Autism spectrum disorder detection from semi-structured and unstructured medical data.

Area of Science:

• Medical Informatics
• Artificial Intelligence
• Developmental Pediatrics

Background:

• Autism spectrum disorder (ASD) diagnosis is complex, time-consuming, and requires specialized expertise.
• Early intervention is crucial for improving outcomes in individuals with ASD.
• Current diagnostic methods present significant accessibility challenges.

Purpose of the Study:

• To develop a machine learning-based system for early autism spectrum disorder (ASD) detection.
• To leverage Natural Language Processing (NLP) techniques for analyzing medical forms.
• To enhance accessibility and efficiency in the ASD diagnostic process.

Main Methods:

• Digitization and preprocessing of semi-structured and unstructured medical forms.
• Application of NLP for document representation learning.
• Classification models trained on extracted medical data for ASD detection.

Main Results:

• The proposed system achieved an accuracy of 83.4% in detecting ASD.
• A recall rate of 91.1% was obtained, indicating high sensitivity in identifying potential cases.
• The framework demonstrated promising results in simplifying and shortening the diagnostic procedure.

Conclusions:

• The developed machine learning framework offers a promising approach for efficient and accessible autism spectrum disorder (ASD) detection.
• NLP techniques applied to medical forms can significantly aid in the early identification of ASD.
• This system has the potential to improve patient access to timely interventions.