Selected Publications

Smartphones are capable of passively capturing persons’ social interactions, movement patterns, physiological activation, and physical environment. Nevertheless, little research has examined whether momentary anxiety symptoms can be accurately assessed using these methodologies. In this research, we utilize smartphone sensors and personalized deep learning models to predict future anxiety symptoms among a sample reporting clinical anxiety disorder symptoms. Participants (N = 32) with generalized anxiety disorder and/or social anxiety disorder (based on self-report) installed a smartphone application and completed ecological momentary assessment symptoms assessing their anxiety and avoidance symptoms hourly for the course of one week (T = 2007 assessments). During the same period, the smartphone app collected information about physiological activation (heart rate and heart rate variability), exposure to light, social contact, and GPS location. GPS locations were coded to reveal the type of location and the weather information. Personalized deep learning models using the smartphone sensor data were capable of predicting the majority of total variation in anxiety symptoms (R2 = 0.748) and predicting a large proportion of within-person variation at the hour-by-hour level (mean R2 = 0.385). These results suggest that personalized deep learning models using smartphone sensor data are capable of accurately predicting future anxiety disorder symptom changes.
Behaviour Research and Therapy, 2021

Journal of Medical Internet Research (JMIR), 2021

Background: The current COVID-19 coronavirus pandemic is an emergency on a global scale, with huge swathes of the population required to remain indoors for prolonged periods to tackle the virus. In this new context, individuals’ health-promoting routines are under greater strain, contributing to poorer mental and physical health. Additionally, individuals are required to keep up to date with latest health guidelines about the virus, which may be confusing in an age of social-media disinformation and shifting guidelines. To tackle these factors, we developed Elena+, a smartphone-based and conversational agent (CA) delivered pandemic lifestyle care intervention. Methods: Elena+ utilizes varied intervention components to deliver a psychoeducation-focused coaching program on the topics of: COVID-19 information, physical activity, mental health (anxiety, loneliness, mental resources), sleep and diet and nutrition. Over 43 subtopics, a CA guides individuals through content and tracks progress over time, such as changes in health outcome assessments per topic, alongside user-set behavioral intentions and user-reported actual behaviors. Ratings of the usage experience, social demographics and the user profile are also captured. Elena+ is available for public download on iOS and Android devices in English, European Spanish and Latin American Spanish with future languages and launch countries planned, and no limits on planned recruitment. Panel data methods will be used to track user progress over time in subsequent analyses. The Elena+ intervention is open-source under the Apache 2 license (MobileCoach software) and the Creative Commons 4.0 license CC BY-NC-SA (intervention logic and content), allowing future collaborations; such as cultural adaptions, integration of new sensor-related features or the development of new topics. Discussion: Digital health applications offer a low-cost and scalable route to meet challenges to public health. As Elena+ was developed by an international and interdisciplinary team in a short time frame to meet the COVID-19 pandemic, empirical data are required to discern how effective such solutions can be in meeting real world, emergent health crises. Additionally, clustering Elena+ users based on characteristics and usage behaviors could help public health practitioners understand how population-level digital health interventions can reach at-risk and sub-populations.
Frontiers in Public Health, 2021

Intro: As smartphone usage becomes increasingly prevalent in the workplace, the physical and psychological implications of this behavior warrant consideration. Recent research has investigated associations between workplace smartphone use and fatigue and boredom, yet findings are not conclusive. Methods: To build off recent efforts, we applied an ensemble machine learning model on a previously published dataset of N = 83 graduate students in the Netherlands to predict work boredom and fatigue from passively collected smartphone app use information. Using time-based feature engineering and lagged variations of the data to train, validate, and test idiographic models, we evaluated the efficacy of a lagged-ensemble predictive paradigm on sparse temporal data. Moreover, we probed the relative importance of both derived app use variables and lags within this predictive framework. Results: The ability to predict fatigue and boredom trajectories from app use information was heterogeneous and highly person-specific. Idiographic modeling reflected moderate to high correlative capacity (r > 0.4) in 47% of participants for fatigue and 24% for boredom, with better overall performance in the fatigue prediction task. App use relating to duration, communication, and patterns of use frequency were among the most important features driving predictions across lags, with longer lags contributing more heavily to final ensemble predictions compared with shorter ones. Conclusion: A lag-specific ensemble predictive paradigm is a promising approach to leveraging high-dimensional app use behavioral data for the prediction of work fatigue and boredom. Future research will benefit from evaluating associations on densely collected data across longer time scales.
Computers in Human Behavior, 2021

Objectives: Machine learning models are a promising, yet underutilized tool within the mindfulness field. Accordingly, this work aimed to provide a practical introduction to key machine learning concepts through an illustrative investigation of the association between at-home mindfulness exercise compliance and stress reduction. To further interrogate the currently inconclusive nature of the compliance-outcome association within the mindfulness literature, the illustrative example leveraged a suite of machine learning techniques to highlight the unique affordances and perspectives of the predictive framework. Methods: Foundational information regarding facets of the machine learning analytical process, including model types, data preprocessing, feature engineering, validation, performance evaluation, and model introspection, was presented. With emphasis on providing details and justifications regarding modeling decisions along the way, the work systematically applied these introduced concepts to a real-world data example. This permitted an opportunity to build, introspect, and derive insight from a model tasked to explore dynamics underlying patient compliance to mindfulness exercises within a web-based delivery setting. Results: The constructed machine learning models suggested a moderate correlation of compliance with post-intervention reliable change in stress (r = 0.349 ± 0.018). Model introspection tools further revealed that a combination of both high consistency and high overall average compliance predicts a trend toward greater reduction in self-reported stress. Conclusions: Results of the illustrative study suggested that compliance, in pattern and absolute magnitude, is a significant contributor to online mindfulness therapy outcomes. Moreover, modeling efforts implicate machine learning as a uniquely beneficial paradigm with which to explore nuanced questions in the mindfulness research space.
Mindfulness, 2021

