Machine learning-based classification of autism spectrum disorder from behavioural questionnaires
ASD is a frequently occurring neurodevelopmental disorder that is disabling in terms of social organisation, communication, and behaviour. ASD is important to detect early in order to implement effective interventions and achieve better developmental outcomes. Conventional diagnostic methods and esp...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Conference or Workshop Item |
| Language: | en |
| Published: |
IEEE
2026
|
| Subjects: | |
| Online Access: | https://umpir.ump.edu.my/id/eprint/47000/1/Machine_Learning-Based_Classification_of_Autism_Spectrum_Disorder_from_Behavioural_Questionnaires-2.pdf https://umpir.ump.edu.my/id/eprint/47000/ https://doi.org/10.1109/ICSECS65227.2025.11279195 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | ASD is a frequently occurring neurodevelopmental disorder that is disabling in terms of social organisation, communication, and behaviour. ASD is important to detect early in order to implement effective interventions and achieve better developmental outcomes. Conventional diagnostic methods and especially those that are based on manual observation, are labourintensive, resource- and time-consuming. The paper presents a framework of machine learning classification, where the data in the form of behavioural questionnaires is used to find a solution to the class imbalance problem with the help of SMOTE and an increase in the model performance using Principal Component Analysis (PCA). The models in use will be Random Forest (RF), Decision Tree (DT) and XGBoost. The highest accuracy of 95.50% was recorded by XGBoost as compared to RF, which scored 93.90%. Using PCA in the integration of the model made a great deal of changes in efficiency, in that there was less redundancy in features. The results indicate that this method can be used to successfully, time-efficiently, and cost-effectively detect ASD, and thus it can be used in practice in the case of telemedicine as well. |
|---|