Background The vast majority of people worldwide have been impacted by coronavirus disease (COVID-19). In addition to the millions of individuals who have been infected with the disease, billions of individuals have been asked or required by local and national governments to change their behavioral patterns. Previous research on epidemics or traumatic events suggests that this can lead to profound behavioral and mental health changes; however, researchers are rarely able to track these changes with frequent, near-real-time sampling or compare their findings to previous years of data for the same individuals. Objective By combining mobile phone sensing and self-reported mental health data among college students who have been participating in a longitudinal study for the past 2 years, we sought to answer two overarching questions. First, have the behaviors and mental health of the participants changed in response to the COVID-19 pandemic compared to previous time periods? Second, are these behavior and mental health changes associated with the relative news coverage of COVID-19 in the US media? Methods Behaviors such as the number of locations visited, distance traveled, duration of phone usage, number of phone unlocks, sleep duration, and sedentary time were measured using the StudentLife smartphone sensing app. Depression and anxiety were assessed using weekly self-reported ecological momentary assessments of the Patient Health Questionnaire-4. The participants were 217 undergraduate students, with 178 (82.0%) students providing data during the Winter 2020 term. Differences in behaviors and self-reported mental health collected during the Winter 2020 term compared to previous terms in the same cohort were modeled using mixed linear models. Results During the first academic term impacted by COVID-19 (Winter 2020), individuals were more sedentary and reported increased anxiety and depression symptoms ( P <.001) relative to previous academic terms and subsequent academic breaks. Interactions between the Winter 2020 term and the week of the academic term (linear and quadratic) were significant. In a mixed linear model, phone usage, number of locations visited, and week of the term were strongly associated with increased amount of COVID-19–related news. When mental health metrics (eg, depression and anxiety) were added to the previous measures (week of term, number of locations visited, and phone usage), both anxiety ( P <.001) and depression ( P =.03) were significantly associated with COVID-19–related news. Conclusions Compared with prior academic terms, individuals in the Winter 2020 term were more sedentary, anxious, and depressed. A wide variety of behaviors, including increased phone usage, decreased physical activity, and fewer locations visited, were associated with fluctuations in COVID-19 news repo...

Background Since late 2019, the lives of people across the globe have been disrupted by COVID-19. Millions of people have become infected with the disease, while billions of people have been continually asked or required by local and national governments to change their behavioral patterns. Previous research on the COVID-19 pandemic suggests that it is associated with large-scale behavioral and mental health changes; however, few studies have been able to track these changes with frequent, near real-time sampling or compare these changes to previous years of data for the same individuals. Objective By combining mobile phone sensing and self-reported mental health data in a cohort of college-aged students enrolled in a longitudinal study, we seek to understand the behavioral and mental health impacts associated with the COVID-19 pandemic, measured by interest across the United States in the search terms coronavirus and COVID fatigue. Methods Behaviors such as the number of locations visited, distance traveled, duration of phone use, number of phone unlocks, sleep duration, and sedentary time were measured using the StudentLife mobile smartphone sensing app. Depression and anxiety were assessed using weekly self-reported ecological momentary assessments, including the Patient Health Questionnaire-4. The participants were 217 undergraduate students. Differences in behaviors and self-reported mental health collected during the Spring 2020 term, as compared to previous terms in the same cohort, were modeled using mixed linear models. Results Linear mixed models demonstrated differences in phone use, sleep, sedentary time and number of locations visited associated with the COVID-19 pandemic. In further models, these behaviors were strongly associated with increased interest in COVID fatigue. When mental health metrics (eg, depression and anxiety) were added to the previous measures (week of term, number of locations visited, phone use, sedentary time), both anxiety and depression (P<.001) were significantly associated with interest in COVID fatigue. Notably, these behavioral and mental health changes are consistent with those observed around the initial implementation of COVID-19 lockdowns in the spring of 2020. Conclusions In the initial lockdown phase of the COVID-19 pandemic, people spent more time on their phones, were more sedentary, visited fewer locations, and exhibited increased symptoms of anxiety and depression. As the pandemic persisted through the spring, people continued to exhibit very similar changes in both mental health and behaviors. Although these large-scale shifts in mental health and behaviors are unsurprising, understanding them is critical in disrupting the negative consequences to mental health during the ongoing pandemic.