Researchers have held a long-standing debate regarding the validity of discrete emotions versus global affect. The current manuscript tries to integrate these perspectives by explicitly examining the structures of state emotions and trait affect across time. Across three samples (sample 1: N = 176 Unites States undergraduates in a 50 day daily diary study, total observations = 7,504; sample 2: N = 2,104 in a 30 day daily diary study within a community sample in Germany; total observations = 28,090; sample 3: N = 245, ecological momentary assessment study within the United States from an outpatient psychiatry clinic completing five measurements per day for 21 days; total observations = 29,950), participants completed the Positive and Negative Affect Schedule. An exploratory multilevel factor analysis in sample 1 allowed for the simultaneous estimation of state factors (i.e., within-person factor analysis) and trait factors (i.e., between-person factor analysis). Confirmatory multilevel factor models examined the generalizability of the multilevel factor solutions to samples 2 and 3. Across all samples, the results suggested strong support for a two-factor solution for trait affect and a seven-factor solution for state emotion. Taken together, these results suggest that positive affect and negative affect can be used to describe differences across people, but at least seven differentiated emotions are experienced within persons across time.
Emotion, 2021

Background: The number of smartphone apps that focus on the prevention, diagnosis, and treatment of depression is increasing. A promising approach to increase the effectiveness of the apps while reducing the individual’s burden is the use of just-in-time adaptive intervention (JITAI) mechanisms. JITAIs are designed to improve the effectiveness of the intervention and reduce the burden on the person using the intervention by providing the right type of support at the right time. The right type of support and the right time are determined by measuring the state of vulnerability and the state of receptivity, respectively. Objective: The aim of this study is to systematically assess the use of JITAI mechanisms in popular apps for individuals with depression. Methods: We systematically searched for apps addressing depression in the Apple App Store and Google Play Store, as well as in curated lists from the Anxiety and Depression Association of America, the United Kingdom National Health Service, and the American Psychological Association in August 2020. The relevant apps were ranked according to the number of reviews (Apple App Store) or downloads (Google Play Store). For each app, 2 authors separately reviewed all publications concerning the app found within scientific databases (PubMed, Cochrane Register of Controlled Trials, PsycINFO, Google Scholar, IEEE Xplore, Web of Science, ACM Portal, and Science Direct), publications cited on the app’s website, information on the app’s website, and the app itself. All types of measurements (eg, open questions, closed questions, and device analytics) found in the apps were recorded and reviewed. Results: None of the 28 reviewed apps used JITAI mechanisms to tailor content to situations, states, or individuals. Of the 28 apps, 3 (11%) did not use any measurements, 20 (71%) exclusively used self-reports that were insufficient to leverage the full potential of the JITAIs, and the 5 (18%) apps using self-reports and passive measurements used them as progress or task indicators only. Although 34% (2368) of the reviewed publications investigated the effectiveness of the apps and 21% (1468) investigated their efficacy, no publication mentioned or evaluated JITAI mechanisms. Conclusions: Promising JITAI mechanisms have not yet been translated into mainstream depression apps. Although the wide range of passive measurements available from smartphones were rarely used, self-reported outcomes were used by 71% (2028) of the apps. However, in both cases, the measured outcomes were not used to tailor content and timing along a state of vulnerability or receptivity. Owing to this lack of tailoring to individual, state, or situation, we argue that the apps cannot be considered JITAIs. The lack of publications investigating whether JITAI mechanisms lead to an increase in the effectiveness or efficacy of the apps highlights the need for further research, especially in real-world apps
Journal of Medical Internet Research (JMIR), 2021

Introduction: Online social networking data (SN) is a contextually and temporally rich data stream that has shown promise in the prediction of suicidal thought and behavior. Despite the clear advantages of this digital medium, predictive modeling of acute suicidal ideation (SI) currently remains underdeveloped. SN data, in conjunction with robust machine learning algorithms, may offer a promising way forward. Methods: We applied an ensemble machine learning model on a previously published dataset of adolescents on Instagram with a prior history of lifetime SI (N = 52) to predict SI within the past month. Using predictors that capture language use and activity within this SN, we evaluated the performance of our out-of-sample, cross-validated model against previous efforts and leveraged a model explainer to further probe relative predictor importance and subject-level phenomenology. Results: Linguistic and SN data predicted acute SI with an accuracy of 0.702 (sensitivity = 0.769, specificity = 0.654, AUC = 0.775). Model introspection showed a higher proportion of SN-derived predictors with substantial impact on prediction compared with linguistic predictors from structured interviews. Further analysis of subject-specific predictor importance uncovered potentially informative trends for future acute SI risk prediction. Conclusion: Application of ensemble learning methodologies to SN data for the prediction of acute SI may mitigate the complexities and modeling challenges of SI that exist within these time scales. Future work is needed on larger, more heterogeneous populations to fine-tune digital biomarkers and more robustly test external validity.
Internet Interventions, 2021

Background: Online guided self-help may be an effective and scalable intervention for symptoms of generalized anxiety disorder (GAD) among university students in India. Methods: Based on an online screen for GAD administered at four Indian universities, 222 students classified as having clinical (DSM-5 criteria) or subthreshold (GAD-Q-IV score ≥ 5.7) GAD were randomly assigned to receive either three months of guided self-help cognitive-behavioral therapy (n = 117) or a waitlist control condition (n = 105). Results: Guided self-help participants recorded high program usage on average across all participants enrolled (M = 9.99 hours on the platform; SD = 20.87). Intent-to-treat analyses indicated that participants in the guided self-help condition experienced significantly greater reductions than participants in the waitlist condition on GAD symptom severity (d = -0.40), worry (d = -0.43), and depressive symptoms (d = -0.53). No usage variables predicted symptom change in the guided self-help condition. Participants on average reported that the program was moderately helpful, and a majority (82.1%) said they would recommend the program to a friend. Conclusions: Guided self-help appears to be a feasible and efficacious intervention for university students in India who meet clinical or subthreshold GAD criteria.
Psychotherapy, 2021

