Abstract
The affordability and miniaturization of sensors create a revolution in wearable wireless solutions deployed to collect physiological parameters to assist in diseases/disorders diagnosis. Electroencephalography (EEG), a recording of the brain's electrical activity, is a promising physiological measure for autism spectrum disorder detection. It can reveal the irregularity of the neural system that is associated with autism. Wireless sensors represent a suitable infrastructure that can be deployed for signal transmission to the processing center. However, streaming EEG signals remotely for classification could shorten the lifetime of the wireless sensor and might question the viability of the application. Therefore, reducing the data transmission might preserve the sensor's energy and increase the wireless sensor network lifetime. This paper proposes the design of a sensor-based scheme for early-age autism detection in children. The proposed scheme uses low-complexity algorithms for on-node EEG processing, relevant features extraction and classification. The proposed processing scheme of the EEG signal is based on Haar wavelet transform and dynamics features. The experimental results show 93% accuracy, 86% sensitivity, and 100% specificity.