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

Diagnostics, Journal Year: 2024, Volume and Issue: 14(22), P. 2504 - 2504

Published: Nov. 8, 2024

Background: As the demand for early and accurate diagnosis of autism spectrum disorder (ASD) increases, integration machine learning (ML) explainable artificial intelligence (XAI) is emerging as a critical advancement that promises to revolutionize intervention strategies by improving both accuracy transparency. Methods: This paper presents method combines XAI techniques with rigorous data-preprocessing pipeline improve interpretability ML-based diagnostic tools. Our preprocessing included outlier removal, missing data handling, selecting pertinent features based on clinical expert advice. Using R caret package (version 6.0.94), we developed compared several ML algorithms, validated using 10-fold cross-validation optimized grid search hyperparameter tuning. were employed model transparency, offering insights into how contribute predictions, thereby enhancing clinician trust. Results: Rigorous improved models’ generalizability real-world applicability across diverse datasets, ensuring robust performance. Neural networks extreme gradient boosting models achieved best performance in terms accuracy, precision, recall. demonstrated behavioral significantly influenced leading greater interpretability. Conclusions: study successfully highly precise interpretable ASD diagnosis, connecting advanced methods practical application supporting adoption AI-driven tools healthcare professionals. study’s findings personalized practices, ultimately outcomes quality life individuals ASD.

Language: Английский

Improving the Generalization Abilities of Constructed Neural Networks with the Addition of Local Optimization Techniques

Algorithms, Journal Year: 2024, Volume and Issue: 17(10), P. 446 - 446

Published: Oct. 6, 2024

Constructed neural networks with the assistance of grammatical evolution have been widely used in a series classification and data-fitting problems recently. Application areas this innovative machine learning technique include solving differential equations, autism screening, measuring motor function Parkinson’s disease. Although has given excellent results, many cases, it is trapped local minimum cannot perform satisfactorily problems. For purpose, considered necessary to find techniques avoid minima, one periodic application minimization that will adjust parameters constructed artificial network while maintaining already existing architecture created by evolution. The shown significant reduction both found relevant literature.

Language: Английский