Post-traumatic stress disorder (PTSD) is characterized by complex, heterogeneous symptomology, thus detection outside traditional clinical contexts is difficult. Fortunately, advances in mobile technology, passive sensing, and analytics offer promising avenues for research and development. The present study examined the ability to utilize Global Positioning System (GPS) data, derived passively from a smartphone across seven days, to detect PTSD diagnostic status among a cohort (N = 185) of high-risk, previously traumatized women. Using daily time spent away and maximum distance traveled from home as a basis for model feature engineering, the results suggested that diagnostic group status can be predicted out-of-fold with high performance (AUC = 0.816, balanced sensitivity = 0.743, balanced specificity = 0.8, balanced accuracy = 0.771). Results further implicate the potential utility of GPS information as a digital biomarker of the PTSD behavioral repertoire. Future PTSD research will benefit from application of GPS data within larger, more diverse populations.
Scientific Reports, 2021

Background: Recent studies have demonstrated that passive smartphone and wearable sensor data collected throughout daily life can predict anxiety symptoms cross-sectionally. However, to date, no research has demonstrated the capacity for these digital biomarkers to predict long-term prognosis. Methods: We utilized deep learning models based on wearable sensor technology to predict long-term (17–18year) deterioration in generalized anxiety disorder and panic disorder symptoms from actigraphy data on daytime movement and nighttime sleeping patterns. As part of Midlife in the United States (MIDUS), a national longitudinal study of health and well-being, subjects (N = 265) (i) completed a phone-based interview that assessed generalized anxiety disorder and panic disorder symptoms at enrollment, (ii) participated in a one-week actigraphy study 9–14 years later, and (iii) completed a long-term follow-up, phone-based interview to quantify generalized anxiety disorder and panic disorder symptoms 17–18 years from initial enrollment. A deep auto- encoder paired with a multi-layered ensemble deep learning model was leveraged to predict whether participants experienced increased anxiety disorder symptoms across this 17–18 year period. Results: Out-of-sample cross-validated results suggested that wearable movement data could significantly predict which individuals would experience symptom deterioration (AUC = 0.696, CI [0.598, 0.793], 84.6% sensitivity, 52.7% specificity, balanced accuracy = 68.7%). Conclusions: Passive wearable actigraphy data could be utilized to predict long-term deterioration of anxiety disorder symptoms. Future studies should examine whether these methods could be implemented to prevent deterioration of anxiety disorder symptoms.
Journal of Affective Disorders, 2021

Objectives: Using two intensive longitudinal data sets with different timescales (90 minutes, daily), we examined emotion network density, a metric of emotional inflexibility, as a predictor of clinical-level anxiety and depression. Design: Mobile-based intensive longitudinal assessments. Methods: 119 participants (61 anxious and depressed, 58 healthy controls) completed ecological momentary assessment (EMA) to rate a variety of negative (NE) and positive emotions (PE) 9 times per day for 8 days using a mobile phone application. 169 participants (97 anxious and depressed and 72 healthy controls) completed an online daily diary on their NE and PE for 50 days. Multilevel vector autoregressive models were run to compute NE and PE network densities in each data set. Results: In the EMA data set, both NE and PE network densities significantly predicted participants’ diagnostic status above and beyond demographics and the mean and standard deviation of NE and PE. Greater NE network density and lower PE network density were associated with anxiety and depression diagnoses. In the daily diary data set, NE and PE network densities did not significantly predict the diagnostic status. Conclusions: Greater inflexibility of NE and lower inflexibility of PE, indexed by emotion network density, are potential clinical markers of anxiety and depressive disorders when assessed at intra-daily levels as opposed to daily levels. Considering emotion network density, as well as the mean level and variability of emotions in daily life, may contribute to diagnostic prediction of anxiety and depressive disorders.
British Journal of Clinical Psychology, 2021

Background: Generalized anxiety disorder (GAD) and major depressive disorder (MDD) are highly prevalent and impairing problems, but frequently go undetected, leading to substantial treatment delays. Electronic health records (EHRs) collect a great deal of biometric markers and patient characteristics that could foster the detection of GAD and MDD in primary care settings. Methods: We approached the problem of predicting MDD and GAD using a novel machine learning pipeline to re-analyze data from an observational study. The pipeline constitutes an ensemble of algorithmically distinct machine learning methods, including deep learning. A sample of 4,184 undergraduate students completed the study, undergoing a general health screening and completing a psychiatric assessment for MDD and GAD. After explicitly excluding all psychiatric information, 59 biomedical and demographic features from the general health survey in addition to a set of engineered features were used for model training. Results: We assessed the model’s performance on a held-out test set and found an AUC of 0.73 (sensitivity: 0.66, specificity: 0.7) and 0.67 (sensitivity: 0.55, specificity: 0.7) for GAD, and MDD, respectively. Additionally, we used advanced techniques (SHAP values) to illuminate which features had the greatest impact on prediction for each disease. The top predictive features for MDD were being satisfied with living conditions and having public health insurance. The top predictive features for GAD were vaccinations being up to date and marijuana use. Conclusions: Our results indicate moderate predictive performance for the application of machine learning methods in detection of GAD and MDD based on EHR data. By identifying biomarkers of GAD and MDD, these results may be used in future research to aid in the early detection of MDD and GAD.
Scientific Reports, 2020

While digital psychiatric interventions reduce treatment barriers, not all persons benefit from this type of treatment. Research is needed to preemptively identify who is likely to benefit from these digital treatments in order to redirect those people to a higher level of care. The current manuscript used an ensemble of machine learning methods to predict changes in major depressive and generalized anxiety disorder symptoms from pre to 9-month follow-up in a randomized controlled trial of a transdiagnostic digital intervention based on participants’ (N=632) pre-treatment data. The results suggested that baseline characteristics could accurately predict changes in depressive symptoms in both treatment groups (r=0.482, 95% CI[0.394, 0.561]; r=0.477, 95% CI[0.385, 0.560]) and anxiety symptoms in both treatment groups (r=0.569, 95% CI[0.491, 0.638]; r=0.548, 95% CI[0.464, 0.622]). These results suggest that machine learning models are capable of preemptively predicting a person’s responsiveness to digital treatments, which would enable personalized decision-making about which persons should be directed towards standalone digital interventions or towards blended stepped-care.
Psychiatry Research, 2020

