Autism severity level detection using fuzzy expert system

Autism is a neuro developmental disorder that is recently well known among Malaysian. Many researches on autism detection have been conducted worldwide. However, there is lack of research conducted in detecting autism severity level. Therefore, this paper focuses on autism severity level detection using fuzzy expert system. Two main autistic behavioral criteria are selected which are social communication impairment and restricted repetitive behavior. Data acquisition was based on interview sessions with clinical psychologist and distribution of 36 questionnaires to teachers and parents that have autistic children. It was then analyzed and the cut off points for each severity level; level 1 (mild), level 2 (moderate), and level 3 (severe) is determined. The fuzzy expert system processes are employed to detect the severity levels. The processes involve Fuzzy system architecture, fuzzification, rules evaluation, rules evaluation and defuzzification. The finding demonstrates that the system is able to detect autism severity level with a good accuracy. This system also accommodates with suitable recommendation based on the generated result whether the suggestion is to go for speech therapy or behavior therapy.