Development of Psychological Stress Detection System Using Bio-Signals

Abstract

Psychological stress is an inevitable part of the modern-day lifestyle that affects human cognitive abilities. The relation between stress and a host of behavioural and somatic pathological conditions is well-established. The low doctor-to-patient ratio in under-developed and developing economies hinders access to expert diagnosis. This emphasizes the need for computer-aided timely detection of psychological stress. The methods like Electroencephalography (EEG) and Electrocardiography (ECG) provide important biophysical diagnostic measures for psychological stress detection, however, these methods are expensive or require a proper clinical setup. Whereas, the acoustic heart sound or Phonocardiography (PCG) signals carry significant information and can be easily acquired. The purpose of this research work is to present a novel framework for psychological stress detection using PCG signal that can serve as a first-level screening method at places where EEG or ECG are not available. In this research work, the pre-competitive (or exam-related) psychological stress is used a as real-life stressor. For this study, the simultaneous ECG and PCG data of five minutes duration is acquired from 33 healthy male students in the age group of 18 to 25 years (mean = 20.11, standard deviation = 2.30) of Thapar Institute of Engineering and Technology, TIET campus who are attempting professional education institute examination. Two readings are acquired from every subject, one approximately two hours before the start of the exam and considered as the signals of subjects under psychological stress, whereas, second reading forms the baseline values for subject-specific template formation and recorded once the students returned from holidays after exams. In this study, the ECG signal is used as a reference signal for S1 peak detection of PCG signal and later for comparison of the results with that obtained from PCG-based method. The State-Trait Anxiety Inventory (STAI Form Y) self-report questionnaire is used in the study as the scores on the S-Anxiety scale increase when used under psychological stress and decrease after relaxation. The psychological stress is detected from the S1-S1 interval of PCG signal, referred to as an inter-beat interval (IBI) signal. The empirical mode decomposition (EMD) technique is used for decomposing IBI signal to intrinsic mode functions (IMFs). The EMD technique has been found suitable for non-linear and non-stationary signal analysis. The non-linear features namely- Area of Analytic Signal Representation (AASR), Log of Area of ellipse from Second-order Difference Plot (LASODP), Root Mean Square value of IMF (RmsIMF), Shannon Entropy (ShEnt) and vii Fuzzy Entropy (FzEnt) were evaluated from IMFs of IBI signals. The first stage of this study comprises of deviation analysis in stressed signals from mean baseline values of the features in non-stressed signals. Thereafter, in the second stage of the study, Kruskal-Wallis statistical test has been used to check the significance and discrimination ability of the features. Subsequently, the features which showed maximum deviation and are statistically significant have been selected and fed to least-square support vector machine (LS-SVM) classifier. The 10-fold cross-validation has been used to make the system more reliable and robust. In this work, the average accuracy of 93.14% in classifying stressed and non-stressed signals has been achieved using Radial Basis Function (RBF) kernel. The novelty of this study is the use of PCG signals for psychological stress detection and the use of subject-specific baseline template to incorporate the individual cardiovascular characteristic behaviour and stress responses. The applicability of another set of non-linear entropy-based features namely- Permutation Entropy (PEn), Fuzzy Entropy (FzEn) and K-Nearest Neighbour (K-NN) entropy estimator is explored in EMD domain. In order to optimise the system, the ranking methods including Entropy method, Bhattacharya space algorithm, Receiver Operating Characteristic (ROC) method and Wilcoxon method are used. The highest-ranked features are fed to LS-SVM for classification. This method showed significant improvement in accuracy and the highest accuracy, sensitivity and specificity obtained using the proposed system is 96.67%, 100% and 93.33% respectively. The results indicate that the proposed features provide better discrimination ability than well-documented low-frequency to high-frequency power ratio (LF/HF) parameter of ECG signal on the dataset. The proposed novel methodology of using PCG signals for psychological stress detection is cost-effective and is suitable for home-care, telemedicine and in rural health care centres especially in developing countries. The proposed system opens a new research area of using PCG signal for psychological stress detection.