Automatic migraine classification using artificial neural networks.

Area of Science:

• Neurology
• Artificial Intelligence
• Medical Diagnostics

Background:

• Traditional migraine classification relies on complex brain wave analysis, limiting accessibility.
• Early migraine diagnosis is often challenging due to time constraints and physician inexperience, leading to misclassification and suboptimal treatment.
• Accurate and timely diagnosis is crucial to prevent inappropriate therapies and protect patient health.

Purpose of the Study:

• To design and validate an early migraine classification system utilizing patient-reported symptoms.
• To differentiate between seven distinct types of migraines for improved diagnostic accuracy.
• To leverage artificial intelligence for efficient and reliable migraine diagnosis.

Main Methods:

• Data collection involved patient symptoms and physician diagnoses.
• Feature selection identified the most pertinent diagnostic variables.
• Artificial neural network (ANN) models were employed for automated classification.
• Model performance was evaluated based on accuracy and precision metrics.

Main Results:

• ANN models demonstrated superior classification performance compared to traditional methods.
• Accuracy and precision levels exceeded 97% for the ANN models.
• The proposed system outperformed logistic regression, support vector machines, nearest neighbor, and decision trees.

Conclusions:

• Artificial neural networks offer a powerful tool for migraine classification.
• The system significantly reduces the time required for accurate and reliable clinical diagnoses.
• Implementation of this AI-driven approach enhances the timeliness and precision of migraine patient care.