Recognition of positive and negative valence states in children with autism spectrum disorder (ASD) using discrete wavelet transform (DWT) analysis of electrocardiogram signals (ECG)

Children with autism spectrum disorder (ASD) are defcit in communication, social skills, empathy, emotional responsivenessand have signifcant behavioral pattern. They have difculty in understanding other feelings and their own emotions. Thisleads to the sudden emotional outburst and aggressive behavior in these children. Parents, caretakers and doctors fnd it verydifcult to prevent such extreme behaviors. Learning the positive and negative valence leads in determining the early indica-tions before the onset of emotional outbursts in children with ASD. The present study measures the psycho physiologicalelectrocardiogram (ECG) signal from the typically developed (TD) childrenand children with ASD in the age group of 5–11years.Personalized protocol was developed for every child with ASD to induce positive and negative valence and ECG datawas collected using wearable Shimmer ECG device. The heart rate ariability (HRV) and the QRS amplitude were derivedfrom ECG signal using Pan–Tompkins algorithm and eleven features were extracted using DWT (db2, db4 and db8) motherwavelet. The signifcant features of ECG, HRV and QRS amplitude were classifed using the K nearest neighbor (KNN),support vector machine (SVM) and ensemble classifer. Ensemble and KNN classifer achieved maximum accuracy of 81%and 76.2% for children with ASD and Ensemble and SVM classifers obtained maximum accuracy of 87.4% and 83.8% forTD children using HRV data.