Background Across college campuses, the prevalence of clinically relevant depression or anxiety is affecting more than 27% of the college population at some point between entry to college and graduation. Stress and self-esteem have both been hypothesized to contribute to depression and anxiety levels. Although contemporaneous relationships between these variables have been well-defined, the causal relationship between these mental health factors is not well understood, as frequent sampling can be invasive, and many of the current causal techniques are not well suited to investigate correlated variables. Objective This study aims to characterize the causal and contemporaneous networks between these critical mental health factors in a cohort of first-year college students and then determine if observed results replicate in a second, distinct cohort. Methods Ecological momentary assessments of depression, anxiety, stress, and self-esteem were obtained weekly from two cohorts of first-year college students for 40 weeks (1 academic year). We used the Peter and Clark Momentary Conditional Independence algorithm to identify the contemporaneous (t) and causal (t-1) network structures between these mental health metrics. Results All reported results are significant at P<.001 unless otherwise stated. Depression was causally influenced by self-esteem (t-1 rp, cohort 1 [C1]=–0.082, cohort 2 [C2]=–0.095) and itself (t-1 rp, C1=0.388, C2=0.382) in both cohorts. Anxiety was causally influenced by stress (t-1 rp, C1=0.095, C2=0.104), self-esteem (t-1 rp, C1=–0.067, C2=–0.064, P=.002), and itself (t-1 rp, of C1=0.293, C2=0.339) in both cohorts. A causal link between anxiety and depression was observed in the first cohort (t-1 rp, C1=0.109) and only observed in the second cohort with a more liberal threshold (t-1 rp, C2=0.044, P=.03). Self-esteem was only causally influenced by itself (t-1 rp, C1=0.389, C2=0.393). Stress was only causally influenced by itself (t-1 rp, C1=0.248, C2=0.273). Anxiety had positive contemporaneous links to depression (t rp, C1=0.462, C2=0.444) and stress (t rp, C1=0.354, C2=0.358). Self-esteem had negative contemporaneous links to each of the other three mental health metrics, with the strongest negative relationship being stress (t rp, C1=–0.334, C2=–0.340), followed by depression (t rp, C1=–0.302, C2=–0.274) and anxiety (t rp, C1=–0.256, C2=–0.208). Depression had positive contemporaneous links to anxiety (previously mentioned) and stress (t rp, C1=0.250, C2=0.231). Conclusions This paper is an initial attempt to describe the contemporaneous and causal relationships among these four mental health metrics in college students. We replicated previous research identifying concurrent relationships between these variables and extended them by identifying causal links among these metrics. These results provide support for the vulnerability model of depression and anxiety. Understanding how causal factors impact the evolution of these mental states over time may provide key information for targeted treatment or, perhaps more importantly, preventative interventions for individuals at risk for depression and anxiety.

There is a growing body of research revealing that longitudinal passive sensing data from smartphones and wearable devices can capture daily behavior signals for human behavior modeling, such as depression detection. Most prior studies build and evaluate machine learning models using data collected from a single population. However, to ensure that a behavior model can work for a larger group of users, its generalizability needs to be verified on multiple datasets from different populations. We present the first work evaluating cross-dataset generalizability of longitudinal behavior models, using depression detection as an application. We collect multiple longitudinal passive mobile sensing datasets with over 500 users from two institutes over a two-year span, leading to four institute-year datasets. Using the datasets, we closely re-implement and evaluated nine prior depression detection algorithms. Our experiment reveals the lack of model generalizability of these methods. We also implement eight recently popular domain generalization algorithms from the machine learning community. Our results indicate that these methods also do not generalize well on our datasets, with barely any advantage over the naive baseline of guessing the majority. We then present two new algorithms with better generalizability. Our new algorithm, Reorder, significantly and consistently outperforms existing methods on most cross-dataset generalization setups. However, the overall advantage is incremental and still has great room for improvement. Our analysis reveals that the individual differences (both within and between populations) may play the most important role in the cross-dataset generalization challenge. Finally, we provide an open-source benchmark platform GLOBEM- short for Generalization of Longitudinal BEhavior Modeling - to consolidate all 19 algorithms. GLOBEM can support researchers in using, developing, and evaluating different longitudinal behavior modeling methods. We call for researchers' attention to model generalizability evaluation for future longitudinal human behavior modeling studies.

