Large-scale wearable data reveal digital phenotypes for daily-life stress detection

Stress and cardiovascular disease: an update

Psychological stress is generally accepted to be associated with an increased risk of cardiovascular disease (CVD), but results have varied in terms of how stress is measured and the strength of the association. Additionally, the mechanisms and potential causal links have remained speculative despite decades of research. The physiological responses to stress are well characterized, but their contribution to the development and progression of CVD has received little attention in empirical studies. Evidence suggests that physiological responses to stress have a fundamental role in the risk of CVD and that haemodynamic, vascular and immune perturbations triggered by stress are especially implicated. Stress response physiology is regulated by the corticolimbic regions of the brain, which have outputs to the autonomic nervous system. Variation in these regulatory pathways might explain interindividual differences in vulnerability to stress. Dynamic perturbations in autonomic, immune and vascular functions are probably also implicated as CVD risk mechanisms of chronic, recurring and cumulative stressful exposures, but more data are needed from prospective studies and from assessments in real-life situations. Psychological assessment remains insufficiently recognized in clinical care and prevention. Although stress-reduction interventions might mitigate perceived stress levels and potentially reduce cardiovascular risk, more data from randomized trials are needed.

Predicting the severity of mood and neuropsychiatric symptoms from digital biomarkers using wearable physiological data and deep learning

Neuropsychiatric symptoms (NPS) and mood disorders are common in individuals with mild cognitive impairment (MCI) and increase the risk of progression to dementia. Wearable devices collecting physiological and behavioral data can help in remote, passive, and continuous monitoring of moods and NPS, overcoming limitations and inconveniences of current assessment methods. In this longitudinal study, we examined the predictive ability of digital biomarkers based on sensor data from a wrist-worn wearable to determine the severity of NPS and mood disorders on a daily basis in older adults with predominant MCI. In addition to conventional physiological biomarkers, such as heart rate variability and skin conductance levels, we leveraged deep-learning features derived from physiological data using a self-supervised convolutional autoencoder. Models combining common digital biomarkers and deep features predicted depression severity scores with a correlation of r = 0.73 on average, total severity of mood disorder symptoms with r = 0.67, and mild behavioral impairment scores with r = 0.69 in the study population. Our findings demonstrated the potential of physiological biomarkers collected from wearables and deep learning methods to be used for the continuous and unobtrusive assessments of mental health symptoms in older adults, including those with MCI. TRIAL REGISTRATION: This trial was registered with ClinicalTrials.gov (NCT05059353) on September 28, 2021, titled "Effectiveness and Safety of a Digitally Based Multidomain Intervention for Mild Cognitive Impairment".

A machine-learning approach for stress detection using wearable sensors in free-living environments

Stress is a psychological condition resulting from the body's response to challenging situations, which can negatively impact physical and mental health if experienced over prolonged periods. Early detection of stress is crucial to prevent chronic health problems. Wearable sensors offer an effective solution for continuous and real-time stress monitoring due to their non-intrusive nature and ability to monitor vital signs, e.g., heart rate and activity. Typically, most existing research has focused on data collected in controlled environments. Yet, our study aims to propose a machine learning-based approach for detecting stress in a free-living environment using wearable sensors. We utilized the SWEET dataset, which includes data from 240 subjects collected via electrocardiography (ECG), skin temperature (ST), and skin conductance (SC). We assessed four machine learning models, i.e., K-Nearest Neighbors (KNN), Support Vector Classification (SVC), Decision Tree (DT), Random Forest (RF), and XGBoost (XGB) in four different settings. This study evaluates the performance of various machine learning models for stress classification using the SWEET dataset. The analysis included two binary classification scenarios (with and without SMOTE) and two multi-class classification scenarios (with and without SMOTE). The Random Forest model demonstrated superior performance in the binary classification without SMOTE, achieving an accuracy of 98.29 % and an F1-score of 97.89 %. For binary classification with SMOTE, the K-Nearest Neighbors model performed best, with an accuracy of 95.70 % and an F1-score of 95.70 %. In the three-level classification without SMOTE, the Random Forest model again excelled, achieving an accuracy of 97.98 % and an F1-score of 97.22 %. For three-level classification with SMOTE, XGBoost showed the highest performance, with an accuracy and F1-score of 98.98 %. These results highlight the effectiveness of different models under various conditions, emphasizing the importance of model selection and preprocessing techniques in enhancing classification performance.

Recent Progress in Biosensors for Depression Monitoring-Advancing Personalized Treatment

Depression is currently a major contributor to unnatural deaths and the healthcare burden globally, and a patient's battle with depression is often a long one. Because the causes, symptoms, and effects of medications are complex and highly individualized, early identification and personalized treatment of depression are key to improving treatment outcomes. The development of wearable electronics, machine learning, and other technologies in recent years has provided more possibilities for the realization of this goal. Conducting regular monitoring through biosensing technology allows for a more comprehensive and objective analysis than previous self-evaluations. This includes identifying depressive episodes, distinguishing somatization symptoms, analyzing etiology, and evaluating the effectiveness of treatment programs. This review summarizes recent research on biosensing technologies for depression. Special attention is given to technologies that can be portable or wearable, with the potential to enable patient use outside of the hospital, for long periods.

Mobile sensing-based depression severity assessment in participants with heterogeneous mental health conditions

Mobile sensing-based depression severity assessment could complement the subjective questionnaires-based assessment currently used in practice. However, previous studies on mobile sensing for depression severity assessment were conducted on homogeneous mental health condition participants; evaluation of possible generalization across heterogeneous groups has been limited. Similarly, previous studies have not investigated the potential of free-living audio data for depression severity assessment. Audio recordings from free-living could provide rich sociability features to characterize depressive states. We conducted a study with 11 healthy individuals, 13 individuals with major depressive disorder, and eight individuals with schizoaffective disorders. Communication logs and location data from the participants' smartphones and continuous audio recordings of free-living from a wearable audioband were obtained over a week for each participant. The depression severity prediction model trained using communication log and location data features had a root mean squared error (rmse) of 6.80. Audio-based sociability features further reduced the rmse to 6.07 (normalized rmse of 0.22). Audio-based sociability features also improved the F1 score in the five-class depression category classification model from 0.34 to 0.46. Thus, free-living audio-based sociability features complement the commonly used mobile sensing features to improve depression severity assessment. The prediction results obtained with mobile sensing-based features are better than the rmse of 9.83 (normalized rmse of 0.36) and the F1 score of 0.25 obtained with a baseline model. Additionally, the predicted depression severity had a significant correlation with reported depression severity (correlation coefficient of 0.76, p < 0.001). Thus, our work shows that mobile sensing could model depression severity across participants with heterogeneous mental health conditions, potentially offering a screening tool for depressive symptoms monitoring in the broader population.

Stress-induced increase in heart-rate during sleep as an indicator of PTSD risk among combat soldiers

STUDY OBJECTIVES

Discerning the differential contribution of sleep behavior and sleep physiology to the subsequent development of posttraumatic-stress-disorder (PTSD) symptoms following military operational service among combat soldiers.

METHODS

Longitudinal design with three measurement time points: during basic training week (T1), during intensive stressed training week (T2), and following military operational service (T3). Participating soldiers were all from the same unit, ensuring equivalent training schedules and stress exposures. During measurement weeks soldiers completed the Depression Anxiety and Stress Scale (DASS) and the PTSD Checklist for DSM-5 (PCL-5). Sleep physiology (sleep heart-rate) and sleep behavior (duration, efficiency) were monitored continuously in natural settings during T1 and T2 weeks using wearable sensors.

RESULTS

Repeated measures ANOVA revealed a progressive increase in PCL-5 scores from T1 and T2 to T3, suggesting an escalation in PTSD symptom severity following operational service. Hierarchical linear regression analysis uncovered a significant relation between the change in DASS stress scores from T1 to T2 and subsequent PCL-5 scores at T3. Incorporating participants' sleep heart-rate markedly enhanced the predictive accuracy of the model, with increased sleep heart-rate from T1 to T2 emerging as a significant predictor of elevated PTSD symptoms at T3, above and beyond the contribution of DASS stress scores. Sleep behavior did not add to the accuracy of the model.

CONCLUSION

Findings underscore the critical role of sleep physiology, specifically elevated sleep heart-rate following stressful military training, in indicating subsequent PTSD risk following operational service among combat soldiers. These findings may contribute to PTSD prediction and prevention efforts.

Quantifying Posttraumatic Stress Disorder Symptoms During Traumatic Memories Using Interpretable Markers of Respiratory Variability

BACKGROUND

Posttraumatic stress disorder (PTSD) causes heightened fight-or-flight responses to traumatic memories (i.e., hyperarousal). Although hyperarousal is hypothesized to cause irregular breathing (i.e., respiratory variability), no quantitative markers of respiratory variability have been shown to correspond with PTSD symptoms in humans.

