سامانه همایش های دانشگاه آزاد اسلامی واحد نجف آباد

صفحه اصلی / International Conference on Artificial Intelligence; City, Industry and Health

A Review of Machine Learning Methods for Autism Diagnosis

نویسندگان :

Ali Emami¹ Nasim Noorafza²

1- Department of Computer Engineering, Na.C, Islamic Azad University, Najafabad, Iran 2- Department of Computer Engineering, Na.C, Islamic Azad University, Najafabad, Iran

کلمات کلیدی :

Autism Spectrum Disorder،Ensemble Methods،Machine Learning،Machine Learning in Medicine

چکیده :

Autism Spectrum Disorder (ASD) is a neurological condition characterized by difficulties in social interactions, communication, and repetitive behaviors. Although its primary cause is genetic, early detection is critically important, and the application of machine learning offers a promising solution for faster and more cost-effective diagnosis. However, studies in the field of ASD diagnosis using machine learning face several challenges. One major limitation is the reliance on single models, which often struggle to simultaneously capture the complex features of autism due to limited generalizability. In addition, many classification models depend on labeled data, and their performance significantly degrades when the data is noisy or of small volume. Moreover, most machine learning methods—especially those based on single algorithms—are unable to integrate information from different aspects of the data. In contrast, combining multiple perspectives can enhance the accuracy and robustness of detection. Given these limitations, the use of ensemble methods, which combine multiple machine learning models to leverage the strengths of each, presents an effective solution. In this paper, we review classical and standalone machine learning methods, and finally, we present our proposed approach.