Background: Sleep disruption is a common precursor to deterioration and relapse in people living with psychotic disorders. Understanding the temporal relationship between sleep and psychopathology is important for identifying and developing interventions which target key variables that contribute to relapse. Methods: We used a purpose-built digital platform to sample self-reported sleep and psychopathology variables over one year, in thirty-six individuals with schizophrenia. Once-daily measures of sleep duration and sleep quality, and fluctuations in psychopathology (positive and negative affect, cognition, and psychotic symptoms) were captured. We examined the temporal relationship between these variables using the Differential Time-Varying Effect (DTVEM) hybrid exploratory-confirmatory model. Results: Poorer sleep quality and shorter sleep duration maximally predicted deterioration in psychosis symptoms over the subsequent 1 – 8 and 1-12 days, respectively. These relationships were also mediated by negative affect and cognitive symptoms. Psychopathology variables also predicted sleep quality, but not sleep duration, and the effect sizes were smaller and of shorter lag duration. Conclusions: Reduced sleep duration and poorer sleep quality anticipate exacerbation of psychotic symptoms by approximately 1-2 weeks, and negative affect and cognitive symptoms mediate this relationship. We also observed a reciprocal relationship, that was of shorter duration and smaller magnitude. Sleep disturbance may play a causal role in symptom exacerbation and relapse, and represents an important and tractable target for intervention. It warrants greater attention as an early warning sign of deterioration, and low-burden, user-friendly digital tools may play a role in its early detection.
Journal of Medical Internet Research (JMIR): Formative Research, 2020

Background: Chatbots could be a scalable solution that provides an interactive means of engaging users in behavioral health interventions driven by artificial intelligence. Although some chatbots have shown promising early efficacy results, there is limited information about how people use these chatbots. Understanding the usage patterns of chatbots for depression represents a crucial step toward improving chatbot design and providing information about the strengths and limitations of the chatbots. Objective: This study aims to understand how users engage and are redirected through a chatbot for depression (Tess) to provide design recommendations. Methods: Interactions of 354 users with the Tess depression modules were analyzed to understand chatbot usage across and within modules. Descriptive statistics were used to analyze participant flow through each depression module, including characters per message, completion rate, and time spent per module. Slide plots were also used to analyze the flow across and within modules. Results: Users sent a total of 6220 messages, with a total of 86,298 characters, and, on average, they engaged with Tess depression modules for 46 days. There was large heterogeneity in user engagement across different modules, which appeared to be affected by the length, complexity, content, and style of questions within the modules and the routing between modules. Conclusions: Overall, participants engaged with Tess; however, there was a heterogeneous usage pattern because of varying module designs. Major implications for future chatbot design and evaluation are discussed in the paper.
Journal of Medical Internet Research (JMIR): Formative Research, 2020

Introduction: Most people with psychiatric illnesses do not receive treatment for almost a decade after disorder onset. Online mental health screens reflect one mechanism designed to shorten this lag in help-seeking, yet there has been limited research on the effectiveness of screening tools in naturalistic settings. Material and methods: We examined a cohort of persons directed to a mental health screening tool via the Bing search engine (n = 126,060). We evaluated the impact of tool content on later searches for mental health selfreferences, self-diagnosis, care seeking, psychoactive medications, suicidal ideation, and suicidal intent. Website characteristics were evaluated by pairs of independent raters to ascertain screen type and content. These included the presence/absence of a suggestive diagnosis, a message on interpretability, as well as referrals to digital treatments, in-person treatments, and crisis services. Results: Using machine learning models, the results suggested that screen content predicted later searches with mental health self-references (AUC = 0.73), mental health self-diagnosis (AUC = 0.69), mental health care seeking (AUC = 0.61), psychoactive medications (AUC = 0.55), suicidal ideation (AUC = 0.58), and suicidal intent (AUC = 0.60). Cox-proportional hazards models suggested individuals utilizing tools with in-person care referral were significantly more likely to subsequently search for methods to actively end their life (HR = 1.727,p = 0.007). Discussion: Online screens may influence help-seeking behavior, suicidal ideation, and suicidal intent. Websites with referrals to in-person treatments could put persons at greater risk of active suicidal intent. Further evaluation using large-scale randomized controlled trials is needed.
Journal of Psychiatric Research, 2020

Introduction: Generalized anxiety disorder (GAD) is prevalent among college students. Smartphone-based interventions may be a low-cost treatment method. Method: College students with self-reported GAD were randomized to receive smartphone-based guided self-help (n = 50), or no treatment (n = 50). Post-treatment and six-month follow-up outcomes included the Depression Anxiety Stress Scales-Short Form Stress Subscale (DASS Stress), the Penn State Worry Questionnaire (PSWQ-11), and the State-Trait Anxiety Inventory-Trait (STAI-T), as well as diagnostic status assessed by the GAD-Questionnaire, 4th edition. Results: From pre- to post-treatment, participants who received guided self-help (vs. no treatment) experienced significantly greater reductions on the DASS Stress (d = -0.408) and a greater probability of remission from GAD (d = -0.445). There was no significant between-group difference in change on the PSWQ-11 (d = -0.208) or STAI-T (d = -0.114). From post to six-month follow-up there was no significant loss of gains on DASS Stress scores (d = -0.141) and of those who had remitted, 78.6% remained remitted. Yet rates of remitted participants no longer differed significantly between conditions at follow-up (d = -0.229). Conclusion: Smartphone-based interventions may be efficacious in treating some aspects of GAD. Methods for improving symptom reduction and long-term outcome are discussed.
Psychotherapy Research, 2020

