Ensemble machine learning model trained on a new synthesized dataset generalizes well for stress prediction using wearable devices.

INTRODUCTION: Advances in wearable sensor technology have enabled the collection of biomarkers that may correlate with levels of elevated stress. While significant research has been done in this domain, specifically in using machine learning to detect elevated levels of stress, the challenge of producing a machine learning model capable of generalizing well for use on new, unseen data remain. Acute stress response has both subjective, psychological and objectively measurable, biological components that can be expressed differently from person to person, further complicating the development of a generic stress measurement model. Another challenge is the lack of large, publicly available datasets labeled for stress response that can be used to develop robust machine learning models. In this paper, we first investigate the generalization ability of models built on datasets containing a small number of subjects, recorded in single study protocols. Next, we propose and evaluate methods combining these datasets into a single, large dataset to study the generalization capability of machine learning models built on larger datasets. Finally, we propose and evaluate the use of ensemble techniques by combining gradient boosting with an artificial neural network to measure predictive power on new, unseen data. In favor of reproducible research and to assist the community advance the field, we make all our experimental data and code publicly available through Github at https://github.com/xalentis/Stress. This paper's in-depth study of machine learning model generalization for stress detection provides an important foundation for the further study of stress response measurement using sensor biomarkers, recorded with wearable technologies. METHODS: Sensor biomarker data from six public datasets were utilized in this study. Exploratory data analysis was performed to understand the physiological variance between study subjects, and the complexity it introduces in building machine learning models capable of detecting elevated levels of stress on new, unseen data. To test model generalization, we developed a gradient boosting model trained on one dataset (SWELL), and tested its predictive power on two datasets previously used in other studies (WESAD, NEURO). Next, we merged four small datasets, i.e. (SWELL, NEURO, WESAD, UBFC-Phys), to provide a combined total of 99 subjects, and applied feature engineering to generate additional features utilizing statistical summaries, with sliding windows of 25 s. We name this large dataset, StressData. In addition, we utilized random sampling on StressData combined with another dataset (EXAM) to build a larger training dataset consisting of 200 synthesized subjects, which we name SynthesizedStressData. Finally, we developed an ensemble model that combines our gradient boosting model with an artificial neural network, and tested it using Leave-One-Subject-Out (LOSO) validation, and on two additional, unseen publicly available stress biomarker datasets (WESAD and Toadstool). RESULTS: Our results show that previous models built on datasets containing a small number (<50) of subjects, recorded in single study protocols, cannot generalize well to new, unseen datasets. Our presented methodology for generating a large, synthesized training dataset by utilizing random sampling to construct scenarios closely aligned with experimental conditions demonstrate significant benefits. When combined with feature-engineering and ensemble learning, our method delivers a robust stress measurement system capable of achieving 85% predictive accuracy on new, unseen validation data, achieving a 25% performance improvement over single models trained on small datasets. The resulting model can be used as both a classification or regression predictor for estimating the level of perceived stress, when applied on specific sensor biomarkers recorded using a wearable device, while further allowing researchers to construct large, varied datasets for training machine learning models that closely emulate their exact experimental conditions. CONCLUSION: Models trained on small, single study protocol datasets do not generalize well for use on new, unseen data and lack statistical power. Machine learning models trained on a dataset containing a larger number of varied study subjects capture physiological variance better, resulting in more robust stress detection. Feature-engineering assists in capturing these physiological variance, and this is further improved by utilizing ensemble techniques by combining the predictive power of different machine learning models, each capable of learning unique signals contained within the data. While there is a general lack of large, labeled public datasets that can be utilized for training machine learning models capable of accurately measuring levels of acute stress, random sampling techniques can successfully be applied to construct larger, varied datasets from these smaller sample datasets, for building robust machine learning 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.

The role of combined modifiable lifestyle behaviors in the longitudinal association between stressful life events and allostatic load in Australian adults.

