Stress response index for adverse childhood experience based on fusion of hypothalamus pituitary adrenocorticol and autonomic nervous system biomakers

Abstract

Early life exposure to stress such as adverse childhood experiences has been suggested to cause changes in physiological processes and alteration in stress response magnitude which might have significant impact on health later in life. For this reason, detection of this altered stress response can be used as an indicator for future health. To date, there is no study that utilized this information to indicate future health. In order to detect the altered stress response, biomarkers that represent both Autonomic Nervous System (ANS) and Hypothalamic-Pituitary-Adrenocorticol (HPA) is proposed. Among the available biomarkers, Heart Rate Variability (HRV) has been proven as a powerful biomarker that represents ANS. Meanwhile, salivary cortisol has been suggested as a biomarker that reflects the HPA. Even though many studies used multiple biomarkers to measure the stress response, the results for each biomarker were analysed separately. Therefore, this study fuses the biomarker that represents both ANS and HPA as a single measure, proposes a new method to classify the stress response based on adverse childhood experience in the form of stress response index as a future health indicator. Electrocardiograph, blood pressure, pulse rate and Salivary Cortisol (SCort) were collected from 23 participants, 12 participants who had adverse childhood experience while the remaining 11 act as the control group. The recording session was done during a Paced Auditory Serial Addition Test (PASAT). HRV features were then extracted from the electrocardiograph (ECG) using time, frequency, time-frequency analysis, and wavelet transform. Following this, genetic algorithm was implemented to select a subset of 12 HRV features from 83 features. Next, the selected HRV features were combined with other biomarkers using parallel and serial fusion for performance comparison. Using Support Vector Machine (SVM), results showed that fused feature of the parallel fusion, so-called Euclidean distance (ed), demonstrated the highest performance with 80.0% accuracy, 83.3% sensitivity and 78.3% specificity. Finally, the fused feature of the Euclidean distance was fed into SVM in order to model the stress response index as an indicator for future health. This index was validated using all samples and achieved 91.3% accuracy. From this study, a new method based on HRV-SCort biomarker using Euclidean distance and SVM named as ed-SVM was proven to be an effective method to classify the stress response and could further be used to model a stress response index. This index can then be benefited as an indicator for future health to improve the health care management in adulthood.