes, including problematic smartphone use (PSU), remains unclear. The current study explored and identified subtypes of Internet Gaming Disorder (IGD) severity and estimated the association between these subtypes and other disorders. Participants completed online questionnaires assessing the severity of IGD, PSU, depression, and anxiety during COVID-19. We conducted a latent class analysis of IGD symptoms among 1,305 Chinese adolescents (mean age = 15.2; male = 58.5%) from 11 secondary schools in Macao (SAR), China. Multinomial logistic regression estimated correlates of latent class membership and PSU. A 4-class model adequately described the sample subgroups. Classes were labeled as normative gamers (30.9%), occasional gamers (42.4%), problematic gamers (22.7%), and addictive gamers (4.1%). Relative to normative gamers, PSU severity, depression, and being male were significantly higher among problematic gamers, addictive gamers, and occasional gamers. Only problematic gamers showed significant positive associations with anxiety severity compared to the other groups. The study revealed the differences in severity of gaming disorder and its association with psychopathology outcomes. Application in screening for IGD and comorbidity is discussed.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12144-022-03133-8.

Nomophobia and relationships with latent classes of solitude

Recently, nomophobia (separation anxiety from mobile phone) has become a common phenomenon. The authors' main purpose was to explore latent classes of solitude behaviors and how they are related to nomophobia. Chinese versions of the Nomophobia Scale and the Solitude Behaviour Scale were used in a sample of college students (351 female and 327 male). Latent class analysis, analysis of variance, and regression analysis were employed to classify solitude behaviors and explore the relationship between solitude and nomophobia. A six-class model best fit the data (BIC = 60086.49). Significant differences among the classes were found on nomophobia. Loneliness, social avoidance, and eccentricity significantly predicted nomophobia. Solitude behaviors of college students can be divided into six latent classes. The classes with a high response preference for solitude scored higher on nomophobia, especially the fear of losing an Internet connection. Not self-determined solitude and negative-solitude had a positive effect on nomophobia.

Willingness to accept (WTA), willingness to pay (WTP), and the WTA/WTP disparity in Chinese social media platforms: Descriptive statistics and associations with personality and social media use

The amount of money individuals were willing to accept (WTA) to discontinue using prominent Chinese social media platforms (WeChat/QQ), the willingness to pay (WTP) for using these platforms, as well as WTA/WTP disparities were investigated in between-groups and within-subjects design studies to examine their existence, size, and psychological correlates in the form of personality and social media use habits. Individuals were recruited at Chinese universities in three separate surveys. For between-groups investigations, four samples were investigated: WTA and WTP samples for investigations in the context of WeChat as well as WTA and WTP samples for QQ. For within-subjects investigations, individuals completed items on WTA and WTP for WeChat/QQ, the Big Five Inventory, time spent on WeChat/QQ, and the short Bergen Social Media Addiction Scale. Two samples providing data on WeChat and QQ, respectively, were investigated. Across study designs and for both WeChat and QQ we found evidence for high WTA and comparatively low WTP scores, thus, large WTA/WTP disparities. Individual differences in the disparities were negatively associated with Openness across social media platforms. The results reveal a generally low acceptance to pay for social media use, which is important against the background of discussions on monetary payment models. Moreover, a complex interplay between individual characteristics, characteristics of the service, and how and why the service is used seems to underly WTA and the WTA/WTP disparity. Finally, methodological implications of the present results for forthcoming studies assessing valuation (WTA, WTP) in the context of social media are discussed.