Allostatic load is a model that is used to quantify the physiological damage from exposure to stressors. Stressful life events are chronic stressors that can lead to an elevated allostatic load through the physiological and behavioral stress responses. However, there is limited empirical studies that has tested the proposed behavioural pathway. Our study addresses this gap by examining the mediating role of combined modifiable lifestyle behaviors in the 12-years longitudinal association between stressful life events and allostatic load among participants from the Australian Diabetes, Obesity and Lifestyle (AusDiab) Study cohort. A latent profile analysis was performed to identify latent subgroups with distinct behavioral clusters based on five modifiable lifestyle behaviors (smoking, sedentary behavior, physical activity, alcohol consumption, and diet quality). We then used a sequential mediation model design with path analysis to test the mediating effect of these latent subgroups in the associations between stressful life events and three measures of allostatic load. Indirect effects were estimated using the product of coefficient approach and the statistical significance was determined by the 95% bias-corrected bootstrap confidence intervals with 1000 replications. We identified three latent subgroups: "least healthy lifestyle" (12%; n = 396), "moderately healthy lifestyle" (78.7%; n = 2599), and "most healthy lifestyle" (9.2%; n = 306). Exposure to stressful life events was not associated with the allocation of participants in latent subgroups. Compared to the "moderately healthy lifestyle" subgroups, we found that the "least healthy lifestyle" behavioral cluster was not associated with allostatic load. However, there was a significant inverse association between the "most healthy lifestyle" behavioral cluster and allostatic load. Overall, we did not find significant indirect effects between stressful life events and three measures of allostatic load via the "least healthy lifestyle" and the "most healthy lifestyle" groups. In summary, the combinations of modifiable lifestyle behaviors did not explain the association between stressful life events and allostatic load. More longitudinal studies are needed to replicate our study to confirm this finding.

A Cross-Sectional Study Investigating Canadian and Australian Adolescents' Perceived Experiences of COVID-19: Gender Differences and Mental Health Implications.

The coronavirus (COVID-19) disease pandemic has been associated with adverse psychological outcomes. This cross-cultural study (N = 1326, 71% female) aimed to investigate Canadian and Australian adolescents' subjective experiences of COVID-19, gender differences, and psychological implications. Mixed-methods analyses were used to examine differences in COVID-19 experiences and mental health outcomes between country and gender in a Canadian (N = 913, 78% female) and an Australian sample (N = 413, 57% female) of adolescents. Canadian adolescents reported increased COVID-19 discussions and more concerns related to their COVID-19 experiences compared to Australian adolescents. Girls consistently reported more concerns related to COVID-19 and poorer psychological outcomes compared to boys. School lockdown for the Canadian sample may have played a role in these country differences. Further, girls might be at significantly more risk for mental health concerns during COVID-19, which should be considered in adolescent mental health initiatives during the pandemic. Although school disruption and separation of peers due to the pandemic likely have a role in adolescent perceived stressors and mental health, the differences between Canadian and Australian adolescents were less clear and future investigations comparing more objective pre-COVID-19 data to current data are needed.

Adverse childhood experiences and allostatic load: A systematic review.

Adverse Childhood Experiences (ACEs) are stressful and/or traumatic experiences associated with an increased lifetime risk of negative health outcomes. The Allostatic Load (AL) is a measure of multisystem dysregulation, resulted by chronic stress. We systematically reviewed the English language literature on the association between ACEs and AL to identify the clinical risk profile, with the exclusion of reviews and preclinical studies. Searches covered the publication period up to the 1st of February 2022 and identified 25 studies in which ACEs such as maltreatment, abuse, poverty, psychological abuse, and discrimination were investigated in the context of AL. The selected studies used different sets of AL biomarkers resulting in substantial heterogenicity of calculating the AL index. Overall, we found that ACEs are associated with elevated AL and poorer health outcomes in adulthood. Furthermore, health risk behaviors, social support, and coping resources either moderate or mediate this association. These findings suggest that targeting individuals at risk and starting interventions early might reduce AL and its deleterious health consequences.

Allostatic load and systemic comorbidities in psychiatric disorders.

Psychiatric disorders are complex, disabling, and chronic conditions that are often accompanied by one or more systemic medical comorbidities. In this narrative review, we provide an overview of the allostatic load concept, which represents a multi-system dysregulation in response to chronic stress and link it to systemic comorbidities associated with psychiatric disorders. We synthesized published literature gathered using Medline (Ovid), Scopus, and PsychInfo and identified a high frequency of systemic comorbidities for both mood and psychotic disorders. The identified cardiovascular, metabolic, and immune comorbidities may represent the result of chronic wear and tear caused by a complex interaction between chronic psychosocial stress, health risk behaviors, pharmacological stressors, and the biological systems involved in the development of allostatic load. These findings support the notion that psychiatric disorders should be re-conceptualized as systemic disorders, affecting the brain and systemic biological pathways in an interconnected fashion to result in systemic comorbidities. We suggest that the multi-systemic and multi-dimensional approach that drives the allostatic load concept should be considered for understanding comorbidities in vulnerable psychiatric patients.