Prior research has recently shown that passively collected sensor data collected within the contexts of persons daily lives via smartphones and wearable sensors can distinguish those with major depressive disorder (MDD) from controls, predict MDD severity, and predict changes in MDD severity across days and weeks. Nevertheless, very little research has examined predicting depressed mood within a day, which is essential given the large amount of variation occurring within days. The current study utilized passively collected sensor data collected from a smartphone application to future depressed mood from hour-to-hour in an ecological momentary assessment study in a sample reporting clinical levels of depression (N = 31). Using a combination of nomothetic and idiographically-weighted machine learning models, the results suggest that depressed mood can be accurately predicted from hour to hour with an average correlation between out of sample predicted depressed mood levels and observed depressed mood of 0.587, CI [0.552, 0.621]. This suggests that passively collected smartphone data can accurately predict future depressed mood among a sample reporting clinical levels of depression. If replicated in other samples, this modeling framework may allow just-in-time adaptive interventions to treat depression as it changes in the context of daily life.
Sensors, 2020

Researchers have increasingly begun to use consumer wearables or wrist-worn smartwatches and fitness monitors for measurement of cardiovascular psychophysiological processes related to mental and physical health outcomes. These devices have strong appeal because they allow for continuous, scalable, unobtrusive, and ecologically valid data collection of cardiac activity in “big data” studies. However, replicability and reproducibility may be hampered moving forward due to the lack of standardization of data collection and processing procedures and inconsistent reporting of technological factors (e.g., device type, firmware versions, sampling rate), biobehavioral variables (e.g., body mass index, wrist dominance and circumference), and participant demographic characteristics, such as skin tone, that may influence heart rate measurement. These limitations introduce unnecessary noise into measurement, which can cloud interpretation and generalizability of findings. This paper provides a brief overview of research using commercial wearable devices to measure heart rate, reviews literature on device accuracy, and outlines the challenges that non-standardized reporting pose for the field. We also discuss study design, technological, biobehavioral, and demographic factors that can impact the accuracy of the passive sensing of heart rate measurements, and provide guidelines and corresponding handouts for future study data collection and design, data cleaning and processing, analysis, and reporting that may help ameliorate some of these barriers and inconsistencies in the literature.
npj Digital Medicine, 2020

Background: The coronavirus disease (COVID-19) has led to dramatic changes worldwide in people’s everyday lives. To combat the pandemic, many governments have implemented social distancing, quarantine, and stay-at-home orders. There is limited research on the impact of such extreme measures on mental health. Objective: The goal of this study was to examine whether stay-at-home orders produced differential changes in mental health symptoms using internet search queries on a national scale. Methods: In the United States, individual states vary in their adoption of measures to reduce the spread of COVID-19; as of March 23, 2020, 11 of the 50 states had issued stay-at-home orders. The staggered rollout of stay-at-home measures across the United States allows us to investigate whether these measures impact mental health by exploring variations in mental health search queries across the states. This paper examines the changes in mental health search queries on Google between March 16-23, 2020, across each state and Washington, DC. Specifically, this paper examines differential changes in mental health searches based on patterns of search activity following issuance of stay-at-home orders in these states compared to all other states. The participants were all the people who searched mental health terms in Google between March 16-23. Between March 16-23, 11 states underwent stay-at-home orders to prevent the transmission of COVID-19. Outcomes included search terms measuring anxiety, depression, obsessive-compulsive, negative thoughts, irritability, fatigue, anhedonia, concentration, insomnia, and suicidal ideation. Results: Analyzing over 10 million search queries using generalized additive mixed models, the results suggested that the implementation of stay-at-home orders are associated with a significant flattening of the curve for searches for suicidal ideation, anxiety, negative thoughts, and sleep disturbances, with the most prominent flattening associated with suicidal ideation and anxiety. Conclusions: These results suggest that, despite decreased social contact, mental health search queries increased rapidly prior to the issuance of stay-at-home orders, and these changes dissipated following the announcement and enactment of these orders. Although more research is needed to examine sustained effects, these results suggest mental health symptoms were associated with an immediate leveling off following the issuance of stay-at-home orders.
JMIR Mental Health, 2020

Background: Social anxiety disorder is a highly prevalent and burdensome condition. Persons with social anxiety frequently avoid seeking physician support and rarely receive treatment. Social anxiety symptoms are frequently under-reported and under-recognized, creating a barrier to the accurate assessment of these symptoms. Consequently, more research is needed to identify passive biomarkers of social anxiety symptom severity. Digital phenotyping, the use of passive sensor data to inform healthcare decisions, offers a possible method of addressing this assessment barrier. Objective: To determine whether passive sensor data acquired from smartphone data can accurately predict social anxiety symptom severity using a publicly available dataset. Methods: In this study, participants (N = 59) completed self-report assessments of their social anxiety symptom severity, depressive symptom severity, positive affect, and negative affect. Next, participants installed an application which passively collected data about their movement (accelerometers) and social contact (incoming and outgoing calls and texts) across two weeks. Next this passive sensor data was used to form digital biomarkers which were paired with machine learning models to predict participants’ social anxiety symptom severity. Results: The results suggested that this passive sensor data could be utilized to accurately predict participants’ social anxiety symptom severity (r = 0.702 between predicted and observed symptom severity), and demonstrated discriminant validity between depression, negative affect, and positive affect. Conclusions: These results suggest that smartphone sensor data may be utilized to accurately detect social anxiety symptom severity and discriminate social anxiety symptom severity from depressive symptoms, negative affect, and positive affect.
Journal of Medicial Internet Research (JMIR), 2020

