Fuzzy Logic for Autism Screening Test

Abstract

Autistic Spectrum Disorder (ASD) can faze brain development, growth, and social behaviour. Teens with ASD tend to become bullying object. Autism can be recognized and diagnosis using a test with a questionnaire. The answers from the questionnaire usually converted into an integer value using the Linkers scale, whereas this answer still contains biased and this bias cannot be captured using this scale. Through our initial test results shows that if a user uses an answer choice slightly agree or slightly disagree, the results value is low enough it's about 6% to 20%. Autistic according to the CDC, it has eight symptoms, and we apply it to form fuzzy memberships. Previous Researchers had been shared autism screening dataset to the UCI machine learning repository, this dataset we filtered for "Who is completing the test" attribute with "self"value. We used the classification algorithms in WEKA to find a model, then this model we applied in fuzzification. This system is still a simulation study and has not been clinically tested. This system is assigned to diagnose autism based on the recognized symptoms, where the number of asked questions can vary according to needs but still refers to the symptoms. This paper proposed an autism screening test diagnosis using the fuzzy logic with a better interpellation and precision.