OBJECTIVE

In this study, we define interpretable markers of respiration pattern variability (RPV) and investigate whether these markers respond during traumatic memories, correlate with PTSD symptoms, and differ in patients with PTSD.

METHODS

We recruited 156 veterans from the Vietnam-Era Twin Registry to participate in a trauma recall protocol. From respiratory effort and electrocardiogram measurements, we extracted respiratory timings and rate using a robust quality assessment and fusion approach. We then quantified RPV using the interquartile range and compared RPV between baseline and trauma recall conditions, correlated PTSD symptoms to the difference between trauma recall and baseline RPV (i.e., ∆RPV), and compared ∆RPV between patients with PTSD and trauma-exposed controls. Leveraging a subset of 116 paired twins, we then uniquely controlled for factors shared by co-twins via within-pair analysis for further validation.

RESULTS

We found RPV was increased during traumatic memories (p .001), ∆ RPV was positively correlated with PTSD symptoms (p .05), and patients with PTSD exhibited higher ∆ RPV than trauma-exposed controls (p . 05).

CONCLUSIONS

This paper is the first to elucidate RPV markers that respond during traumatic memories, especially in patients with PTSD, and correlate with PTSD symptoms.

SIGNIFICANCE

These findings encourage future studies outside the clinic, where interpretable markers of respiratory variability are used to track hyperarousal.

Five million nights: temporal dynamics in human sleep phenotypes

Sleep monitoring has become widespread with the rise of affordable wearable devices. However, converting sleep data into actionable change remains challenging as diverse factors can cause combinations of sleep parameters to differ both between people and within people over time. Researchers have attempted to combine sleep parameters to improve detecting similarities between nights of sleep. The cluster of similar combinations of sleep parameters from a night of sleep defines that night's sleep phenotype. To date, quantitative models of sleep phenotype made from data collected from large populations have used cross-sectional data, which preclude longitudinal analyses that could better quantify differences within individuals over time. In analyses reported here, we used five million nights of wearable sleep data to test (a) whether an individual's sleep phenotype changes over time and (b) whether these changes elucidate new information about acute periods of illness (e.g., flu, fever, COVID-19). We found evidence for 13 sleep phenotypes associated with sleep quality and that individuals transition between these phenotypes over time. Patterns of transitions significantly differ (i) between individuals (with vs. without a chronic health condition; chi-square test; p-value < 1e-100) and (ii) within individuals over time (before vs. during an acute condition; Chi-Square test; p-value < 1e-100). Finally, we found that the patterns of transitions carried more information about chronic and acute health conditions than did phenotype membership alone (longitudinal analyses yielded 2-10× as much information as cross-sectional analyses). These results support the use of temporal dynamics in the future development of longitudinal sleep analyses.

Physiological Profiling of Agitation in Dementia: Insights From Wearable Sensor Data

Background and Objectives

The number of people with dementia is expected to triple to 152 million in 2050, with 90% having accompanying behavioral and psychological symptoms (BPSD). Agitation is among the most critical BPSD and can lead to decreased quality of life for people with dementia and their caregivers. This study aims to explore objective quantification of agitation in people with dementia by analyzing the relationships between physiological and movement data from wearables and observational measures of agitation.

Research Design and Methods

The data presented here is from 30 people with dementia, each included for 1 week, collected following our previously published multimodal data collection protocol. This observational protocol has a cross-sectional repeated measures design, encompassing data from both wearable and fixed sensors. Generalized linear mixed models were used to quantify the relationship between data from different wearable sensor modalities and agitation, as well as motor and verbal agitation specifically.

Results

Several features from wearable data are significantly associated with agitation, at least the p < .05 level (absolute β: 0.224-0.753). Additionally, different features are informative depending on the agitation type or the patient the data were collected from. Adding context with key confounding variables (time of day, movement, and temperature) allows for a clearer interpretation of feature differences when a person with dementia is agitated.

Discussion and Implications

The features shown to be significantly different, across the study population, suggest possible autonomic nervous system activation when agitated. Differences when splitting the data by agitation type point toward a need for future detection models to tailor to the primary type of agitation expressed. Finally, patient-specific differences in features indicate a need for patient- or group-level model personalization. The findings reported in this study both reinforce and add to the fundamental understanding of and can be used to drive the objective quantification of agitation.

Psychogenic Aging: A Novel Prospect to Integrate Psychobiological Hallmarks of Aging

Psychological factors are amongst the most robust predictors of healthspan and longevity, yet are rarely incorporated into scientific and medical frameworks of aging. The prospect of characterizing and integrating the psychological influences of aging is therefore an unmet step for the advancement of geroscience. Psychogenic Aging research is an emerging branch of biogerontology that aims to address this gap by investigating the impact of psychological factors on human longevity. It is an interdisciplinary field that integrates complex psychological, neurological, and molecular relationships that can be best understood with precision medicine methodologies. This perspective argues that psychogenic aging should be considered an integral component of the Hallmarks of Aging framework, opening the doors for future biopsychosocial integration in longevity research. By providing a unique perspective on frequently overlooked aspects of organismal aging, psychogenic aging offers new insights and targets for anti-aging therapeutics on individual and societal levels that can significantly benefit the scientific and medical communities.

Respiratory modulation of the heart rate: A potential biomarker of cardiorespiratory function in human

In recent decade, wearable digital devices have shown potentials for the discovery of novel biomarkers of humans' physiology and behavior. Heart rate (HR) and respiration rate (RR) are most crucial bio-signals in humans' digital phenotyping research. HR is a continuous and non-invasive proxy to autonomic nervous system and ample evidence pinpoints the critical role of respiratory modulation of cardiac function. In the present study, we recorded longitudinal (7 days, 4.63 ± 1.52) HR and RR of 89 freely behaving human subjects (Female: 39, age 57.28 ± 5.67, Male: 50, age 58.48 ± 6.32) and analyzed their dynamics using linear models and information theoretic measures. While HR's linear and nonlinear characteristics were expressed within the plane of the HR-RR directed flow of information (HR→RR - RR→HR), their dynamics were determined by its RR→HR axis. More importantly, RR→HR quantified the effect of alcohol consumption on individuals' cardiorespiratory function independent of their consumed amount of alcohol, thereby signifying the presence of this habit in their daily life activities. The present findings provided evidence for the critical role of the respiratory modulation of HR, which was previously only studied in non-human animals. These results can contribute to humans' phenotyping research by presenting RR→HR as a digital diagnosis/prognosis marker of humans' cardiorespiratory pathology.

Relieving the burden of intensive labeling for stress monitoring in the wild by using semi-supervised learning

Stress, a natural process affecting individuals' wellbeing, has a profound impact on overall quality of life. Researchers from diverse fields employ various technologies and methodologies to investigate it and alleviate the negative effects of this phenomenon. Wearable devices, such as smart bands, capture physiological data, including heart rate variability, motions, and electrodermal activity, enabling stress level monitoring through machine learning models. However, labeling data for model accuracy assessment poses a significant challenge in stress-related research due to incomplete or inaccurate labels provided by individuals in their daily lives. To address this labeling predicament, our study proposes implementing Semi-Supervised Learning (SSL) models. Through comparisons with deep learning-based supervised models and clustering-based unsupervised models, we evaluate the performance of our SSL models. Our experiments show that our SSL models achieve 77% accuracy with a classifier trained on an augmented dataset prepared using the label propagation (LP) algorithm. Additionally, our deep autoencoder network achieves 76% accuracy. These results highlight the superiority of SSL models over unsupervised learning techniques and their comparable performance to supervised learning models, even with limited labeled data. By relieving the burden of labeling in daily life stress recognition, our study advances stress-related research, recognizing stress as a natural process rather than a disease. This facilitates the development of more efficient and accurate stress monitoring methods in the wild.

Dissecting the heterogeneity of "in the wild" stress from multimodal sensor data

Stress is associated with numerous chronic health conditions, both mental and physical. However, the heterogeneity of these associations at the individual level is poorly understood. While data generated from individuals in their day-to-day lives "in the wild" may best represent the heterogeneity of stress, gathering these data and separating signals from noise is challenging. In this work, we report findings from a major data collection effort using Digital Health Technologies (DHTs) and frontline healthcare workers. We provide insights into stress "in the wild", by using robust methods for its identification from multimodal data and quantifying its heterogeneity. Here we analyze data from the Stress and Recovery in Frontline COVID-19 Workers study following 365 frontline healthcare workers for 4-6 months using wearable devices and smartphone app-based measures. Causal discovery is used to learn how the causal structure governing an individual's self-reported symptoms and physiological features from DHTs differs between non-stress and potential stress states. Our methods uncover robust representations of potential stress states across a population of frontline healthcare workers. These representations reveal high levels of inter- and intra-individual heterogeneity in stress. We leverage multiple stress definitions that span different modalities (from subjective to physiological) to obtain a comprehensive view of stress, as these differing definitions rarely align in time. We show that these different stress definitions can be robustly represented as changes in the underlying causal structure on and off stress for individuals. This study is an important step toward better understanding potential underlying processes generating stress in individuals.