The application of digital technology to psychiatry research is rapidly leading to new discoveries and capabilities in the field of mobile health. However, the increase in opportunities to passively collect vast amounts of detailed information on study participants coupled with advances in statistical techniques that enable machine learning models to process such information has raised novel ethical dilemmas regarding researchers’ duties to: (i) monitor adverse events and intervene accordingly; (ii) obtain fully informed, voluntary consent; (iii) protect the privacy of participants; and (iv) increase the transparency of powerful, machine learning models to ensure they can be applied ethically and fairly in psychiatric care. This review highlights emerging ethical challenges and unresolved ethical questions in mobile health research and provides recommendations on how mobile health researchers can address these issues in practice. Ultimately, the hope is that this review will facilitate continued discussion on how to achieve best practice in mobile health research within psychiatry.
Bulletin of the World Health Organization, 2020

Our field has come a long way in establishing cognitive behavioral therapy as the empirically-supported treatment of choice for a wide range of mental and behavioral health problems. Nevertheless, most individuals with mental disorders do not receive any care at all, and those who do often have difficulty accessing care that is consistently high in quality. Addressing these issues is complex and costly and thus progress has been slow. We are entering an exciting stage in which emerging technologies might offer novel solutions to the treatment gap. This paper discusses a number of technology-enabled solutions to our field’s challenges, including internet-based and smartphone-based cognitive behavioral therapy. Nevertheless, we must remain attentive to potential pitfalls of these emerging technologies. The manuscript incorporates suggestions for how the field may approach these potential pitfalls and provides a vision for how we might develop powerful, scalable, precisely timed, personalized interventions to enhance global mental health.
Behavior Therapy, 2019

Persons living with HIV (PLWH) report experiencing disproportionally severe and chronic pain and worry. However, no objective biomarkers of these subjective experiences have been developed. To address the lack of objective measures and assist in treatment planning, the current study examined whether digital biomarkers of pain severity, pain chronicity, and worry could be developed using passive wearable sensors continuously monitoring movement. Results suggest that digital biomarkers can predict pain severity (r(35) = 0.690), pain chronicity (74.63% accuracy), and worry severity (r(65) = 0.642) with high precision, suggesting that objective digital biomarkers alone accurately capture internal symptom experiences in PLWH.
The British Journal of Psychiatry, 2019

Development of digital biomarkers holds promise for enabling scalable, time-sensitive, and cost-effective strategies to monitor symptom severity among those with major depressive disorder. The current study examined the use of passive movement and light data from wearable devices to assess depression severity in 15 patients with major depressive disorder. Using over one week of movement data, we were able to significantly assess depression severity with high precision for self-reported (r = 0.855, 95% CI 0.610 to 0.950, p = 4.95x10-5) and clinician-rated (r = 0.604, 95% CI 0.133 to 0.894, p = .017) symptom severity. Pending replication, the present data suggests that the use of passive wearable sensors to inform healthcare decisions holds considerable promise.
Journal of Nervous and Mental Disease, 2019

Current approaches to psychiatric assessment are resource-intensive, requiring time-consuming evaluation by a trained clinician. Development of digital biomarkers holds promise for enabling scalable, time-sensitive, and cost-effective assessment of both psychiatric diagnosis and symptom change. The present study aimed to identify robust digital biomarkers of diagnostic status and changes in symptom severity over ~2 weeks, through re-analysis of public-use actigraphy data collected in patients with major depressive or bipolar disorder and healthy controls. Results suggest that participants’ diagnostic group status (i.e., mood disorder, Q1 control) can be predicted with a high degree of accuracy (predicted correctly 89% of the time, kappa = 0.773), using features extracted from actigraphy data alone. Results also suggest that actigraphy data can be used to predict symptom change across ~2 weeks (r = 0.782, p = 1.04e-05). Through inclusion of digital biomarkers in our statistical model, which are generalizable to new samples, the results may be replicated by other research groups in order to validate and extend this work.
Nature Partner Journal (npj) Digital Medicine, 2019

Objective: Although recent research has begun to examine the impact of elevated anxiety on evolutionary fitness, no prior research has examined anxiety across a continuum. Such research is important as the effect of traits across a continuum on fertility hold important implications for the levels and distribution of the traits in later generations. Method: In a three-generational sample (N = 2,657) the linear and quadratic relationship between anxiety and the number of children, grandchildren, and great-grandchildren 15 years later was examined. Results: The findings suggested that anxiety had a positive quadratic relationship with the number of children, grandchildren, and great-grandchildren 15 years later. These relationships were not significantly moderated by sex. Moreover, most of the variance between anxiety and the number of great-grandchildren was explained by anxiety’s influence on the number of children and grandchildren, as opposed to anxiety having an independent direct impact on the number of great-grandchildren. Conclusion: These findings suggest that extreme values from the mean anxiety are associated with increased evolutionary fitness within the modern environment.
Journal of Psychiatric Research, 2018

Background: The Contrast Avoidance Model (CAM) suggests that worry increases and sustains negative emotion to prevent a negative emotional contrast (sharp upward shift in negative emotion) and to increase the probability of a positive contrast (shift toward positive emotion). Method: In Study 1, we experimentally validated momentary assessment items (N=25). In Study 2, participants with generalized anxiety disorder (GAD) (N=31) and controls (N=37) were prompted once per hour regarding their worry, thought valence, and arousal 10x/day for 8 days. Results: Higher worry duration, negative thought valence, and uncontrollable train of thoughts predicted feeling more keyed up concurrently and sustained anxious activation one hour later. More worry, feeling keyed up, and uncontrollable train of thoughts predicted lower likelihood of a negative emotional contrast in thought valence, and higher likelihood of a positive emotional contrast in thought valence one hour later. Conclusions: Findings support the prospective ecological validity of CAM.
Clinical Psychological Science, 2018

