RESUMEN

Migraine is a chronic headache disorder. A prototype mobile app-based system was implemented to test data collection and improve data coverage for the Empatica E4 biometric sensor device. Results from the prototype testing are reported. Future iteration of the app will be tested with patients with migraine to predict events and potentially reduce event duration and therefore disease burden.

Asunto(s)

RESUMEN

Cometrics is a Microsoft Windows compatible clinical tool for the collection and recording of frequency- and duration-based target behaviors, physiological signals, and video data. This software package is designed to record in-vivo observational and physiological data. In addition, we have included features that allow observers to capture video from real-time camera feeds and import saved video for retroactive data collection. By using Microsoft Excel-based spreadsheets, also called keystroke files, assessment and treatment sessions are exported into a single document using the click of a button. Integrated interobserver agreement metrics allow comparisons across primary and reliability observers, with the output exported into a spreadsheet for easy reference. All file system interactions are handled by the user interface, so files and folders are created and managed without manual intervention. This software is available free-of-charge through the Microsoft Store for Windows 10 and 11 and the source code is publicly available on GitHub. Supplementary Information: The online version contains supplementary material available at 10.1007/s40617-023-00817-w.

RESUMEN

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.

Asunto(s)

RESUMEN

BACKGROUND: Physiological signals such as heart rate and electrodermal activity can provide insight into an individual's mental state, which are invaluable information for mental health care. Using recordings of physiological signals from wearable devices in the wild can facilitate objective monitoring of symptom severity and evaluation of treatment progress. OBJECTIVE: We designed a study to evaluate the feasibility of predicting obsessive-compulsive disorder (OCD) events from physiological signals recorded using wrist-worn devices in the wild. Here, we present an analysis plan for the study to document our a priori hypotheses and increase the robustness of the findings of our planned study. METHODS: In total, 18 children and adolescents aged between 8 and 16 years were included in this study. Nine outpatients with an OCD diagnosis were recruited from a child and adolescent mental health center. Nine youths without a psychiatric diagnosis were recruited from the catchment area. Patients completed a clinical interview to assess OCD severity, types of OCD, and number of OCD symptoms in the clinic. Participants wore a biosensor on their wrist for up to 8 weeks in their everyday lives. Patients were asked to press an event tag button on the biosensor when they were stressed by OCD symptoms. Participants without a psychiatric diagnosis were asked to press this button whenever they felt really scared. Before and after the 8-week observation period, participants wore the biosensor under controlled conditions of rest and stress in the clinic. Features are extracted from 4 different physiological signals within sliding windows to predict the distress event logged by participants during data collection. We will test the prediction models within participants across time and multiple participants. Model selection and estimation using 2-layer cross-validation are outlined for both scenarios. RESULTS: Participants were included between December 2021 and December 2022. Participants included 10 female and 8 male participants with an even sex distribution between groups. Patients were aged between 10 and 16 years, and adolescents without a psychiatric diagnosis were between the ages of 8 and 16 years. Most patients had moderate to moderate to severe OCD, except for 1 patient with mild OCD. CONCLUSIONS: The strength of the planned study is the investigation of predictions of OCD events in the wild. Major challenges of the study are the inherent noise of in-the-wild data and the lack of contextual knowledge associated with the recorded signals. This preregistered analysis plan discusses in detail how we plan to address these challenges and may help reduce interpretation bias of the upcoming results. If the obtained results from this study are promising, we will be closer to automated detection of OCD events outside of clinical experiments. This is an important tool for the assessment and treatment of OCD in youth. TRIAL REGISTRATION: ClinicalTrials.gov NCT05064527; https://clinicaltrials.gov/study/NCT05064527. INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): DERR1-10.2196/48571.

RESUMEN

In response to challenging circumstances, the human body can experience marked levels of anxiety and distress. To prevent stress-related complications, timely identification of stress symptoms is crucial, necessitating the need for continuous stress monitoring. Wearable devices offer a means of real-time and ongoing data collection, facilitating personalized stress monitoring. Based on our protocol for data pre-processing, this study proposes to analyze signals obtained from the Empatica E4 bracelet using machine-learning algorithms (Random Forest, SVM, and Logistic Regression) to determine the efficacy of the abovementioned techniques in differentiating between stressful and non-stressful situations. Photoplethysmographic and electrodermal activity signals were collected from 29 subjects to extract 27 features which were then fed into three different machine-learning algorithms for binary classification. Using MATLAB after applying the chi-square test and Pearson's correlation coefficient on WEKA for features' importance ranking, the results demonstrated that the Random Forest model has the highest stability (accuracy of 76.5%) using all the features. Moreover, the Random Forest applying the chi-test for feature selection reached consistent results in terms of stress evaluation based on precision, recall, and F1-measure (71%, 60%, 65%, respectively).