Smartphone sensor data estimate alcohol craving in a cohort of patients with alcohol-associated liver disease and alcohol use disorder

BACKGROUND

Sensors within smartphones, such as accelerometer and location, can describe longitudinal markers of behavior as represented through devices in a method called digital phenotyping. This study aimed to assess the feasibility of digital phenotyping for patients with alcohol-associated liver disease and alcohol use disorder, determine correlations between smartphone data and alcohol craving, and establish power assessment for future studies to prognosticate clinical outcomes.

METHODS

A total of 24 individuals with alcohol-associated liver disease and alcohol use disorder were instructed to download the AWARE application to collect continuous sensor data and complete daily ecological momentary assessments on alcohol craving and mood for up to 30 days. Data from sensor streams were processed into features like accelerometer magnitude, number of calls, and location entropy, which were used for statistical analysis. We used repeated measures correlation for longitudinal data to evaluate associations between sensors and ecological momentary assessments and standard Pearson correlation to evaluate within-individual relationships between sensors and craving.

RESULTS

Alcohol craving significantly correlated with mood obtained from ecological momentary assessments. Across all sensors, features associated with craving were also significantly correlated with all moods (eg, loneliness and stress) except boredom. Individual-level analysis revealed significant relationships between craving and features of location entropy and average accelerometer magnitude.

CONCLUSIONS

Smartphone sensors may serve as markers for alcohol craving and mood in alcohol-associated liver disease and alcohol use disorder. Findings suggest that location-based and accelerometer-based features may be associated with alcohol craving. However, data missingness and low participant retention remain challenges. Future studies are needed for further digital phenotyping of relapse risk and progression of liver disease.

Simulation of ambulatory electrodermal activity and the handling of low-quality segments

BACKGROUND AND OBJECTIVES

Monitoring electrodermal activity (EDA) in daily life requires effective handling of low-quality segments, which are common in ambulatory EDA data. Although several low-quality handling methods have been implemented, systematic comparison of these methods, which requires a large annotated dataset, is lacking.

METHODS

Therefore, we proposed the simulation of realistic ambulatory EDA data starting from high-quality EDA signals, which were subsequently contaminated with varying concentrations of artifacts. Subsequently, three approaches for handling low-quality data were evaluated regarding the preservation of several EDA-derived features: removing all artifacts, interpolating over removed artifacts, and retaining all artifacts. Specifically, multiple EDA features were assessed, derived from response detection (evaluated using F1, precision, recall) as well as EDA, phasic, and tonic features (assessed using absolute error), by comparing the simulated EDA data with and without the inserted artifacts, using the latter as ground truth.

RESULTS

For response detection, retaining artifacts resulted in the highest F1-scores, while interpolating over removed artifacts achieved the highest F1-scores for the phasic signal. The approaches did significantly differ in the mean error for the phasic but not for the tonic component and raw EDA.

CONCLUSION

This work generated ambulatory EDA datasets of 200 h, containing 0.125 to 3 artifacts per minute, and showed that interpolation over removed artifacts was an effective approach to reconstruct phasic-derived features up to 2 artifacts per minute. The proposed simulation and evaluation methodology, which are easily customizable, offer opportunities for future research to develop and systematically compare signal quality indicators, decomposition methods, and response detectors for processing ambulatory EDA.

Mental Health for All: The Case for Investing in Digital Mental Health to Improve Global Outcomes, Access, and Innovation in Low-Resource Settings

Mental health disorders are an increasing global public health concern that contribute to morbidity, mortality, disability, and healthcare costs across the world. Biomedical and psychological research has come a long way in identifying the importance of mental health and its impact on behavioral risk factors, physiological health, and overall quality of life. Despite this, access to psychological and psychiatric services remains widely unavailable and is a challenge for many healthcare systems, particularly those in developing countries. This review article highlights the strengths and opportunities brought forward by digital mental health in narrowing this divide. Further, it points to the economic and societal benefits of effectively managing mental illness, making a case for investing resources into mental healthcare as a larger priority for large non-governmental organizations and individual nations across the globe.

Detecting Prolonged Stress in Real Life Using Wearable Biosensors and Ecological Momentary Assessments: Naturalistic Experimental Study

BACKGROUND

Increasing efforts toward the prevention of stress-related mental disorders have created a need for unobtrusive real-life monitoring of stress-related symptoms. Wearable devices have emerged as a possible solution to aid in this process, but their use in real-life stress detection has not been systematically investigated.

OBJECTIVE

We aimed to determine the utility of ecological momentary assessments (EMA) and physiological arousal measured through wearable devices in detecting ecologically relevant stress states.

METHODS

Using EMA combined with wearable biosensors for ecological physiological assessments (EPA), we investigated the impact of an ecological stressor (ie, a high-stakes examination week) on physiological arousal and affect compared to a control week without examinations in first-year medical and biomedical science students (51/83, 61.4% female). We first used generalized linear mixed-effects models with maximal fitting approaches to investigate the impact of examination periods on subjective stress exposure, mood, and physiological arousal. We then used machine learning models to investigate whether we could use EMA, wearable biosensors, or the combination of both to classify momentary data (ie, beeps) as belonging to examination or control weeks. We tested both individualized models using a leave-one-beep-out approach and group-based models using a leave-one-subject-out approach.

RESULTS

During stressful high-stakes examination (versus control) weeks, participants reported increased negative affect and decreased positive affect. Intriguingly, physiological arousal decreased on average during the examination week. Time-resolved analyses revealed peaks in physiological arousal associated with both momentary self-reported stress exposure and self-reported positive affect. Mediation models revealed that the decreased physiological arousal in the examination week was mediated by lower positive affect during the same period. We then used machine learning to show that while individualized EMA outperformed EPA in its ability to classify beeps as originating from examinations or from control weeks (1603/4793, 33.45% and 1648/4565, 36.11% error rates, respectively), a combination of EMA and EPA yields optimal classification (1363/4565, 29.87% error rate). Finally, when comparing individualized models to group-based models, we found that the individualized models significantly outperformed the group-based models across all 3 inputs (EMA, EPA, and the combination).

CONCLUSIONS

This study underscores the potential of wearable biosensors for stress-related mental health monitoring. However, it emphasizes the necessity of psychological context in interpreting physiological arousal captured by these devices, as arousal can be related to both positive and negative contexts. Moreover, our findings support a personalized approach in which momentary stress is optimally detected when referenced against an individual's own data.

Towards PPG-based anger detection for emotion regulation

BACKGROUND

Anger dyscontrol is a common issue after traumatic brain injury (TBI). With the growth of wearable physiological sensors, there is new potential to facilitate the rehabilitation of such anger in the context of daily life. This potential, however, depends on how well physiological markers can distinguish changing emotional states and for such markers to generalize to real-world settings. Our study explores how wearable photoplethysmography (PPG), one of the most widely available physiological sensors, could be used detect anger within a heterogeneous population.

METHODS

This study collected the TRIEP (Toronto Rehabilitation Institute Emotion-Physiology) dataset, which comprised of 32 individuals (10 TBI), exposed to a variety of elicitation material (film, pictures, self-statements, personal recall), over two day sessions. This complex dataset allowed for exploration into how the emotion-PPG relationship varied over changes in individuals, endogenous/exogenous drivers of emotion, and day-to-day differences. A multi-stage analysis was conducted looking at: (1) times-series visual clustering, (2) discriminative time-interval features of anger, and (3) out-of-sample anger classification.

RESULTS

Characteristics of PPG are largely dominated by inter-subject (between individuals) differences first, then intra-subject (day-to-day) changes, before differentiation into emotion. Both TBI and non-TBI individuals showed evidence of linear separable features that could differentiate anger from non-anger classes within time-interval analysis. However, what is more challenging is that these separable features for anger have various degrees of stability across individuals and days.

CONCLUSION

This work highlights how there are contextual, non-stationary challenges to the emotion-physiology relationship that must be accounted for before emotion regulation technology can perform in real-world scenarios. It also affirms the need for a larger breadth of emotional sampling when building classification models.

