Abstract:Stress has been identified as one of the major causes of automobile crashes which then lead to high rates of fatalities and injuries each year. Stress can be measured via physiological measurements and in this study the focus will be based on the features that can be extracted by common wearable devices. Hence, the study will be mainly focusing on heart rate variability (HRV). This study is aimed at investigating the role of HRV-derived features as stress markers. This is achieved by developing a good predicti… Show more
“…There are several machine learning approaches that have explored physiologically derived markers in general for stress recognition (MacLaughlin et al, 2011;Dhama et al, 2019;Šalkevicius et al, 2019). From feature-based machine learning paradigms which classify various features extracted from wearable sensors, i. e., sleep quality, and percentage of screen time (Sano and Picard, 2013), or heart rate variability (HRV) (Dalmeida and Masala, 2021), and thermal-video recognition of the Initial Systolic Time Interval (Kumar S. et al, 2021), applying the state-of-the-art StressNet. StressNet consists of a Long Short-Term Memory (LSTM)-based architecture to harness spatial-temporal aspects of a continuous signal.…”
Life in modern societies is fast-paced and full of stress-inducing demands. The development of stress monitoring methods is a growing area of research due to the personal and economic advantages that timely detection provides. Studies have shown that speech-based features can be utilised to robustly predict several physiological markers of stress, including emotional state, continuous heart rate, and the stress hormone, cortisol. In this contribution, we extend previous works by the authors, utilising three German language corpora including more than 100 subjects undergoing a Trier Social Stress Test protocol. We present cross-corpus and transfer learning results which explore the efficacy of the speech signal to predict three physiological markers of stress—sequentially measured saliva-based cortisol, continuous heart rate as beats per minute (BPM), and continuous respiration. For this, we extract several features from audio as well as video and apply various machine learning architectures, including a temporal context-based Long Short-Term Memory Recurrent Neural Network (LSTM-RNN). For the task of predicting cortisol levels from speech, deep learning improves on results obtained by conventional support vector regression—yielding a Spearman correlation coefficient (ρ) of 0.770 and 0.698 for cortisol measurements taken 10 and 20 min after the stress period for the two corpora applicable—showing that audio features alone are sufficient for predicting cortisol, with audiovisual fusion to an extent improving such results. We also obtain a Root Mean Square Error (RMSE) of 38 and 22 BPM for continuous heart rate prediction on the two corpora where this information is available, and a normalised RMSE (NRMSE) of 0.120 for respiration prediction (−10: 10). Both of these continuous physiological signals show to be highly effective markers of stress (based on cortisol grouping analysis), both when available as ground truth and when predicted using speech. This contribution opens up new avenues for future exploration of these signals as proxies for stress in naturalistic settings.
“…There are several machine learning approaches that have explored physiologically derived markers in general for stress recognition (MacLaughlin et al, 2011;Dhama et al, 2019;Šalkevicius et al, 2019). From feature-based machine learning paradigms which classify various features extracted from wearable sensors, i. e., sleep quality, and percentage of screen time (Sano and Picard, 2013), or heart rate variability (HRV) (Dalmeida and Masala, 2021), and thermal-video recognition of the Initial Systolic Time Interval (Kumar S. et al, 2021), applying the state-of-the-art StressNet. StressNet consists of a Long Short-Term Memory (LSTM)-based architecture to harness spatial-temporal aspects of a continuous signal.…”
Life in modern societies is fast-paced and full of stress-inducing demands. The development of stress monitoring methods is a growing area of research due to the personal and economic advantages that timely detection provides. Studies have shown that speech-based features can be utilised to robustly predict several physiological markers of stress, including emotional state, continuous heart rate, and the stress hormone, cortisol. In this contribution, we extend previous works by the authors, utilising three German language corpora including more than 100 subjects undergoing a Trier Social Stress Test protocol. We present cross-corpus and transfer learning results which explore the efficacy of the speech signal to predict three physiological markers of stress—sequentially measured saliva-based cortisol, continuous heart rate as beats per minute (BPM), and continuous respiration. For this, we extract several features from audio as well as video and apply various machine learning architectures, including a temporal context-based Long Short-Term Memory Recurrent Neural Network (LSTM-RNN). For the task of predicting cortisol levels from speech, deep learning improves on results obtained by conventional support vector regression—yielding a Spearman correlation coefficient (ρ) of 0.770 and 0.698 for cortisol measurements taken 10 and 20 min after the stress period for the two corpora applicable—showing that audio features alone are sufficient for predicting cortisol, with audiovisual fusion to an extent improving such results. We also obtain a Root Mean Square Error (RMSE) of 38 and 22 BPM for continuous heart rate prediction on the two corpora where this information is available, and a normalised RMSE (NRMSE) of 0.120 for respiration prediction (−10: 10). Both of these continuous physiological signals show to be highly effective markers of stress (based on cortisol grouping analysis), both when available as ground truth and when predicted using speech. This contribution opens up new avenues for future exploration of these signals as proxies for stress in naturalistic settings.
