Multimodal data integration for early autism detection and LLM-driven personalized intervention: A review.

Area of Science:

• Neurodevelopmental Disorders
• Artificial Intelligence in Healthcare
• Biomedical Data Integration

Background:

• Autism Spectrum Disorder (ASD) presents significant early detection and intervention challenges.
• Current AI applications struggle with multimodal data integration for early ASD detection and intervention planning.
• Advancements in AI are vital for improving diagnostic accuracy and treatment personalization in ASD.

Purpose of the Study:

• To examine the role of AI in integrating diverse data modalities for a comprehensive understanding of autism.
• To discuss the application of generative AI, specifically LLMs, in creating personalized early intervention plans.
• To provide a holistic overview of AI advancements and future directions in early ASD detection and intervention.

Main Methods:

• Reviewing the emerging role of AI in processing multimodal data (image, audio, video) for autism analysis.
• Exploring the use of generative models like LLMs for personalized treatment plan generation.
• Synthesizing current research on AI-driven early autism detection and intervention strategies.

Main Results:

• AI shows promise in integrating multimodal data for a deeper understanding of autistic behaviors.
• LLMs can facilitate the creation of individualized medical reports for early intervention planning.
• Effective multimodal data integration remains a challenge for current AI in early ASD detection.

Conclusions:

• AI integration of multimodal data is essential for comprehensive autism understanding and early detection.
• Generative AI, particularly LLMs, offers a novel approach to personalized intervention planning for ASD.
• Further research is needed to overcome current limitations in AI-driven early autism detection and intervention.