Schizophrenia: a disorder of broken brain bioenergetics.

A substantial and diverse body of literature suggests that the pathophysiology of schizophrenia is related to deficits of bioenergetic function. While antipsychotics are an effective therapy for the management of positive psychotic symptoms, they are not efficacious for the complete schizophrenia symptom profile, such as the negative and cognitive symptoms. In this review, we discuss the relationship between dysfunction of various metabolic pathways across different brain regions in relation to schizophrenia. We contend that several bioenergetic subprocesses are affected across the brain and such deficits are a core feature of the illness. We provide an overview of central perturbations of insulin signaling, glycolysis, pentose-phosphate pathway, tricarboxylic acid cycle, and oxidative phosphorylation in schizophrenia. Importantly, we discuss pharmacologic and nonpharmacologic interventions that target these pathways and how such interventions may be exploited to improve the symptoms of schizophrenia.

Cannabidiol for at risk for psychosis youth: A randomized controlled trial.

BACKGROUND: No biological treatment has been firmly established for the at-risk stage of psychotic disorder. In this study we aim to test if subthreshold psychotic symptoms can be effectively treated with cannabidiol (CBD), a non-psychoactive compound of the plant Cannabis sativa. The question has taken on increased importance in the wake of evidence questioning both the need and efficacy of specific pharmacological interventions in the ultra-high risk (UHR) for psychosis group. METHODS: Three-arm randomized controlled trial of 405 patients (135 per arm) aged 12-25 years who meet UHR for psychosis criteria. The study includes a 6-week lead-in phase during which 10% of UHR individuals are expected to experience symptom remission. Participants will receive CBD (per oral) at doses 600 or 1000 mg per day (fixed schedule) for 12 weeks. Participants in the third arm of the trial will receive matching placebo capsules. Primary outcome is severity of positive psychotic symptoms as measured by the Comprehensive Assessment of At-Risk Mental States at 12 weeks. We hypothesize that CBD will be significantly more effective than placebo in improving positive psychotic symptoms in UHR patients. All participants will also be followed up 6 months post baseline to evaluate if treatment effects are sustained. CONCLUSION: This paper reports on the rationale and protocol of the Cannabidiol for At Risk for psychosis Youth (CanARY) study. This study will test CBD for the first time in the UHR phase of psychotic disorder.

Childhood adversity, allostatic load, and adult mental health: Study protocol using the Avon Longitudinal Study of Parents and Children birth cohort.

Introduction: The cumulative burden of chronic stress and life events has been termed allostatic load. Elevated allostatic load indices are associated with different mental health conditions in adulthood. To date, however, the association between elevated allostatic load in childhood and later development of mental health conditions has not been investigated. Methods: Using data from the Avon Longitudinal Study of Parents and Children (ALSPAC), we will calculate allostatic load indices using biomarkers representing the cardiovascular, metabolic, immune, and neuroendocrine systems, at the ages of 9 and 17 years. Bivariate and multivariable logistic regression models will be used to investigate the association between allostatic load and psychiatric disorders in adulthood. Furthermore, the role of adverse childhood experiences as a modifier will be investigated. Discussion: This protocol describes a strategy for investigating the association between elevated allostatic load indices in childhood at the age of 9 years old and psychiatric disorders in adulthood at 24 years old.

Building brain capital.

Brains are indispensable drivers of human progress. Why not invest more heavily in them? We seek to place Brain Capital at the center of a new narrative to fuel economic and societal recovery and resilience.

Within-treatment changes in a novel addiction treatment program using traditional Amazonian medicine.

AIMS: The therapeutic use of psychedelics is regaining scientific momentum, but similarly psychoactive ethnobotanical substances have a long history of medical (and other) uses in indigenous contexts. Here we aimed to evaluate patient outcomes in a residential addiction treatment center that employs a novel combination of Western and traditional Amazonian methods. METHODS: The study was observational, with repeated measures applied throughout treatment. All tests were administered in the center, which is located in Tarapoto, Peru. Data were collected between 2014 and 2015, and the study sample consisted of 36 male inpatients who were motivated to seek treatment and who entered into treatment voluntarily. Around 58% of the sample was from South America, 28% from Europe, and the remaining 14% from North America. We primarily employed repeated measures on a psychological test battery administered throughout treatment, measuring perceived stress, craving frequency, mental illness symptoms, spiritual well-being, and physical and emotional health. Addiction severity was measured on intake, and neuropsychological performance was assessed in a subsample from intake to at least 2 months into treatment. RESULTS: Statistically significant and clinically positive changes were found across all repeated measures. These changes appeared early in the treatment and were maintained over time. Significant improvements were also found for neuropsychological functioning. CONCLUSION: These results provide evidence for treatment safety in a highly novel addiction treatment setting, while also suggesting positive therapeutic effects.

