Abstract

The presence of Autism Spectrum Disorder (ASD) in a person makes them lag in social interaction. Thus, early detection and diagnosis solution is needed for ASD patients. Hence, an SFP-ANFIS-based VR intervention for the early diagnosis of ASD is proposed. Initially, the EEG signal of an ASD patient is taken and pre-processed, from which features are extracted. Based on the extracted features, the SH-KPA groups the patients with autism and their corresponding emotion. Similarly, from the behavioral images, the features are extracted after landmarking the facial points. Then, the features are classified with SH-KPA. From the SH-KPA results of EEG signal and behavioral images, the similarity is estimated to recognize the ASD child emotion. Then, based on the previous health records of patients, the state of the patient, estimated emotions, and similarity score, the SFP-ANFIS predicts the level of ASD and recommends a possible solution to be conducted with a VR device. Finally, through experiments, the efficacy of the proposed model is proved.

Key words: Autism Spectrum Disorder (ASD); Virtual Reality (VR); Electroencephelogram (EEG); Semi-Hausdroff-based K-Prototype Algorithm (SH-KPA); ASD solution recommendation