Generalizable machine learning for stress monitoring from wearable devices: A systematic literature review

INTRODUCTION

Wearable sensors have shown promise as a non-intrusive method for collecting biomarkers that may correlate with levels of elevated stress. Stressors cause a variety of biological responses, and these physiological reactions can be measured using biomarkers including Heart Rate Variability (HRV), Electrodermal Activity (EDA) and Heart Rate (HR) that represent the stress response from the Hypothalamic-Pituitary-Adrenal (HPA) axis, the Autonomic Nervous System (ANS), and the immune system. While Cortisol response magnitude remains the gold standard indicator for stress assessment [1], recent advances in wearable technologies have resulted in the availability of a number of consumer devices capable of recording HRV, EDA and HR sensor biomarkers, amongst other signals. At the same time, researchers have been applying machine learning techniques to the recorded biomarkers in order to build models that may be able to predict elevated levels of stress.

OBJECTIVE

The aim of this review is to provide an overview of machine learning techniques utilized in prior research with a specific focus on model generalization when using these public datasets as training data. We also shed light on the challenges and opportunities that machine learning-enabled stress monitoring and detection face.

METHODS

This study reviewed published works contributing and/or using public datasets designed for detecting stress and their associated machine learning methods. The electronic databases of Google Scholar, Crossref, DOAJ and PubMed were searched for relevant articles and a total of 33 articles were identified and included in the final analysis. The reviewed works were synthesized into three categories of publicly available stress datasets, machine learning techniques applied using those, and future research directions. For the machine learning studies reviewed, we provide an analysis of their approach to results validation and model generalization. The quality assessment of the included studies was conducted in accordance with the IJMEDI checklist [2].

RESULTS

A number of public datasets were identified that are labeled for stress detection. These datasets were most commonly produced from sensor biomarker data recorded using the Empatica E4 device, a well-studied, medical-grade wrist-worn wearable that provides sensor biomarkers most notable to correlate with elevated levels of stress. Most of the reviewed datasets contain less than twenty-four hours of data, and the varied experimental conditions and labeling methodologies potentially limit their ability to generalize for unseen data. In addition, we discuss that previous works show shortcomings in areas such as their labeling protocols, lack of statistical power, validity of stress biomarkers, and model generalization ability.

CONCLUSION

Health tracking and monitoring using wearable devices is growing in popularity, while the generalization of existing machine learning models still requires further study, and research in this area will continue to provide improvements as newer and more substantial datasets become available.

Mild daily stress, in interaction with NR3C1 DNA methylation levels, is linked to alterations in the HPA axis and ANS response to acute stress in early adolescents

BACKGROUND

Daily Hassles (DH) or daily stress - is a mild type of stressor with unique contributions to psychological distress. Yet, most prior studies that investigate the effects of stressful life experiences focus on childhood trauma or on early life stress and little is known about the effects of DH on epigenetic changes in stress system related genes and on the physiological response to social stressors.

METHODS

In the present study, conducted among 101 early adolescents (mean age = 11.61; SD = 0.64), we investigated whether Autonomic Nervous System (ANS) (namely heart rate and heart rate variability) and Hypothalamic-Pituitary-Adrenal (HPA) axis functioning (measured as cortisol stress reactivity and recovery) are associated with DNA methylation (DNAm) in the glucocorticoid receptor gene (NR3C1), the level of DH and their interaction. To assess the stress system functioning the TSST protocol was used.

RESULTS

Our findings show that higher NR3C1 DNAm in interaction with higher levels of daily hassles, is associated with blunted HPA axis reactivity to psychosocial stress. In addition, higher levels of DH are associated with extended HPA axis stress recovery. In addition, participants with higher NR3C1 DNAm had lower ANS adaptability to stress, specifically lower parasympathetic withdrawal; for heart rate variability this effect was strongest for participants with higher level of DH.

CONCLUSIONS

The observation that interaction effects between NR3C1 DNAm levels and daily stress on the functioning of the stress-systems, are already detectable in young adolescents, highlights the importance of early interventions, not only in the case of trauma, but also daily stress. This might help to prevent stress-induced mental and physical disorders later in life.

Stress Estimation Using Biometric and Activity Indicators to Improve QoL of the Elderly

It is essential to estimate the stress state of the elderly to improve their QoL. Stress states change every day and hour, depending on the activities performed and the duration/intensity. However, most existing studies estimate stress states using only biometric information or specific activities (e.g., sleep duration, exercise duration/amount, etc.) as explanatory variables and do not consider all daily living activities. It is necessary to link various daily living activities and biometric information in order to estimate the stress state more accurately. Specifically, we construct a stress estimation model using machine learning with the answers to a stress status questionnaire obtained every morning and evening as the ground truth and the biometric data during each of the performed activities and the new proposed indicator including biological and activity perspectives as the features. We used the following methods: Baseline Method 1, in which the RRI variance and Lorenz plot area for 4 h after waking and 24 h before the questionnaire were used as features; Baseline Method 2, in which sleep time was added as a feature to Baseline Method 1; the proposed method, in which the Lorenz plot area per activity and total time per activity were added. We compared the results with the proposed method, which added the new indicators as the features. The results of the evaluation experiments using the one-month data collected from five elderly households showed that the proposed method had an average estimation accuracy of 59%, 7% better than Baseline Method 1 (52%) and 4% better than Baseline Method 2 (55%).

Stressors and Destressors in Working From Home Based on Context and Physiology From Self-Reports and Smartwatch Measurements: International Observational Study Trial

Background

The COVID-19 pandemic has greatly boosted working from home as a way of working, which is likely to continue for most companies in the future, either in fully remote or in hybrid form. To manage stress levels in employees working from home, insights into the stressors and destressors in a home office first need to be studied.

Objective

We present an international remote study with employees working from home by making use of state-of-the-art technology (ie, smartwatches and questionnaires through smartphones) first to determine stressors and destressors in people working from home and second to identify smartwatch measurements that could represent these stressors and destressors.

Methods

Employees working from home from 3 regions of the world (the United States, the United Kingdom, and Hong Kong) were asked to wear a smartwatch continuously for 7 days and fill in 5 questionnaires each day and 2 additional questionnaires before and after the measurement week. The entire study was conducted remotely. Univariate statistical analyses comparing variable distributions between low and high stress levels were followed by multivariate analysis using logistic regression, considering multicollinearity by using variance inflation factor (VIF) filtering.

Results

A total of 202 people participated, with 198 (98%) participants finishing the experiment. Stressors found were other people and daily life getting in the way of work (P=.05), job intensity (P=.01), a history of burnout (P=.03), anxiety toward the pandemic (P=.04), and environmental noise (P=.01). Destressors found were access to sunlight (P=.02) and fresh air (P<.001) during the workday and going outdoors (P<.001), taking breaks (P<.001), exercising (P<.001), and having social interactions (P<.001). The smartwatch measurements positively related to stress were the number of active intensity periods (P<.001), the number of highly active intensity periods (P=.04), steps (P<.001), and the SD in the heart rate (HR; P<.001). In a multivariate setting, only a history of burnout (P<.001) and family and daily life getting in the way of work (P<.001) were positively associated with stress, while self-reports of social activities (P<.001) and going outdoors (P=.03) were negatively associated with stress. Stress prediction models based on questionnaire data had a similar performance (F1=0.51) compared to models based on automatic measurable data alone (F1=0.47).

Conclusions

The results show that there are stressors and destressors when working from home that should be considered when managing stress in employees. Some of these stressors and destressors are (in)directly measurable with unobtrusive sensors, and prediction models based on these data show promising results for the future of automatic stress detection and management.

Trial Registration

Netherlands Trial Register NL9378; https://trialsearch.who.int/Trial2.aspx?TrialID=NL9378

Real-time mental stress detection using multimodality expressions with a deep learning framework

Mental stress is becoming increasingly widespread and gradually severe in modern society, threatening people’s physical and mental health. To avoid the adverse effects of stress on people, it is imperative to detect stress in time. Many studies have demonstrated the effectiveness of using objective indicators to detect stress. Over the past few years, a growing number of researchers have been trying to use deep learning technology to detect stress. However, these works usually use single-modality for stress detection and rarely combine stress-related information from multimodality. In this paper, a real-time deep learning framework is proposed to fuse ECG, voice, and facial expressions for acute stress detection. The framework extracts the stress-related information of the corresponding input through ResNet50 and I3D with the temporal attention module (TAM), where TAM can highlight the distinguishing temporal representation for facial expressions about stress. The matrix eigenvector-based approach is then used to fuse the multimodality information about stress. To validate the effectiveness of the framework, a well-established psychological experiment, the Montreal imaging stress task (MIST), was applied in this work. We collected multimodality data from 20 participants during MIST. The results demonstrate that the framework can combine stress-related information from multimodality to achieve 85.1% accuracy in distinguishing acute stress. It can serve as a tool for computer-aided stress detection.

