A new computational intelligence approach to detect autistic features for autism screening

Area of Science:

• Neurodevelopmental Disorders
• Computational Intelligence
• Medical Diagnostics

Background:

• Autism Spectrum Disorder (ASD) is a rapidly growing developmental disability diagnosis.
• General Practitioners (GPs) are often the first point of contact for ASD concerns.
• Many families and adults remain unaware of ASD traits, delaying necessary diagnostic services.

Purpose of the Study:

• To identify fewer, influential features in common ASD screening methods for efficient screening.
• To address the urgent need for evaluating item influences on ASD within existing tools.
• To enhance early detection and treatment of ASD through accessible screening tools.

Main Methods:

• A computational intelligence method, Variable Analysis (Va), was proposed.
• Va considers feature-to-class correlations and reduces feature-to-feature correlations.
• Results were verified using machine learning algorithms for automated classification systems.

Main Results:

• Va derived fewer features from adult, adolescent, and child ASD screening methods.
• The method maintained competitive predictive accuracy, sensitivity, and specificity rates.
• Va's performance was comparable to other common filtering methods.

Conclusions:

• Variable Analysis (Va) offers an efficient approach to ASD screening.
• The method can improve the identification of key ASD traits.
• This contributes to earlier diagnosis and intervention for individuals with ASD.