Reliable Autism Spectrum Disorder Diagnosis for Pediatrics Using Machine Learning and Explainable AI

Area of Science:

• Artificial Intelligence
• Machine Learning
• Neuroscience

Background:

• Increasing demand for early and accurate autism spectrum disorder (ASD) diagnosis.
• Emerging role of machine learning (ML) and explainable artificial intelligence (XAI) in improving diagnostic accuracy and transparency.
• Potential of AI to revolutionize ASD intervention strategies.

Purpose of the Study:

• To present a method combining XAI with data preprocessing for accurate and interpretable ML-based ASD diagnostic tools.
• To enhance the transparency of ML models for clinical applications.
• To improve clinician trust in AI-driven diagnostic tools.

Main Methods:

• Rigorous data preprocessing: outlier removal, missing data handling, feature selection.
• Development and comparison of ML algorithms using R and caret package.
• Validation via 10-fold cross-validation and hyperparameter tuning with grid search.
• Application of XAI techniques for model interpretability.

Main Results:

• Data preprocessing enhanced model generalizability and applicability across diverse datasets.
• Neural networks and extreme gradient boosting models showed superior performance (accuracy, precision, recall).
• XAI revealed significant influence of behavioral features on predictions, increasing interpretability.
• Enhanced clinician trust through transparent AI insights.

Conclusions:

• Successfully developed precise and interpretable ML models for ASD diagnosis.
• Bridged advanced ML methods with clinical practice for AI adoption.
• Findings support personalized interventions and early diagnostic practices for improved ASD outcomes.
• Facilitated better quality of life for individuals with ASD through AI-driven tools.