Analysis of wearable time series data in endocrine and metabolic research

Many hormones in the body oscillate with different frequencies and amplitudes, creating a dynamic environment that is essential to maintain health. In humans, disruptions to these rhythms are strongly associated with increased morbidity and mortality. While mathematical models can help us understand rhythm misalignment, translating this insight into personalised healthcare technologies requires solving additional challenges. Here, we discuss how combining minimally invasive, high-frequency biosampling technologies with wearable devices can assist the development of hormonal surrogates. We review bespoke algorithms that can help analyse multidimensional, noisy, time series data and identify wearable signals that could constitute clinical proxies of endocrine rhythms. These techniques can support the development of computational biomarkers to support the diagnosis and management of endocrine and metabolic conditions.

Measuring Health-Related Quality of Life With Multimodal Data: Viewpoint

The ability to objectively measure aspects of performance and behavior is a fundamental pillar of digital health, enabling digital wellness products, decentralized trial concepts, evidence generation, digital therapeutics, and more. Emerging multimodal technologies capable of measuring several modalities simultaneously and efforts to integrate inputs across several sources are further expanding the limits of what digital measures can assess. Experts from the field of digital health were convened as part of a multi-stakeholder workshop to examine the progress of multimodal digital measures in two key areas: detection of disease and the measurement of meaningful aspects of health relevant to the quality of life. Here we present a meeting report, summarizing key discussion points, relevant literature, and finally a vision for the immediate future, including how multimodal measures can provide value to stakeholders across drug development and care delivery, as well as three key areas where headway will need to be made if we are to continue to build on the encouraging progress so far: collaboration and data sharing, removal of barriers to data integration, and alignment around robust modular evaluation of new measurement capabilities.

The digital transformation of hepatology: The patient is logged in

The rise in innovative digital health technologies has led a paradigm shift in health care toward personalized, patient-centric medicine that is reaching beyond traditional brick-and-mortar facilities into patients' homes and everyday lives. Digital solutions can monitor and detect early changes in physiological data, predict disease progression and health-related outcomes based on individual risk factors, and manage disease intervention with a range of accessible telemedicine and mobile health options. In this review, we discuss the unique transformation underway in the care of patients with liver disease, specifically examining the digital transformation of diagnostics, prediction and clinical decision-making, and management. Additionally, we discuss the general considerations needed to confirm validity and oversight of new technologies, usability and acceptability of digital solutions, and equity and inclusivity of vulnerable populations.

Machine intelligence in non-invasive endocrine cancer diagnostics

Artificial intelligence (AI) has illuminated a clear path towards an evolving health-care system replete with enhanced precision and computing capabilities. Medical imaging analysis can be strengthened by machine learning as the multidimensional data generated by imaging naturally lends itself to hierarchical classification. In this Review, we describe the role of machine intelligence in image-based endocrine cancer diagnostics. We first provide a brief overview of AI and consider its intuitive incorporation into the clinical workflow. We then discuss how AI can be applied for the characterization of adrenal, pancreatic, pituitary and thyroid masses in order to support clinicians in their diagnostic interpretations. This Review also puts forth a number of key evaluation criteria for machine learning in medicine that physicians can use in their appraisals of these algorithms. We identify mitigation strategies to address ongoing challenges around data availability and model interpretability in the context of endocrine cancer diagnosis. Finally, we delve into frontiers in systems integration for AI, discussing automated pipelines and evolving computing platforms that leverage distributed, decentralized and quantum techniques.

Predicting Chronic Stress among Healthy Females Using Daily-Life Physiological and Lifestyle Features from Wearable Sensors

Background

Chronic stress is a highly prevalent condition that may stem from different sources and can substantially impact physiology and behavior, potentially leading to impaired mental and physical health. Multiple physiological and behavioral lifestyle features can now be recorded unobtrusively in daily-life using wearable sensors. The aim of the current study was to identify a distinct set of physiological and behavioral lifestyle features that are associated with elevated levels of chronic stress across different stress sources.

Methods

For that, 140 healthy female participants completed the Trier inventory for chronic stress (TICS) before wearing the Fitbit Charge3 sensor for seven consecutive days while maintaining their daily routine. Physiological and lifestyle features that were extracted from sensor data, alongside demographic features, were used to predict high versus low chronic stress with support vector machine classifiers, applying out-of-sample model testing.

Results

The model achieved 79% classification accuracy for chronic stress from a social tension source. A mixture of physiological (resting heart-rate, heart-rate circadian characteristics), lifestyle (steps count, sleep onset and sleep regularity) and non-sensor demographic features (smoking status) contributed to this classification.

Conclusion

As wearable technologies continue to rapidly evolve, integration of daily-life indicators could improve our understanding of chronic stress and its impact of physiology and behavior.

The cumulative effect of chronic stress and depressive symptoms affects heart rate in a working population

BACKGROUND

Chronic stress and depressive symptoms have both been linked to increased heart rate (HR) and reduced HR variability. However, up to date, it is not clear whether chronic stress, the mechanisms intrinsic to depression or a combination of both cause these alterations. Subclinical cases may help to answer these questions. In a healthy working population, we aimed to investigate whether the effect of chronic stress on HR circadian rhythm depends on the presence of depressive symptoms and whether chronic stress and depressive symptoms have differential effects on HR reactivity to an acute stressor.

METHODS

1,002 individuals of the SWEET study completed baseline questionnaires, including psychological information, and 5 days of electrocardiogram (ECG) measurements. Complete datasets were available for 516 individuals. In addition, a subset (n = 194) of these participants completed a stress task on a mobile device. Participants were grouped according to their scores for the Depression Anxiety Stress Scale (DASS) and Perceived Stress Scale (PSS). We explored the resulting groups for differences in HR circadian rhythm and stress reactivity using linear mixed effect models. Additionally, we explored the effect of stress and depressive symptoms on night-time HR variability [root mean square of successive differences (RMSSD)].

RESULTS

High and extreme stress alone did not alter HR circadian rhythm, apart from a limited increase in basal HR. Yet, if depressive symptoms were present, extreme chronic stress levels did lead to a blunted circadian rhythm and a lower basal HR. Furthermore, blunted stress reactivity was associated with depressive symptoms, but not chronic stress. Night-time RMSSD data was not influenced by chronic stress, depressive symptoms or their interaction.

CONCLUSION

The combination of stress and depressive symptoms, but not chronic stress by itself leads to a blunted HR circadian rhythm. Furthermore, blunted HR reactivity is associated with depressive symptoms and not chronic stress.

An Alternative to the Light Touch Digital Health Remote Study: The Stress and Recovery in Frontline COVID-19 Health Care Workers Study

Background

Several app-based studies share similar characteristics of a light touch approach that recruit, enroll, and onboard via a smartphone app and attempt to minimize burden through low-friction active study tasks while emphasizing the collection of passive data with minimal human contact. However, engagement is a common challenge across these studies, reporting low retention and adherence.

Objective

This study aims to describe an alternative to a light touch digital health study that involved a participant-centric design including high friction app-based assessments, semicontinuous passive data from wearable sensors, and a digital engagement strategy centered on providing knowledge and support to participants.

Methods

The Stress and Recovery in Frontline COVID-19 Health Care Workers Study included US frontline health care workers followed between May and November 2020. The study comprised 3 main components: (1) active and passive assessments of stress and symptoms from a smartphone app, (2) objective measured assessments of acute stress from wearable sensors, and (3) a participant codriven engagement strategy that centered on providing knowledge and support to participants. The daily participant time commitment was an average of 10 to 15 minutes. Retention and adherence are described both quantitatively and qualitatively.

Results

A total of 365 participants enrolled and started the study, and 81.0% (n=297) of them completed the study for a total study duration of 4 months. Average wearable sensor use was 90.6% days of total study duration. App-based daily, weekly, and every other week surveys were completed on average 69.18%, 68.37%, and 72.86% of the time, respectively.

Conclusions

This study found evidence for the feasibility and acceptability of a participant-centric digital health study approach that involved building trust with participants and providing support through regular phone check-ins. In addition to high retention and adherence, the collection of large volumes of objective measured data alongside contextual self-reported subjective data was able to be collected, which is often missing from light touch digital health studies.