BACKGROUND Worldwide, the vast majority of people have been impacted by COVID-19. While millions of individuals have become infected, billions of individuals have been asked or required by local and national governments to change their behavioral patterns. Previous research on epidemics or traumatic events suggest this can lead to profound behavioral and mental health changes, but rarely are researchers able to track these changes with frequent, near real-time sampling or compare these to previous years of data on the same individuals. OBJECTIVE We seek to answer two overarching questions by combining mobile phone sensing and self-reported mental health data among college students participating in a longitudinal study for the past two years. First, have behaviors and mental health changed in response to the COVID-19 pandemic as compared to previous time periods within the same participants? Second, did behavior and mental health changes track the relative news coverage of COVID-19 in the US media? METHODS Behaviors such as the number of locations visited, distance traveled, duration of phone usage, number of phone unlocks, sleep duration and sedentary time were measured using the StudentLife mobile smartphone sensing app. Depression and anxiety were assessed using weekly self-reported Ecological Momentary Assessments (EMAs) of the Patient Health Questionnaire-4 (PHQ-4). Participants were 217 undergraduate students, with 178 students having data during the Winter 2020 term. Differences in behaviors and self-reported mental health collected during the Winter 2020 term (the term in which the coronavirus pandemic started), as compared to previous terms in the same cohort, were modeled using mixed linear models. RESULTS During the initial COVID-19 impacted academic term (Winter 2020), individuals were more sedentary and reported increased anxiety and depression symptoms (P<.001), relative to the previous academic terms and subsequent academic breaks. Interactions between the Winter 2020 term and week of academic term (linear and quadratic) were significant. In a mixed linear model, phone usage, number of locations visited, and week of the term, were strongly associated with increased coronavirus-related news. When mental health metrics (e.g., depression and anxiety) were added to the previous measures (week of term, number of locations visited, and phone usage), both anxiety (P<.001) and depression (P=.029) were significantly associated with coronavirus-related news. CONCLUSIONS Compared with prior academic terms, individuals in Winter 2020 were more sedentary, anxious, and depressed. A wide variety of behaviors, including increased phone usage, decreased physical activity, and fewer locations visited, are associated with fluctuations in COVID-19 news reporting. While this large-scale shift in mental health and behavior is unsurprising, its characterization is particularly important to help guide the development of methods that could reduce the impact of future catastrophic events on the mental health of the population. CLINICALTRIAL

Background Similar to other populations with highly stigmatized medical or psychiatric conditions, people who hear voices (ie, experience auditory verbal hallucinations [AVH]) are often difficult to identify and reach for research. Technology-assisted remote research strategies reduce barriers to research recruitment; however, few studies have reported on the efficiency and effectiveness of these approaches. Objective This study introduces and evaluates the efficacy of technology-assisted remote research designed for people who experience AVH. Methods Our group developed an integrated, automated and human complementary web-based recruitment and enrollment apparatus that incorporated Google Ads, web-based screening, identification verification, hybrid automation, and interaction with live staff. We examined the efficacy of that apparatus by examining the number of web-based advertisement impressions (ie, number of times the web-based advertisement was viewed); clicks on that advertisement; engagement with web-based research materials; and the extent to which it succeeded in representing a broad sample of individuals with AVH, assessed through the self-reported AVH symptom severity and demographic representativeness (relative to the US population) of the sample recruited. Results Over an 18-month period, our Google Ads advertisement was viewed 872,496 times and clicked on 11,183 times. A total amount of US $4429.25 was spent on Google Ads, resulting in 772 individuals who experience AVH providing consent to participate in an entirely remote research study (US $0.40 per click on the advertisement and US $5.73 per consented participant) after verifying their phone number, passing a competency screening questionnaire, and providing consent. These participants reported high levels of AVH frequency (666/756, 88.1% daily or more), distress (689/755, 91.3%), and functional interference (697/755, 92.4%). They also represented a broad sample of diversity that mirrored the US population demographics. Approximately one-third (264/756, 34.9%) of the participants had never received treatment for their AVH and, therefore, were unlikely to be identified via traditional clinic-based research recruitment strategies. Conclusions Web-based procedures allow for time saving, cost-efficient, and representative recruitment of individuals with AVH and can serve as a model for future studies focusing on hard-to-reach populations.