With the recent growth in intensive longitudinal designs and corresponding demand for methods to analyze such data, there has never been a more pressing need for user-friendly analytic tools that can identify and estimate optimal time lags in intensive longitudinal data. Available standard exploratory methods to identify optimal time lags within univariate and multivariate multiple subject time series are greatly under-powered at the group (i.e., population) level. We describe a hybrid exploratory-confirmatory tool, referred to herein as the Differential Time-Varying Effect Model (DTVEM), which features a convenient user-accessible function to identify optimal time lags and estimate these lags within a state-space framework. Data from an empirical ecological momentary assessment study are used to demonstrate the utility of the proposed tool in identifying the optimal time lag for studying the linkages between nervousness and heart rate in a group of undergraduate students. Using a simulation study, we illustrate the effectiveness of DTVEM in identifying optimal lag structures in multiple-subject, time series data with missingness, as well as its strengths and limitations as a hybrid exploratory-confirmatory approach compared to other existing approaches.
Behavior Research Methods, 2018

Not only do anxiety and depression diagnoses tend to co-occur, but their symptoms are highly correlated. Although a plethora of research has examined longitudinal associations between anxiety and depression, these data have not yet been effectively synthesized. To address this need, the current study undertook a systematic review and meta-analysis of 66 studies involving 88,336 persons examining the prospective relationship between anxiety and depression at both symptom and disorder levels. Using mixed-effect models, results suggested that all types of anxiety symptoms predicted later depressive symptoms (r = .34), and all types of depressive symptoms predicted later anxiety symptoms (r = .31). Although anxiety symptoms more strongly predicted depressive symptoms than vice versa, the difference in effect size for this analysis was very small and likely not clinically meaningful. Additionally, all types of diagnosed anxiety disorders predicted all types of later depressive disorders (OR = 2.77), and all depressive disorders predicted later anxiety disorders (OR = 2.73). Most anxiety and depressive disorders predicted each other with similar degrees of strength, but depressive disorders more strongly predicted social anxiety disorder (OR = 6.05) and specific phobia (OR = 2.93) than vice versa. Contrary to conclusions of prior reviews, our findings suggest that depressive disorders may be prodromes for social and specific phobia, whereas other anxiety and depressive disorders are bidirectional risk factors for one another.
Psychological Bulletin, 2017

Background. Prior research has shown that anxiety symptoms predict later depression symptoms following bereavement. Nevertheless, no research has investigated mechanisms of the temporal relationship between anxiety and later depressive symptoms or examined the impact of depressive symptoms on later anxiety symptoms following bereavement. Methods. The current study examined perceived emotional social support as a possible mediator between anxiety and depressive symptoms in a bereaved sample of older adults (N =250). Anxiety and depressive symptoms were measured at Wave 1 (immediately after bereavement), social support was measured at Wave 2 (18 months after bereavement), and anxiety and depressive symptoms were also measured at Wave 3 (48 months after bereavement). Results. Using Bayesian structural equation models, when controlling for baseline depression, anxiety symptoms significantly positively predicted depressive symptoms 48 months later, Further, perceived emotional social support significantly mediated the relationship between anxiety symptoms and later depressive symptoms, such that anxiety symptoms significantly negatively predicted later emotional social support, and emotional social support significantly negatively predicted later depressive symptoms. Also, when controlling for baseline anxiety, depressive symptoms positively predicted anxiety symptoms 48 months later. However, low emotional social support failed to mediate this relationship. Conclusions. Low perceived emotional social support may be a mechanism by which anxiety symptoms predict depressive symptoms 48 months later for bereaved individuals.
Journal of Affective Disorders, 2017

This study sought to evaluate the current evolutionary adaptiveness of psychopathology by examining whether these disorders impact the quantity of offspring or the quality of the parent–child relationship across the life span. Using the National Comorbidity Survey, this study examined whether DSM–III–R anxiety, posttraumatic stress, depressive, bipolar, substance use, antisocial, and psychosis disorders predicted later fertility and the quality of parent–child relationships across the life span in a national sample (N = 8,098). Using latent variable and varying coefficient models, the results suggested that anxiety in males and bipolar pathology in males and females were associated with increased fertility at younger ages. The results suggested almost all other psychopathology was associated with decreased fertility in middle to late adulthood. The results further suggested that all types of psychopathology had negative impacts on the parent–child relationship quality (except for antisocial pathology in males). Nevertheless, for all disorders, the impact of psychopathology on both fertility and the parent–child relationship quality was affected by the age of the participant. The results also showed that anxiety pathology is associated with a high-quantity, low-quality parenting strategy followed by a low-quantity, low-quality parenting strategy. Further, the results suggest that bipolar pathology is associated with an early high-quantity and a continued low-quality parenting strategy. Posttraumatic stress, depression, substance use, antisocial personality, and psychosis pathology are each associated with a low-quantity, low-quality parenting strategy, particularly in mid to late adulthood. These findings suggest that the evolutionary impact of psychopathology depends on the developmental context.
Journal of Abnormal Psychology, 2016

Recent Publications

More Publications

(2021). The Language of the Times: Using the COVID-19 Pandemic to Assess the Influence of News Affect on Online Mental Health-Related Search Behavior across the United States. Journal of Medical Internet Research (JMIR).

(2021). Elena+ Care for COVID-19, a Pandemic Lifestyle Care Intervention: Intervention Design and Study Protocol. Frontiers in Public Health.

PDF Source Document

(2021). Using smartphone app use and lagged-ensemble machine learning for the prediction of work fatigue and boredom. Computers in Human Behavior.

PDF Source Document

(2021). Just-in-time Adaptive Mechanisms of Popular Mobile Applications for Individuals with Depression: Systematic Review. Journal of Medical Internet Research (JMIR).

PDF Source Document

(2021). Predicting Acute Suicidal Ideation on Instagram Using Ensemble Machine Learning Models. Internet Interventions.

PDF Source Document

(2021). A Randomized Controlled Feasibility Trial of Internet-Delivered Guided Self-Help for Generalized Anxiety Disorder (GAD) Among University Students in India. Psychotherapy.

PDF Source Document

Recent Posts

Overview This post documents reproducible code accompanying the manuscript draft “Digital Biomarkers of Mood Disorders and Symptom Change” by Nicholas C. Jacobson, Hilary M. Weingardenm and Sabine Wilhelm (published in Nature Partner Journal (npj): Digital Medicine). This code uses machine learning to predict the diagnostic status and depressive symptom change in a a group of 23 patients with bipolar disorder or major depressive disorder and 32 non-disordered controls using actigraphy data.