Trial Registration

ClinicalTrials.gov NCT04713111; https://clinicaltrials.gov/ct2/show/NCT04713111

A Shrewd Artificial Neural Network-Based Hybrid Model for Pervasive Stress Detection of Students Using Galvanic Skin Response and Electrocardiogram Signals

Mental illness issues are a very common health issue in youths and adults across the world. The usage of real-time data analytics in health care has a great potential to improve and enhance the quality of health care services, including diagnosis and medical prescription. Stress is one of the major health issues these days, which leads to many acute and sometimes incurable diseases to the students of very young age. Stress affects physiological parameters of the human body; due to these, human emotions may also change. This research paper proposes a hybrid model for pervasive stress detection, which deals with imbalance class problems using real-time data analytics and Internet of Things and it also presents a new stress analysis system to detect stressful conditions of the student, and to diagnose whether they are stressed or relaxed by using a designed set of experimental tasks. Regular monitoring of students'/professionals' health, including measurement of Galvanic Skin Response (GSR) and Electrocardiogram (ECG) data, provides a good understanding of their stress level. Data are acquired by using GSR and ECG sensors for 34 participants while undertaking five different tasks discussed in the proposed experiment. The graphical relationship between heart rate, blood pressure, and skin conductance across various experimental activities highlights the fact as to how physiological parameters of the human body get affected along with the mental status of the students. This article performs accuracy computation by using different machine-learning models such as Logistic Regression (LR), Support Vector Machine (SVM), K-Nearest Neighbours (KNN), Bagging Classifiers (BAG), Random Forest (RF), Gradient Boosting (GB), and Artificial Neural Network (ANN) followed by tuning with the best set of hyper parameters for each model. The proposed hybrid classification model deals with the class imbalance problem by using the Synthetic Minority Oversampling Technique. The shrewd ANN-based hybrid model achieves 99.4% accuracy on the self-generated dataset for the mental state classification of the students, which is best among other classifiers such as LR, SVM, KNN, BAG, RF, GB, and ANN. The prediction result of all 34 participants of the experiment is also classified into four categories: relaxed, stressed, partially stressed, and happy.

The association between self-reported stress and cardiovascular measures in daily life: A systematic review

BACKGROUND

Stress plays an important role in the development of mental illness, and an increasing number of studies is trying to detect moments of perceived stress in everyday life based on physiological data gathered using ambulatory devices. However, based on laboratory studies, there is only modest evidence for a relationship between self-reported stress and physiological ambulatory measures. This descriptive systematic review evaluates the evidence for studies investigating an association between self-reported stress and physiological measures under daily life conditions.

METHODS

Three databases were searched for articles assessing an association between self-reported stress and cardiovascular and skin conductance measures simultaneously over the course of at least a day.

RESULTS

We reviewed findings of 36 studies investigating an association between self-reported stress and cardiovascular measures with overall 135 analyses of associations between self-reported stress and cardiovascular measures. Overall, 35% of all analyses showed a significant or marginally significant association in the expected direction. The most consistent results were found for perceived stress, high-arousal negative affect scales, and event-related self-reported stress measures, and for frequency-domain heart rate variability physiological measures. There was much heterogeneity in measures and methods.

CONCLUSION

These findings confirm that daily-life stress-dynamics are complex and require a better understanding. Choices in design and measurement seem to play a role. We provide some guidance for future studies.

Real-Time Assessment of Stress and Stress Response Using Digital Phenotyping: A Study Protocol

Background: Stress is a complex phenomenon that may have a negative influence on health and well-being; consequently, it plays a pivotal role in mental health. Although the incidence of mental disorders has been continuously rising, development of prevention and treatment methods has been rather slow. Through the ubiquitous presence of smartphones and wearable devices, people can monitor stress parameters in everyday life. However, the reliability and validity of such monitoring are still unsatisfactory.

Methods: The aim of this trial is to find a relationship between psychological stress and saliva cortisol levels on the one hand and physiological parameters measured by smartphones in combination with a commercially available wearable device on the other. Participants include cohorts of individuals with and without a psychiatric disorder. The study is conducted in two settings: one naturalistic and one a controlled laboratory environment, combining ecological momentary assessment (EMA) and digital phenotyping (DP). EMA is used for the assessment of challenging and stressful situations coincidentally happening during a whole observation week. DP is used during a controlled stress situation with the Trier Social Stress Test (TSST) as a standardized psychobiological paradigm. Initially, participants undergo a complete psychological screening and profiling using a standardized psychometric test battery. EMA uses a smartphone application, and the participants keep a diary about their daily routine, activities, well-being, sleep, and difficult and stressful situations they may encounter. DP is conducted through wearable devices able to continuously monitor physiological parameters (i.e., heart rate, heart rate variability, skin conductivity, temperature, movement and acceleration). Additionally, saliva cortisol samples are repeatedly taken. The TSST is conducted with continuous measurement of the same parameters measured during the EMA.

Discussion: We aim to identify valid and reliable digital biomarkers for stress and stress reactions. Furthermore, we expect to find a way of early detection of psychological stress in order to evolve new opportunities for interventions reducing stress. That may allow us to find new ways of treating and preventing mental disorders.

Trial Registration: The competing ethics committee of the Canton of Zurich, Switzerland, approved the study protocol V05.1 May 28, 2019 [BASEC: 2019-00814]; the trial was registered at ClinicalTrials.gov [NCT04100213] on September 19, 2019.

Digital Resilience Biomarkers for Personalized Health Maintenance and Disease Prevention

Health maintenance and disease prevention strategies become increasingly prioritized with increasing health and economic burden of chronic, lifestyle-related diseases. A key element in these strategies is the empowerment of individuals to control their health. Self-measurement plays an essential role in achieving such empowerment. Digital measurements have the advantage of being measured non-invasively, passively, continuously, and in a real-world context. An important question is whether such measurement can sensitively measure subtle disbalances in the progression toward disease, as well as the subtle effects of, for example, nutritional improvement. The concept of resilience biomarkers, defined as the dynamic evaluation of the biological response to an external challenge, has been identified as a viable strategy to measure these subtle effects. In this review, we explore the potential of integrating this concept with digital physiological measurements to come to digital resilience biomarkers. Additionally, we discuss the potential of wearable, non-invasive, and continuous measurement of molecular biomarkers. These types of innovative measurements may, in the future, also serve as a digital resilience biomarker to provide even more insight into the personal biological dynamics of an individual. Altogether, digital resilience biomarkers are envisioned to allow for the measurement of subtle effects of health maintenance and disease prevention strategies in a real-world context and thereby give personalized feedback to improve health.

Digital Biomarkers for Depression Screening With Wearable Devices: Cross-sectional Study With Machine Learning Modeling

Background

Depression is a prevalent mental disorder that is undiagnosed and untreated in half of all cases. Wearable activity trackers collect fine-grained sensor data characterizing the behavior and physiology of users (ie, digital biomarkers), which could be used for timely, unobtrusive, and scalable depression screening.

Objective

The aim of this study was to examine the predictive ability of digital biomarkers, based on sensor data from consumer-grade wearables, to detect risk of depression in a working population.

Methods

This was a cross-sectional study of 290 healthy working adults. Participants wore Fitbit Charge 2 devices for 14 consecutive days and completed a health survey, including screening for depressive symptoms using the 9-item Patient Health Questionnaire (PHQ-9), at baseline and 2 weeks later. We extracted a range of known and novel digital biomarkers characterizing physical activity, sleep patterns, and circadian rhythms from wearables using steps, heart rate, energy expenditure, and sleep data. Associations between severity of depressive symptoms and digital biomarkers were examined with Spearman correlation and multiple regression analyses adjusted for potential confounders, including sociodemographic characteristics, alcohol consumption, smoking, self-rated health, subjective sleep characteristics, and loneliness. Supervised machine learning with statistically selected digital biomarkers was used to predict risk of depression (ie, symptom severity and screening status). We used varying cutoff scores from an acceptable PHQ-9 score range to define the depression group and different subsamples for classification, while the set of statistically selected digital biomarkers remained the same. For the performance evaluation, we used k-fold cross-validation and obtained accuracy measures from the holdout folds.

Results

A total of 267 participants were included in the analysis. The mean age of the participants was 33 (SD 8.6, range 21-64) years. Out of 267 participants, there was a mild female bias displayed (n=170, 63.7%). The majority of the participants were Chinese (n=211, 79.0%), single (n=163, 61.0%), and had a university degree (n=238, 89.1%). We found that a greater severity of depressive symptoms was robustly associated with greater variation of nighttime heart rate between 2 AM and 4 AM and between 4 AM and 6 AM; it was also associated with lower regularity of weekday circadian rhythms based on steps and estimated with nonparametric measures of interdaily stability and autocorrelation as well as fewer steps-based daily peaks. Despite several reliable associations, our evidence showed limited ability of digital biomarkers to detect depression in the whole sample of working adults. However, in balanced and contrasted subsamples comprised of depressed and healthy participants with no risk of depression (ie, no or minimal depressive symptoms), the model achieved an accuracy of 80%, a sensitivity of 82%, and a specificity of 78% in detecting subjects at high risk of depression.

