Enhancing Developmental Language Disorder Identification with Artificial Intelligence: Development of an Explainable...

Area of Science:

• Linguistics
• Developmental Psychology
• Artificial Intelligence

Background:

• Developmental Language Disorder (DLD) affects language acquisition in children.
• Accurate and timely diagnosis of DLD is crucial for intervention.
• Existing diagnostic methods can be time-consuming and subjective.

Purpose of the Study:

• To identify linguistic markers differentiating children with DLD from typically developing (TD) peers.
• To develop an explainable Artificial Intelligence (AI)-based screening application for DLD.
• To assess the reliability and feasibility of the AI screening app.

Main Methods:

• A cohort of 30 children (15 DLD, 15 TD, aged 7-10) underwent verbal assessments.
• Linguistic measures included vocabulary, morphosyntax, and sentence repetition.
• A random forest classifier was trained on synthetic data to create the AI screening app.

Main Results:

• Significant linguistic differences were found between DLD and TD groups across all measures.
• The AI screening app demonstrated high concordance with expert clinical diagnoses.
• The app proved reliable in identifying children with DLD.

Conclusions:

• Specific linguistic indicators hold diagnostic value for DLD.
• An AI-driven screening solution is feasible and offers practical advantages.
• The app's interpretability and scalability can improve DLD detection, especially in under-resourced areas.