A machine learning autism classification based on logistic regression analysis.

Area of Science:

• Neurodevelopmental Disorders
• Machine Learning Applications
• Biomedical Data Analysis

Background:

• Autistic Spectrum Disorder (ASD) presents significant healthcare costs, with early diagnosis offering potential cost reductions.
• Effective and accessible screening methods are urgently needed due to the economic impact of autism.
• Limited availability of non-genetic autism screening datasets hinders timely diagnosis.

Purpose of the Study:

• To propose a novel machine learning framework for autism screening in adults and adolescents.
• To identify influential features for autism screening through in-depth data analysis.
• To develop a time-efficient screening tool to aid health professionals and individuals.

Main Methods:

• Development of a machine learning framework for autism screening.
• Utilizing logistic regression for predictive analysis on autism datasets.
• Employing Information Gain (IG) and Chi-square (CHI) testing for feature analysis.

Main Results:

• The machine learning framework demonstrated acceptable classification performance.
• Key features influential in autism screening were identified.
• The proposed methods provide valuable insights for autism screening.

Conclusions:

• Machine learning offers a viable approach for developing effective autism screening tools.
• Feature analysis is critical for optimizing screening tool performance.
• This framework can enhance early identification and diagnosis of ASD.