Asunto(s)

RESUMEN

BACKGROUND: Depressive and manic episodes within bipolar disorder (BD) and major depressive disorder (MDD) involve altered mood, sleep, and activity, alongside physiological alterations wearables can capture. OBJECTIVE: Firstly, we explored whether physiological wearable data could predict (aim 1) the severity of an acute affective episode at the intra-individual level and (aim 2) the polarity of an acute affective episode and euthymia among different individuals. Secondarily, we explored which physiological data were related to prior predictions, generalization across patients, and associations between affective symptoms and physiological data. METHODS: We conducted a prospective exploratory observational study including patients with BD and MDD on acute affective episodes (manic, depressed, and mixed) whose physiological data were recorded using a research-grade wearable (Empatica E4) across 3 consecutive time points (acute, response, and remission of episode). Euthymic patients and healthy controls were recorded during a single session (approximately 48 h). Manic and depressive symptoms were assessed using standardized psychometric scales. Physiological wearable data included the following channels: acceleration (ACC), skin temperature, blood volume pulse, heart rate (HR), and electrodermal activity (EDA). Invalid physiological data were removed using a rule-based filter, and channels were time aligned at 1-second time units and segmented at window lengths of 32 seconds, as best-performing parameters. We developed deep learning predictive models, assessed the channels' individual contribution using permutation feature importance analysis, and computed physiological data to psychometric scales' items normalized mutual information (NMI). We present a novel, fully automated method for the preprocessing and analysis of physiological data from a research-grade wearable device, including a viable supervised learning pipeline for time-series analyses. RESULTS: Overall, 35 sessions (1512 hours) from 12 patients (manic, depressed, mixed, and euthymic) and 7 healthy controls (mean age 39.7, SD 12.6 years; 6/19, 32% female) were analyzed. The severity of mood episodes was predicted with moderate (62%-85%) accuracies (aim 1), and their polarity with moderate (70%) accuracy (aim 2). The most relevant features for the former tasks were ACC, EDA, and HR. There was a fair agreement in feature importance across classification tasks (Kendall W=0.383). Generalization of the former models on unseen patients was of overall low accuracy, except for the intra-individual models. ACC was associated with "increased motor activity" (NMI>0.55), "insomnia" (NMI=0.6), and "motor inhibition" (NMI=0.75). EDA was associated with "aggressive behavior" (NMI=1.0) and "psychic anxiety" (NMI=0.52). CONCLUSIONS: Physiological data from wearables show potential to identify mood episodes and specific symptoms of mania and depression quantitatively, both in BD and MDD. Motor activity and stress-related physiological data (EDA and HR) stand out as potential digital biomarkers for predicting mania and depression, respectively. These findings represent a promising pathway toward personalized psychiatry, in which physiological wearable data could allow the early identification and intervention of mood episodes.

Asunto(s)

RESUMEN

Lab research might benefit from the advantages of wearable devices, such as their ease of use, to estimate pulse rate (PR) and pulse rate variability (PRV) as an equivalent for heart rate (HR) and heart rate variability. However, before implementing them in a lab context, the validity of the PR and PRV, also on ultra-short time scales (e.g., 30s), needs to be confirmed. We recorded heart activity simultaneously with an E4 wristband and an ECG device in a seated resting condition for 5 min. Our results showed that HR, RMSSD, SDNN and LF, but not HF, were validly estimated by the E4 wristband. Furthermore, the E4 wristband could validly estimate PR with recording lengths as short as 10s. RMSSD and SDNN were validly estimated using 30s or 120 s or an average of multiple short intervals (10s), while HF likely requires longer recording intervals. Based on this study, we formulated several recommendations for using the E4 wristband in a lab context.

Asunto(s)

RESUMEN

Portable and wireless devices that collect physiological data are becoming more and more sought after in clinical and psychophysiological research as technology swiftly advances. These devices allow for data collection in interactive states, such as dyadic therapy, with reduced restraints compared to traditional laboratory devices. One such portable device is the Empatica E4 wristband (Empatica Srl, Milan, Italy) which allows quantifying cardiac interbeat intervals (IBIs), heart rate variability (HRV), and electro-dermal activity (EDA), as well as several other acceleration and temperature measures. In the current study, we aimed to assess IBI, HRV, and EDA measures, against the same data collected from the well-validated MindWare mobile impedance cardiograph device (MindWare Technology, Gahanna, OH, United States). We assessed the E4 strictly as a research instrument and not as a clinical tool. We were specifically interested in the wristbands' performance during naturalistic interactive face-to-face conversations which inherently involve more hand movements. We collected data from 30 participants, nested in 15 dyads, which were connected to both devices simultaneously, during rest and during a social conversation. After preprocessing and analyses, we found that mean IBIs obtained by the E4 and the MindWare device, were highly similar during rest and during conversation. Medium to high correlations were found between the devices with respect to several HRV measures, with higher correlations during rest compared to conversation. The E4 failed to produce reliable EDA data. We conclude by discussing the strengths and limitations of the E4 during seated conversational states and suggest optimal ways to collect and analyze data with the E4.

RESUMEN

Fibromyalgia is a medical condition characterized by widespread muscle pain and tenderness and is often accompanied by fatigue and alteration in sleep, mood, and memory. Poor sleep quality and fatigue, as prominent characteristics of fibromyalgia, have a direct impact on patient behavior and quality of life. As such, the detection of extreme cases of sleep quality and fatigue level is a prerequisite for any intervention that can improve sleep quality and reduce fatigue level for people with fibromyalgia and enhance their daytime functionality. In this study, we propose a new supervised machine learning method called Learning Using Concave and Convex Kernels (LUCCK). This method employs similarity functions whose convexity or concavity can be configured so as to determine a model for each feature separately, and then uses this information to reweight the importance of each feature proportionally during classification. The data used for this study was collected from patients with fibromyalgia and consisted of blood volume pulse (BVP), 3-axis accelerometer, temperature, and electrodermal activity (EDA), recorded by an Empatica E4 wristband over the courses of several days, as well as a self-reported survey. Experiments on this dataset demonstrate that the proposed machine learning method outperforms conventional machine learning approaches in detecting extreme cases of poor sleep and fatigue in people with fibromyalgia.