“…Feature selection showed that time domain HRV metrics such as RMSSD and AVNN were important features in stress classification. 39 In our study, tree-based extra-tree classifier had highest sensitivity of 84% and an AUC Score of 84% and an accuracy of 77 %. Feature ranking in our study showed that both demographic features such as mental well-being, higher marriage age, joint family type and HRV features including pNNi20, SDNN and RMSSD were the top classifiers that distinguished between subjects reporting burnout/stress and heathy ones.…”
Section: Discussionmentioning
confidence: 47%
“… 21 These ECG-derived HRV features as markers for stress detection have been previously used in ML algorithms such as K-Nearest Neighbour (KNN), Support Vector Machines (SVM), Multilayer Perceptron (MLP), Random Forest (RF) and Gradient Boosting (GB). 39 In a study aimed to detect stress based on HRV features derived from Apple watch, MLP was the best ML model with 75% AUC, 80% Recall and 72% F1 score. Feature selection showed that time domain HRV metrics such as RMSSD and AVNN were important features in stress classification.…”
Objectives
COVID-19 pandemic has led to unprecedented increase in rates of stress and burn out among healthcare workers (HCWs). Heart rate variability (HRV) has been shown to be reflective of stress and burnout. The present study evaluated the prevalence of burnout and attempted to develop a HRV based predictive machine learning (ML) model to detect burnout among HCWs during COVID-19 pandemic.
Methods
Mini-Z 1.0 survey was collected from 1615 HCWs, of whom 664, 512 and 439 were frontline, second-line and non-COVID HCWs respectively. Burnout was defined as score ≥ 3 on Mini-Z-burnout-item. A 12-lead digitized ECG recording was performed and ECG features of HRV were obtained using feature extraction. A ML model comprising demographic and HRV features was developed to detect burnout.
Results
Burnout rates were higher among second-line workers 20.5% than frontline 14.9% and non-COVID 13.2% workers. In multivariable analyses, features associated with higher likelihood of burnout were feeling stressed (OR = 6.02), feeling dissatisfied with current job (OR = 5.15), working in a chaotic, hectic environment (OR = 2.09) and feeling that COVID has significantly impacted the mental wellbeing (OR = 6.02). HCWs with burnout had a significantly lower HRV parameters like root mean square of successive RR intervals differences (RMSSD) [P<0.0001] and standard deviation of the time interval between successive RR intervals (SDNN) [P<0.001]) as compared to normal subjects. Extra tree classifier was the best performing ML model (sensitivity: 84%)
Conclusion
In this study of HCWs from India, burnout prevalence was lower than reports from developed nations, and was higher among second-line versus frontline workers. Incorporation of HRV based ML model predicted burnout among HCWs with a good accuracy.
“…The authors of [5] presented a study on the physiological parameters (with particular focus on the heart rate variability (HRV)) that can be extracted from wearable devices to detect stress levels in car drivers. The authors developed a predictive model based on different machine learning (ML) methodologies such as K-Nearest Neighbor (KNN), Random Forest (RF), among others that is able to classify the stress level extracted from ECG-derived HRV features [5]. The techniques proposed by the authors show that the HRV features can act as markers for stress level detection, achieving a recall of 80% with the ML models proposed [5].…”
Section: Contributed Papersmentioning
confidence: 99%
“…The authors developed a predictive model based on different machine learning (ML) methodologies such as K-Nearest Neighbor (KNN), Random Forest (RF), among others that is able to classify the stress level extracted from ECG-derived HRV features [5]. The techniques proposed by the authors show that the HRV features can act as markers for stress level detection, achieving a recall of 80% with the ML models proposed [5].…”
The demographic shift of the population toward an increased number of elder citizens, together with the sedentary lifestyle we are adopting, is reflected in the increasingly debilitated physical health of the population [...]
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.