HCBiLSTM-WOA: hybrid convolutional bidirectional long short-term memory with water optimization algorithm for autism...

Area of Science:

• Neurology and Artificial Intelligence
• Developmental Neuroscience
• Machine Learning in Healthcare

Background:

• Autism Spectrum Disorder (ASD) is a complex neurodevelopmental disability requiring early intervention for optimal outcomes.
• Existing ASD detection methods are often cumbersome and insufficient for real-time data analysis.
• Accurate and timely diagnosis is crucial for managing ASD's impact on individuals' physical and mental health.

Purpose of the Study:

• To propose a novel and efficient mechanism for Autism Spectrum Disorder (ASD) detection using real-time datasets.
• To enhance the accuracy and efficiency of ASD classification through a hybrid deep learning and optimization algorithm approach.
• To develop a tool that aids in the early and precise detection of ASD across various age groups.

Main Methods:

• A Hybrid Convolutional Bidirectional Long Short-Term Memory (HCBiLSTM) model was developed for ASD classification.
• The Water Optimization Algorithm (WOA) was employed to tune the hyperparameters of the HCBiLSTM model, optimizing prediction accuracy.
• Real-time ASD datasets encompassing diverse age groups were preprocessed to handle missing values, outliers, and data reduction before classification.

Main Results:

• The proposed HCBiLSTM-WOA method achieved a high ASD prediction accuracy of approximately 98.53%.
• The method demonstrated superior performance in detecting ASD compared to existing techniques across various performance metrics, including accuracy, sensitivity, and specificity.
• The model successfully classified ASD cases into respective stages based on observed autistic traits.

Conclusions:

• The fusion of HCBiLSTM and WOA offers a robust framework for accurate and efficient ASD diagnosis.
• This approach enhances diagnostic precision and improves the interpretation of complex ASD data, supporting early intervention strategies.
• The HCBiLSTM-WOA method provides clinicians with a valuable tool for early and precise ASD detection, ultimately improving patient outcomes.