Conclusions

Digital biomarkers that have been discovered and are based on behavioral and physiological data from consumer wearables could detect increased risk of depression and have the potential to assist in depression screening, yet current evidence shows limited predictive ability. Machine learning models combining these digital biomarkers could discriminate between individuals with a high risk of depression and individuals with no risk.

Comparing stress prediction models using smartwatch physiological signals and participant self-reports

Recent advances in wearable technology have facilitated the non-obtrusive monitoring of physiological signals, creating opportunities to monitor and predict stress. Researchers have utilized machine learning methods using these physiological signals to develop stress prediction models. Many of these prediction models have utilized objective stressor tasks (e.g., a public speaking task or solving math problems). Alternatively, the subjective user responses with self-reports have also been used for measuring stress. In this paper, we describe a methodological approach (a) to compare the prediction performance of models developed using objective markers of stress using participant-reported subjective markers of stress from self-reports; and (b) to develop personalized stress models by accounting for inter-individual differences. Towards this end, we conducted a laboratory-based study with 32 healthy volunteers. Participants completed a series of stressor tasks-social, cognitive and physical-wearing an instrumented commercial smartwatch that collected physiological signals and participant responses using timed self-reports. After extensive data preprocessing using a combination of signal processing techniques, we developed two types of models: objective stress models using the stressor tasks as labels; and subjective stress models using participant responses to each task as the label for that stress task. We trained and tested several machine learning algorithms-support vector machine (SVM), random forest (RF), gradient boosted trees (GBT), AdaBoost, and Logistic Regression (LR)-and evaluated their performance. SVM had the best performance for the models using the objective stressor (i.e., stressor tasks) with an AUROC of 0.790 and an F-1 score of 0.623. SVM also had the highest performance for the models using the subjective stress (i.e., participant self-reports) with an AUROC of 0.719 and an F-1 score of 0.520. Model performance improved with a personalized threshold model to an AUROC of 0.751 and an F-1 score of 0.599. The performance of the stress models using an instrumented commercial smartwatch was comparable to similar models from other state-of-the-art laboratory-based studies. However, the subjective stress models had a lower performance, indicating the need for further research on the use of self-reports for stress-related studies. The improvement in performance with the personalized threshold-based models provide new directions for building stress prediction models.

A faith-based mind-body intervention to improve psychosocial well-being among rural adults

Churches are well positioned to promote better mental health outcomes in underserved populations, including rural adults. Mind-body (MB) practices improve psychological well-being yet are not widely adopted among faith-based groups due to conflicting religious or practice beliefs. Thus, "Harmony & Health" (HH) was developed as a culturally adapted MB intervention to improve psychosocial health in urban churchgoers and was adapted and implemented in a rural church. The purpose of this study was to explore the feasibility, acceptability, and efficacy of HH to reduce psychosocial distress in rural churchgoers. HH capitalized on an existing church partnership to recruit overweight or obese (body mass index [BMI] ≥25.0 kg/m2) and insufficiently active adults (≥18 years old). Eligible adults participated in an 8 week MB intervention and completed self-reported measures of perceived stress, depressive symptoms, anxiety, and positive and negative affect at baseline and postintervention. Participants (mean [M] age = 49.1 ± 14.0 years) were mostly women (84.8%), non-Hispanic white (47.8%) or African American (45.7%), high socioeconomic status (65.2% completed ≥bachelor degree and 37.2% reported an annual household income ≥$80,000), and obese (M BMI = 32.6 ± 5.8 kg/m2). Participants reported lower perceived stress (t = -2.399, p = .022), fewer depressive symptoms (t = -3.547, p = .001), and lower negative affect (t = -2.440, p = .020) at postintervention. Findings suggest that HH was feasible, acceptable, and effective at reducing psychosocial distress in rural churchgoers in the short-term. HH reflects an innovative approach to intertwining spirituality and MB practices to improve physical and psychological health in rural adults, and findings lend to our understanding of community-based approaches to improve mental health outcomes in underserved populations.

Twenty-Four-Hour Heart Rate Is a Trait but Not State Marker for Depression in a Pilot Randomized Controlled Trial With a Single Infusion of Ketamine

Background: Abnormalities of heart rate (HR) and its variability are characteristic of major depressive disorder (MDD). However, circadian rhythm is rarely taken into account when statistically exploring state or trait markers for depression. Methods: A 4-day electrocardiogram was recorded for 16 treatment-resistant patients with MDD and 16 age- and sex-matched controls before, and for the patient group only, after a single treatment with the rapid-acting antidepressant ketamine or placebo (clinical trial registration available on https://www.clinicaltrialsregister.eu/ with EUDRACT number 2016-001715-21). Circadian rhythm differences of HR and the root mean square of successive differences (RMSSD) were compared between groups and were explored for classification purposes. Baseline HR/RMSSD were tested as predictors for treatment response, and physiological measures were assessed as state markers. Results: Patients showed higher HR and lower RMSSD alongside marked reductions in HR amplitude and RMSSD variation throughout the day. Excellent classification accuracy was achieved using HR during the night, particularly between 2 and 3 a.m. (90.6%). A positive association between baseline HR and treatment response (r = 0.55, p = 0.046) pointed toward better treatment outcome in patients with higher HR. Heart rate also decreased significantly following treatment but was not associated with improved mood after a single infusion of ketamine. Limitations: Our study had a limited sample size, and patients were treated with concomitant antidepressant medication. Conclusion: Patients with depression show a markedly reduced amplitude for HR and dysregulated RMSSD fluctuation. Higher HR and lower RMSSD in depression remain intact throughout a 24-h day, with the highest classification accuracy during the night. Baseline HR levels show potential for treatment response prediction but did not show potential as state markers in this study. Clinical trial registration: EUDRACT number 2016-001715-21.

Consumer Wearables and the Integration of New Objective Measures in Oncology: Patient and Provider Perspectives

With one in five adults in the United States owning a smartwatch or fitness tracker, these devices are poised to impact all aspects of medicine by offering a more objective approach to replace self-reported data. Oncology has proved to be a prototypical example, and wearables offer immediate benefits to patients and oncologists with the ability to track symptoms and health metrics in real time. We aimed to review the recent literature on consumer-grade wearables and its current applications in cancer from the perspective of both the patient and the provider. The relevant studies suggested that these devices offer benefits, such as improved medication adherence and accuracy of symptom tracking over self-reported data, as well as insights that increase patient empowerment. Physical activity is consistently correlated with stronger patient outcomes, and a patient's real-time metrics were found to be capable of tracking medication side effects and toxicity. Studies have made associations between wearable data and telomere shortening, cardiovascular disease, alcohol consumption, sleep apnea, and other conditions. The objective data obtained by the wearable presents a more complete picture of an individual's health than the snapshot of a 15-minute office visit and a single set of vital signs. Real-time metrics can be translated into a digital phenotype that identifies risk factors specific to each patient, and shared risk factors across one's social network may uncover common environmental exposures detrimental to one's health. Wearable data and its upcoming integration with social media will be the foundation for the next generation of personalized medicine.

Digital Phenotyping in Livestock Farming

Currently, large volumes of data are being collected on farms using multimodal sensor technologies. These sensors measure the activity, housing conditions, feed intake, and health of farm animals. With traditional methods, the data from farm animals and their environment can be collected intermittently. However, with the advancement of wearable and non-invasive sensing tools, these measurements can be made in real-time for continuous quantitation relating to clinical biomarkers, resilience indicators, and behavioral predictors. The digital phenotyping of humans has drawn enormous attention recently due to its medical significance, but much research is still needed for the digital phenotyping of farm animals. Implications from human studies show great promise for the application of digital phenotyping technology in modern livestock farming, but these technologies must be directly applied to animals to understand their true capacities. Due to species-specific traits, certain technologies required to assess phenotypes need to be tailored efficiently and accurately. Such devices allow for the collection of information that can better inform farmers on aspects of animal welfare and production that need improvement. By explicitly addressing farm animals' individual physiological and mental (affective states) needs, sensor-based digital phenotyping has the potential to serve as an effective intervention platform. Future research is warranted for the design and development of digital phenotyping technology platforms that create shared data standards, metrics, and repositories.

