An Entropy based approach to investigate the heterogeneous sensory pattern in Autism Spectrum Disorders

Sensory hyper-or hypo-responsivity, which imparts heterogeneity to Autism Spectrum Disorders (ASD), adversely affects the communication skills, classroom learning and independent living of such individuals. The need of the hour is to provide targeted medical intervention to the neurobiologically distinct subgroups present in ASD. Hence, the proposed work is aimed at examining the heterogeneity and identifying the hyper-responsive subgroup in ASD. The current study employs entropy-based measure calculated over multiple time scales to study the complexity pattern of the Visual Evoked Potential (VEP) signals recorded from a group of 30 subjects, diagnosed with ASD. The VEP signals are measured in response to visual stimuli consisting of vertical achromatic sinewave gratings presented under low (5%), medium (50%) and high (90%) contrasts. After pre-processing, the VEP signals are decomposed into 5 modes using Variational Mode Extraction as follows: mode 1 (0–4 Hz), mode 2 (4–8 Hz), mode 3 (8–16 Hz), mode 4 (16–32 Hz) and mode 5 (32–64 Hz). Improved Multiscale Permutation Entropy (IMPE) is extracted from these modes and Relative Complexity (RC) analysis is carried out across varying contrasts. The IMPE-based RC analysis exhibits an increasing complexity pattern which is an indication of hyper-responsivity and is further statistically verified through Friedman test. This pattern is observed in 43% of the ASD study population in modes 2 and 3, which correspond to theta and alpha bands of brain signal frequencies.