Prediction and Analysis of Autism Spectrum Disorder Using Machine Learning Techniques.

Area of Science:

• Neurodevelopmental Disorders
• Artificial Intelligence in Healthcare
• Machine Learning Applications

Background:

• Autism spectrum disorder (ASD) is a lifelong neurodevelopmental condition impacting socio-communication and behavior.
• Early recognition of ASD symptoms in children aged 2-3 years is crucial.
• Existing ASD prediction research heavily relies on traditional machine learning algorithms.

Purpose of the Study:

• To investigate and compare various machine learning algorithms for autism spectrum disorder prediction.
• To provide a centralized framework for researchers in the field of ASD prediction.
• To evaluate prediction models based on common parameters like application type, simulation method, comparison methodology, and input data.

Main Methods:

• Comparison of traditional machine learning algorithms including Support Vector Machine, Random Forest, Multiple Layer Perceptron, Naive Bayes, Convolution Neural Network, and Deep Neural Network.
• Validation of proposed models using performance metrics such as accuracy, precision, and recall.
• Analysis of autism spectrum disorder prediction across different parameters.

Main Results:

• The Random Forest algorithm demonstrated superior performance compared to other traditional machine learning algorithms.
• An accuracy of 89.23% was achieved using the Random Forest model for ASD prediction.
• Workflow representations were provided to elucidate the architectures of the investigated frameworks.

Conclusions:

• The Random Forest algorithm is highly effective for autism spectrum disorder prediction.
• The study offers a valuable centralized framework to guide future research in ASD prediction.
• The findings highlight the potential of machine learning in improving the accuracy and efficiency of ASD diagnosis.