CONTINUE READING

Loading in the Dataset This code will illustrate the R package (DTVEM) with simulated data available in the DTVEM package. Click here to download and install the DTVEM package. First load the DTVEM package. library(DTVEM) Next load the simulated data included in the DTVEM package, called exampledat1. data(exampledat1) Get a look at the file structure. head(exampledat1) ## Time X ID ## 1 1 -1.076422 1 ## 2 2 -1.904713 1 ## 3 3 1.

CONTINUE READING

Projects

Mood Triggers

This describes a smartphone application to help users figure out triggers of their anxiety and depression in daily life.

The Differential Time Varying Effect Model

This project describes a novel technique entitled the Differential Time-Varying Effect Model, which is a tool to explore lags in intensive longitudinal data.

Using Intensive Longitudinal Data to Study Affective Dynamics

Studying affective dynamics from intraindividual variability in intensive longitudinal data.

Teaching

I currently teach the following courses at Dartmouth College:

  • QBS 126: Analysis of Densely Collected Longitudinal Data
  • QBS 270: Quantitative Biomedical Sciences Journal Club

I have taught the following courses at Pennsylvania State University:

  • PSYCH 238: Introduction to Personality Psychology
  • PSYCH 301W: Basic Research Methods in Psychology
  • PSYCH 481: Introduction to Clinical Psychology

Contact

Join Our Team

We are currently recruiting:

  1. Undergraduate research assistants
  2. Graduate students





Specific Areas of Involvement for Undergraduate Research Assistants

The current lab has many foci concentrating around:

  1. Developing statistical methods for densely sampled data collection.
  2. The integration of smartphone-based and wearable data to draw inferences about a person’s mental health status.
  3. Creating new smartphone applications to assess and treat anxiety, depression, and substance abuse.

Research assistants do not need to be interested in participating in all projects. For example, someone with a major of the psychology and the brain sciences may be interested in that involve digital mental health, but they are not interested in projects that involve any coding/syntax writing. Alternatively, someone in computer science may be more interested in the creation of a digital platform, and they would not be required to have an interest in mental health or statistics.

Concrete practical benefits to undergraduate research assistants:

  • Opportunity to learn advanced quantitative skills
  • Creation of mentored original work (i.e. potentially being a co-author or first author on journal publication submissions and conference submissions)
  • Development of a strong relationship with a faculty member for (1) letters of recommendations, and/or (2) professional references
  • Excellent exposure to the following fields: (1) mental health (i.e. psychiatry, clinical psychology), (2) data science (i.e. bioinformatics, statistics, information science), and (3) computer science (i.e. computer engineering).

  • Extensions of modern statistical methods. Current smartphone-based and wearable data collection are limited in their ability to draw strong conclusions about the timing of causal processes. In the lab, one of our foci is the creation and validation of new tools, including the timing of naturalistic and causal processes within intensive longitudinal data.

  • Recommended interests: statistics (e.g. the major in Mathematical Data Science, though this is certainly not required).

  • Specific opportunities: writing statistical software code, conducting Monte Carlo simulation studies, and writing manuscripts

  • Digital phenotyping/passive sensing of mental health. In recent years, wearable devices and smartphone sensors have been used to measure constructs related to psychopathology. The current work requires quantitative interests in applying machine learning to smartphone sensor data to determine whether some signals might be fundamentally related to psychopathology processes (e.g. whether we can determine if someone is experiencing an increase in anxiety based on their sleep patterns detected through smartphone accelerometers).

  • Specific opportunities: cleaning, organizing, and synthesizing data; literature reviews; coding; manuscript writing

  • Developing Apps for Mental Health. Very few persons have access to care for their mental health. A major focus of this lab is creating digital solutions to assess and treat mental health problems. The main mental health problems that we treat are anxiety and depressive disorders, although there are opportunities to develop treatments for other types of mental health problems. These include smartphone-based and web-based applications. To date, we have created applications with more than 50,000 installs, and, consequently, this work can have profound and wide impact on persons daily lives.

  • Specific opportunities: writing and altering app code (in Java for Android and/or Swift for iOS); working on the server-side code for data management; creating design prototypes; marketing and promotion of applications; potential opportunities for manuscript writing

Interested students should contact Nick directly (Nicholas.C.Jacobson@dartmouth.edu).





Specific Areas of Involvement for Graduate Students

To be a graduate student, you must first gain admission to a graduate program within Dartmouth College. The lab focuses on problems which are highly interdisciplinary in nature, intersecting psychiatry/clinical psychology, computer science/computer engineering, and data science/statistics. Graduate programs of interest could include, but are not limited to, the following degree programs:

Graduate students in the lab will have the opportunity to choose and build upon their own areas of interest, as long as they align within the broader research areas of the lab. Graduate students will have the opportunity to be trained in paradigms of digital mental health and in the analysis of intensive longitudinal data related to the assessment of mental health in daily life. Graduate students will develop expertise that will prepare them for careers in academia and/or industry.

Graduate students will be expected to both participate in lab projects and develop their own independent areas of study. Graduate students will have the opportunity and be expected to participate fully in the research process, writing and submiting manuscripts, presenting posters or talks at research conferences, and writing graduate student fellowships and/or contributing to preliminary analyses to a full grant proposal. Graduate students will also be expected to mentor undergraduate students.

Graduate students will have access to existing lab data and have the opportunities to collect new original data. Interested students should contact Nick directly (Nicholas.C.Jacobson@dartmouth.edu).





Collaborators

The lab tackles many interdisciplinary problems that are best addressed by team-science approaches, and we welcome the involvement of collaborators within both academia and industry. If you are interested in collaborating with the lab, please contact Nick directly (Nicholas.C.Jacobson@dartmouth.edu).