CLASSIFICATION OF MIGRAINE TYPES BASED ON SYMPTOMS USING ARTIFICIAL NEURAL NETWORKS

Abstract

Migraine is a complex neurological disorder with heterogeneous clinical manifestations, making accurate subtype classification difficult using conventional diagnostic approaches. Diagnostic inaccuracies may result in inappropriate treatment and suboptimal patient outcomes. This study proposes an Artificial Neural Network (ANN) model to classify migraine subtypes based on patient-reported symptoms and clinical characteristics. A publicly available dataset containing 400 instances and 24 features—including demographic data, aura symptoms, neurological and autonomic indicators, genetic history, and disease burden—was utilized. Data preprocessing involved feature standardization, label encoding, and one-hot encoding, followed by an 80:20 split for training and testing. The ANN architecture comprised an input layer with 23 neurons, two hidden layers with 64 and 32 neurons using ReLU activation, and a seven-neuron output layer with softmax activation. The model was trained using the Adam optimizer and categorical cross-entropy loss for 50 epochs. Experimental results showed that the proposed model achieved a training accuracy of 91.56% and a testing accuracy of 93.00%, demonstrating strong generalization performance and effective learning of complex, non-linear symptom patterns. These results indicate that ANN-based classification has significant potential as a clinical decision-support tool for improving migraine subtype diagnosis and enabling more personalized treatment strategies.