Human nail cortisol as a retrospective biomarker of chronic stress: A systematic review.

Cortisol is the primary glucocorticoid produced by the activation of the hypothalamic pituitary adrenal (HPA) axis after a psychological or physiological stressor. The dysregulation of the HPA axis by chronic stress has been associated with psychiatric disorders. Although hair is currently the main validated source of chronic cortisol concentrations, cortisol is also bound to human nails, another keratinised matrix. Therefore, nail cortisol has the potential to be an alternative retrospective chronic measure of HPA activation. The aim of this systematic review was to assess the temporal resolution, methodological issues, HPA correlates, and target populations in nail cortisol investigations. A qualitative synthesis was performed to assess current literature exploring cortisol concentrations from human nails. A total of 18 eligible human studies extracted from Medline (PubMed and Ovid), ProQuest (PsycINFO), and Scopus found that immunoassays and mass spectrometry were the two primarily methods of analysis. However, methodological variability remained evident between studies. Nail cortisol correlated with saliva and hair in some studies and was investigated across multiple developmental periods. Finally, when applied as an outcome measure in health disorders, higher nail cortisol concentrations have been shown to be associated with acute coronary syndrome and depression. In conclusion, nail cortisol may serve as a retrospective biomarker of chronic stress; however, the ability to track how much cortisol is accumulating within nail clippings is complex and may represent a large timespan. Further, very few studies have reported effect sizes and investigated the effects of covariates, such as age, sex, ethnicity, and nail characteristics, which limits the validation of this measure. Further studies are required to validate the utility of nail cortisol as a biomarker of chronic stress across the human lifespan.

Understanding the pathology of psychiatric disorders in refugees.

Displacement of people from their homes, families and countries is a current global crisis, with over 70 million people forcibly on the move. A substantial proportion of these people will end up in regions with a different language and culture, where they are registered as refugees or asylum seekers. Due to the underlying reasons for displacement (including conflicts, persecution or violation of human rights), displaced people are severely stress-exposed, which continues into their post-migration life and increases risk for developing psychiatric disorders such as post-traumatic stress disorder and other anxiety disorders and mood disorders. While landmark studies have illustrated the increased prevalence of psychopathology in asylum seeker and refugee populations following pre-/post-displacement stress, few studies add to our understanding of the basic biological mechanisms underpinning risk to psychiatric disorders in these populations. Additionally, the mechanisms underlying resilience despite significant adversity remain unclear. Understanding the molecular mechanisms underpinning the development of psychiatric disorders in refugees can propel treatments (both drug and non-drug) that are capable of influencing biology at the molecular level, and the design of interventions. In the following review, we summarise the status quo of research investigating the pathophysiology of psychiatric disorders in refugees, and propose new ways to address gaps in knowledge with multidisciplinary research.

The Gut Microbiome in Psychosis From Mice to Men: A Systematic Review of Preclinical and Clinical Studies.

The gut microbiome is rapidly becoming the focus of interest as a possible factor involved in the pathophysiology of neuropsychiatric disorders. Recent understanding of the pathophysiology of schizophrenia emphasizes the role of systemic components, including immune/inflammatory and metabolic processes, which are influenced by and interacting with the gut microbiome. Here we systematically review the current literature on the gut microbiome in schizophrenia-spectrum disorders and in their animal models. We found that the gut microbiome is altered in psychosis compared to healthy controls. Furthermore, we identified potential factors related to psychosis, which may contribute to the gut microbiome alterations. However, further research is needed to establish the disease-specificity and potential causal relationships between changes of the microbiome and disease pathophysiology. This can open up the possibility of. manipulating the gut microbiome for improved symptom control and for the development of novel therapeutic approaches in schizophrenia and related psychotic disorders.

Protocol for the Use of the Ketogenic Diet in Preclinical and Clinical Practice.

