Using Machine Learning to Diagnose Autism Based on Eye Tracking Technology

Area of Science:

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

Background:

• Early diagnosis of autism spectrum disorder (ASD) is critical for effective intervention.
• Current diagnostic methods often rely on clinical observation after age three, delaying crucial early support.
• There is a need for more effective and earlier diagnostic approaches for ASD.

Purpose of the Study:

• To develop an earlier and more effective method for diagnosing autism spectrum disorder (ASD).
• To evaluate the efficacy of artificial intelligence models using eye-tracking datasets for ASD diagnosis.
• To identify autism scores using eye-tracking data.

Main Methods:

• Utilized the Eye Gaze fixes map dataset and the Eye Tracking Scanpath dataset (ETSDS) for ASD diagnosis.
• Employed a hybrid artificial intelligence model for analyzing eye-tracking data.
• A subset of the ETSDS was specifically used for recognizing autism scores.

Main Results:

• The hybrid model achieved high accuracy rates of 96.1% on Eye Gaze fixes map datasets and 98.0% on the ETSDS.
• An accuracy rate of 98.1% was achieved on the ETSDS for recognizing autism scores.
• The proposed method demonstrated superior performance compared to previous studies.

Conclusions:

• Artificial intelligence techniques are effective for diagnosing diseases, including autism spectrum disorder (ASD).
• The study highlights the potential of eye-tracking data analysis for early ASD detection.
• Further research into advanced AI techniques for disease diagnosis is warranted.