Examining Response to Negative Life Events Through Fitness Tracker Data

Negative life events, such as the death of a loved one, are an unavoidable part of life. These events can be overwhelmingly stressful and may lead to the development of mental health disorders. To mitigate these adverse developments, prior literature has utilized measures of psychological responses to negative life events to better understand their effects on mental health. However, psychological changes represent only one aspect of an individual's potential response. We posit measuring additional dimensions of health, such as physical health, may also be beneficial, as physical health itself may be affected by negative life events and measuring its response could provide context to changes in mental health. Therefore, the primary aim of this work was to quantify how an individual's physical health changes in response to negative life events by testing for deviations in their physiological and behavioral state (PB-state). After capturing post-event, PB-state responses, our second aim sought to contextualize changes within known factors of psychological response to negative life events, namely coping strategies. To do so, we utilized a cohort of professionals across the United States monitored for 1 year and who experienced a negative life event while under observation. Garmin Vivosmart-3 devices provided a multidimensional representation of one's PB-state by collecting measures of resting heart rate, physical activity, and sleep. To test for deviations in PB-state following negative life events, One-Class Support Vector Machines were trained on a window of time prior to the event, which established a PB-state baseline. The model then evaluated participant's PB-state on the day of the life event and each day that followed, assigning each day a level of deviance relative to the participant's baseline. Resulting response curves were then examined in association with the use of various coping strategies using Bayesian gamma-hurdle regression models. The results from our objectives suggest that physical determinants of health also deviate in response to negative life events and that these deviations can be mitigated through different coping strategies. Taken together, these observations stress the need to examine physical determinants of health alongside psychological determinants when investigating the effects of negative life events.

Smartphone-Based Passive Sensing for Behavioral and Physical Monitoring in Free-Life Conditions: Technical Usability Study

BACKGROUND

Smartphone use is widely spreading in society. Their embedded functions and sensors may play an important role in therapy monitoring and planning. However, the use of smartphones for intrapersonal behavioral and physical monitoring is not yet fully supported by adequate studies addressing technical reliability and acceptance.

OBJECTIVE

The objective of this paper is to identify and discuss technical issues that may impact on the wide use of smartphones as clinical monitoring tools. The focus is on the quality of the data and transparency of the acquisition process.

METHODS

QuantifyMyPerson is a platform for continuous monitoring of smartphone use and embedded sensors data. The platform consists of an app for data acquisition, a backend cloud server for data storage and processing, and a web-based dashboard for data management and visualization. The data processing aims to extract meaningful features for the description of daily life such as phone status, calls, app use, GPS, and accelerometer data. A total of health subjects installed the app on their smartphones, running it for 7 months. The acquired data were analyzed to assess impact on smartphone performance (ie, battery consumption and anomalies in functioning) and data integrity. Relevance of the selected features in describing changes in daily life was assessed through the computation of a k-nearest neighbors global anomaly score to detect days that differ from others.

RESULTS

The effectiveness of smartphone-based monitoring depends on the acceptability and interoperability of the system as user retention and data integrity are key aspects. Acceptability was confirmed by the full transparency of the app and the absence of any conflicts with daily smartphone use. The only perceived issue was the battery consumption even though the trend of battery drain with and without the app running was comparable. Regarding interoperability, the app was successfully installed and run on several Android brands. The study shows that some smartphone manufacturers implement power-saving policies not allowing continuous sensor data acquisition and impacting integrity. Data integrity was 96% on smartphones whose power-saving policies do not impact the embedded sensor management and 84% overall.

CONCLUSIONS

The main technological barriers to continuous behavioral and physical monitoring (ie, battery consumption and power-saving policies of manufacturers) may be overcome. Battery consumption increase is mainly due to GPS triangulation and may be limited, while data missing because of power-saving policies are related only to periods of nonuse of the phone since the embedded sensors are reactivated by any smartphone event. Overall, smartphone-based passive sensing is fully feasible and scalable despite the Android market fragmentation.

Computer Mouse Movements as an Indicator of Work Stress: Longitudinal Observational Field Study

BACKGROUND

Work stress affects individual health and well-being. These negative effects could be mitigated through regular monitoring of employees' stress. Such monitoring becomes even more important as the digital transformation of the economy implies profound changes in working conditions.

OBJECTIVE

The goal of this study was to investigate the association between computer mouse movements and work stress in the field.

METHODS

We hypothesized that stress is associated with a speed-accuracy trade-off in computer mouse movements. To test this hypothesis, we conducted a longitudinal field study at a large business organization, where computer mouse movements from regular work activities were monitored over 7 weeks; the study included 70 subjects and 1829 observations. A Bayesian regression model was used to estimate whether self-reported acute work stress was associated with a speed-accuracy trade-off in computer mouse movements.

RESULTS

There was a negative association between stress and the two-way interaction term of mouse speed and accuracy (mean -0.32, 95% highest posterior density interval -0.58 to -0.08), which means that stress was associated with a speed-accuracy trade-off. The estimated association was not sensitive to different processing of the data and remained negative after controlling for the demographics, health, and personality traits of subjects.

CONCLUSIONS

Self-reported acute stress is associated with computer mouse movements, specifically in the form of a speed-accuracy trade-off. This finding suggests that the regular analysis of computer mouse movements could indicate work stress.

Evaluation, Acceptance, and Qualification of Digital Measures: From Proof of Concept to Endpoint

To support the successful adoption of digital measures into internal decision making and evidence generation for medical product development, we present a unified lexicon to aid communication throughout this process, and highlight key concepts including the critical role of participant engagement in development of digital measures. We detail the steps of bringing a successful proof of concept to scale, focusing on key decisions in the development of a new digital measure: asking the right question, optimized approaches to evaluating new measures, and whether and how to pursue qualification or acceptance. Building on the V3 framework for establishing verification and analytical and clinical validation, we discuss strategic and practical considerations for collecting this evidence, illustrated with concrete examples of trailblazing digital measures in the field.

Conditional Entropy: A Potential Digital Marker for Stress

Recent decades have witnessed a substantial progress in the utilization of brain activity for the identification of stress digital markers. In particular, the success of entropic measures for this purpose is very appealing, considering (1) their suitability for capturing both linear and non-linear characteristics of brain activity recordings and (2) their direct association with the brain signal variability. These findings rely on external stimuli to induce the brain stress response. On the other hand, research suggests that the use of different types of experimentally induced psychological and physical stressors could potentially yield differential impacts on the brain response to stress and therefore should be dissociated from more general patterns. The present study takes a step toward addressing this issue by introducing conditional entropy (CE) as a potential electroencephalography (EEG)-based resting-state digital marker of stress. For this purpose, we use the resting-state multi-channel EEG recordings of 20 individuals whose responses to stress-related questionnaires show significantly higher and lower level of stress. Through the application of representational similarity analysis (RSA) and K-nearest-neighbor (KNN) classification, we verify the potential that the use of CE can offer to the solution concept of finding an effective digital marker for stress.

Current strategies with sensing technologies to eliminate stress cardiomyopathy

Stress cardiomyopathy refers weakening of heart muscle due to the continuous stress. Generally, the severe status of stress cardiomyopathy has been revealed after damaging the muscles and measured by the physical changes in the heart system. To overcome this issue, biosensor can be used, which could eliminate the late identification stress cardiomyopathy. With biosensors, different stress markers such as epinephrine, dopamine, catecholamine, α-amylase, norepinephrine, serotonin and cortisol have been identified by a wide range of developments. These biosensors are available from laboratory to industry at the ranges of nano to macrodevices. To merge with the identification of stress cardiomyopathy, the above strategies might be utilized properly and can aid to reduce the stress-related problems. This overview gleaned the currently available biosensing methods and the associated biomarkers at various stages of the developments and implementations of stress cardiomyopathy.

Investigating the stress-related fluctuations of level of personality functioning: A critical review and agenda for future research

The Diagnostic and Statistical Manual of Mental Disorders (DSM-5) and the International Classification of Diseases (ICD-11) proposed a dimensional approach to the assessment of personality disorders (PDs). Both models dictate that the clinician first determines PD severity before assessing maladaptive traits, invoking the level of personality functioning (LPF) construct. We consider LPF a promising dimensional construct for translational research because of its clinical importance and conceptual overlap with the Research Domain Criteria (RDoC) Social Processes. We aim to identify biomarkers that co-vary with fluctuations in LPF in adulthood, ultimately to predict persistent decrease in LPF, associated with suicidality and morbidity. However, a theoretical framework to investigate stress-related oscillations in LPF is currently missing. In this article, we aim to fill this hiatus with a critical review about stress and LPF. First, we discuss acute stress and LPF. We briefly present the basics of the neurophysiological stress response and review the literature on momentary and daily fluctuations in LPF, both at a subjective and physiological level. Second, we review the effects of chronic stress on brain function and social behaviour and recapitulate the main findings from prospective cohort studies. This review underlies our suggestions for multimethod assessment of stress-related oscillations in LPF and our theoretical framework for future longitudinal studies, in particular studies using the experience sampling method (ESM).