Many age-related diseases are associated with metabolic abnormalities, and dietary interventions may have some benefit in alleviating symptoms or in delaying disease onset. Here, we review the commonly used best practices involved in applications of the ketogenic diet to facilitate its translation into clinical use. The findings reveal that better education of physicians is essential for applying the optimum diet and monitoring its effects in clinical practice. In addition, investigators should carefully consider potential confounding factors prior to commencing studies involving a ketogenic diet. Most importantly, current studies should improve their reporting on ketone levels as well as on the intake of both macro- and micronutrients. Finally, more detailed studies on the mechanism of action are necessary to help identify potential biomarkers for response prediction and monitoring, and to uncover new drug targets to aid the development of novel treatments.

Testing the Effects of Dietary Seafood Consumption on Depressive Symptoms.

This chapter presents a protocol for assessing the effects dietary seafood consumption on depressive symptoms. We designed a cross-sectional study of 206 participants recruited in two Torres Strait Island communities. Depressive symptoms were assessed using the adapted Patient Health Questionnaire-9 (aPHQ-9), diet was analyzed with a structured questionnaire, omega-3 and omega-6 fatty acid concentrations were measured via a capillary dried blood spot system, and plasma levels of triglycerides and cholesterol were measured by gas-phase chromatography. Finally, we tested the relationship between seafood consumption, blood lipid concentrations, and depression scores using independent samples t-tests and a logistic and quantile regression model.

Ketogenic therapy in neurodegenerative and psychiatric disorders: From mice to men.

Ketogenic diet is a low carbohydrate and high fat diet that has been used for over 100 years in the management of childhood refractory epilepsy. More recently, ketogenic diet has been investigated for a number of metabolic, neurodegenerative and neurodevelopmental disorders. In this comprehensive review, we critically examine the potential therapeutic benefits of ketogenic diet and ketogenic agents on neurodegenerative and psychiatric disorders in humans and translationally valid animal models. The preclinical literature provides strong support for the efficacy of ketogenic diet in a variety of diverse animal models of neuropsychiatric disorders. However, the evidence from clinical studies, while encouraging, particularly in Alzheimer's disease, psychotic and autism spectrum disorders, is limited to case studies and small pilot trials. Firm conclusion on the efficacy of ketogenic diet in psychiatric disorders cannot be drawn due to the lack of randomised, controlled clinical trials. The potential mechanisms of action of ketogenic therapy in these disorders with diverse pathophysiology may include energy metabolism, oxidative stress and immune/inflammatory processes. In conclusion, while ketogenic diet and ketogenic substances hold promise pre-clinically in a variety of neurodegenerative and psychiatric disorders, further studies, particularly randomised controlled clinical trials, are warranted to better understand their clinical efficacy and potential side effects.

Effects of beta-hydroxybutyrate administration on MK-801-induced schizophrenia-like behaviour in mice.

RATIONALE: Impaired cerebral glucose metabolism is a core pathological feature of schizophrenia. We recently demonstrated that a ketogenic diet, causing a shift from glycolysis to ketosis, normalized schizophrenia-like behaviours in an acute N-methyl-D-aspartate (NMDA) receptor antagonist model of the illness. Ketogenic diet produces the ketone body, ß-hydroxybutyrate (BHB), which may serve as an alternative fuel source in its own right without a strict dietary regime. OBJECTIVE: We hypothesized that chronic administration of BHB replicates the therapeutic effects of ketogenic diet in an acute NMDA receptor hypofunction model of schizophrenia in mice. METHODS: C57Bl/6 mice were either treated with acute doses of 2 mmol/kg, 10 mmol/kg, or 20 mmol/kg BHB or received daily intraperitoneal injections of 2 mmol/kg BHB or saline for 3 weeks. Behavioural testing assessed the effect of acute challenge with 0.2 mg/kg MK-801 or saline on open field behaviour, social interaction, and prepulse inhibition of startle (PPI). RESULTS: Acute BHB administration dose-dependently increased BHB plasma levels, whereas the 2 mmol/kg dose increased plasma glucose levels. The highest acute dose of BHB supressed spontaneous locomotor activity, MK-801-induced locomotor hyperactivity and MK-801-induced disruption of PPI. Chronic BHB treatment normalized MK-801-induced hyperlocomotion, reduction of sociability, and disruption of PPI. CONCLUSION: In conclusion, BHB may present a novel treatment option for patients with schizophrenia by providing an alternative fuel source to normalize impaired glucose metabolism in the brain.