Hilbert Huang Transform (HHT) Analysis of Heart Rate Variability (HRV) in Recognition of Emotion in Children with Autism Spectrum Disorder (ASD)

Emotion recognition has become a boon to society since it aids in the early management of uncontrollable sudden emotional outbursts in children with ASD prior to their onset. As a result, developing an alarm system assists caregivers, doctors and therapists in preventing such aggressive behaviours before they occur. Only a few studies have studied the physiological markers in ASD children and adults. The Pan-Tompkins equation was used to determine HRV from the RR index. The Hilbert-Huang transform (HHT) of heart rate variability (HRV) data was used to extract emotional characteristics corresponding to two emotions (positive and negative valence). The k-nearest neighbour (KNN) and ensemble classifiers were used to identify only the statistically important features. The HHT analysis of the HRV data achieved good classification rate with a mean accuracy of 74.5% and 73.5% in children with ASD and in TD children using the ensemble classifier. Further feature reduction was done using the principal component analysis (PCA), and it was seen that the accuracy improved to 77% and 78.3% in children with ASD and TD children, respectively using the ensemble classifier. Though the children with ASD are deficit in emotion recognition, we found no significant impairment in identifying the basic emotions. However, they were deficit in correctly predicting the positive valance state than the TD childrenKeywordsAutism spectrum disorder (ASD)Heart rate variability (HRV)Hilbert transformPrincipal component analysis (PCA)