BACKGROUND Stress levels among college students have been on the rise for the last few decades. Currently, rates of reported stress among college students are at an all-time high. Traditionally, the dominant way to assess stress levels has been through pen-and-paper surveys. OBJECTIVE The aim of this study is to use passive sensing data collected via mobile phones to obtain a rich and potentially less-biased source of data that can be used to help better understand stressors in the college experience. METHODS We used a mobile sensing app, StudentLife, in tandem with a pictorial mobile phone–based measure of stress, the Mobile Photographic Stress Meter, to investigate the situations and contexts that are more likely to precipitate stress. RESULTS Using recently developed methods for handling high-dimensional longitudinal data, penalized generalized estimating equations, we identified a set of mobile sensing features (absolute values of beta >0.001 and robust z>1.96) across the domains of social activity, movement, location, and ambient noise that were predictive of student stress levels. CONCLUSIONS By combining recent statistical methods and mobile phone sensing, we have been able to study stressors in the college experience in a way that is more objective, detailed, and less intrusive than past research. Future work can leverage information gained from passive sensing and use that to develop real-time, targeted interventions for students experiencing a stressful time.

BACKGROUND Across college campuses, the prevalence of clinically relevant depression or anxiety is affecting more than 27% of the college population at some point between entry to college and graduation. Stress and self-esteem have both been hypothesized to contribute to depression and anxiety levels. Although contemporaneous relationships between these variables have been well-defined, the causal relationship between these mental health factors is not well understood, as frequent sampling can be invasive, and many of the current causal techniques are not well suited to investigate correlated variables. OBJECTIVE This study aims to characterize the causal and contemporaneous networks between these critical mental health factors in a cohort of first-year college students and then determine if observed results replicate in a second, distinct cohort. METHODS Ecological momentary assessments of depression, anxiety, stress, and self-esteem were obtained weekly from two cohorts of first-year college students for 40 weeks (1 academic year). We used the Peter and Clark Momentary Conditional Independence algorithm to identify the contemporaneous (t) and causal (t-1) network structures between these mental health metrics. RESULTS All reported results are significant at <i>P</i>&lt;.001 unless otherwise stated. Depression was causally influenced by self-esteem (<i>t</i>-1 <i>r</i>_p, cohort 1 [C1]=–0.082, cohort 2 [C2]=–0.095) and itself (<i>t</i>-1 <i>r</i>_p, C1=0.388, C2=0.382) in both cohorts. Anxiety was causally influenced by stress (<i>t</i>-1 <i>r</i>_p, C1=0.095, C2=0.104), self-esteem (<i>t</i>-1 <i>r</i>_p, C1=–0.067, C2=–0.064, <i>P</i>=.002), and itself (<i>t</i>-1 <i>r</i>_p, of C1=0.293, C2=0.339) in both cohorts. A causal link between anxiety and depression was observed in the first cohort (<i>t</i>-1 <i>r</i>_p, C1=0.109) and only observed in the second cohort with a more liberal threshold (<i>t</i>-1 <i>r</i>_p, C2=0.044, <i>P</i>=.03). Self-esteem was only causally influenced by itself (<i>t</i>-1 <i>r</i>_p, C1=0.389, C2=0.393). Stress was only causally influenced by itself (<i>t</i>-1 <i>r</i>_p, C1=0.248, C2=0.273). Anxiety had positive contemporaneous links to depression (<i>t </i> <i>r</i>_p, C1=0.462, C2=0.444) and stress (<i>t </i> <i>r</i>_p, C1=0.354, C2=0.358). Self-esteem had negative contemporaneous links to each of the other three mental health metrics, with the strongest negative relationship being stress (<i>t </i> <i>r</i>_p, C1=–0.334, C2=–0.340), followed by depression (<i>t </i> <i>r</i>_p, C1=–0.302, C2=–0.274) and anxiety (<i>t </i> <i> r</i>_p, C1=–0.256, C2=–0.208). Depression had positive contemporaneous links to anxiety (previously mentioned) and stress (<i>t </i> <i> r</i>_p, C1=0.250, C2=0.231). CONCLUSIONS This paper is an initial attempt to describe the contemporaneous and causal relationships among these four mental health metrics in college students. We replicated previous research identifying concurrent relationships between these variables and extended them by identifying causal links among these metrics. These results provide support for the vulnerability model of depression and anxiety. Understanding how causal factors impact the evolution of these mental states over time may provide key information for targeted treatment or, perhaps more importantly, preventative interventions for individuals at risk for depression and anxiety.