RECOMMENDATION SYSTEM USING NEURAL NETWORK ALGORITHM FOR EARLY SYMPTOM TREATMENT OF ISCHEMIC STROKE

Abstract

Ischemic stroke is a critical medical condition requiring rapid intervention to prevent permanent neurological damage and reduce global mortality rates. This study develops a recommendation system for early symptom treatment using the Artificial Neural Network (ANN) algorithm to accelerate clinical decision-making. Utilizing a dataset of 500 patients, the model analyzes primary symptoms such as face drooping, arm weakness, and speech difficulty through a Multi-Layer Perceptron architecture. The data is processed using Min-Max Scaling normalization and Adam optimization to ensure high accuracy and rapid model convergence during the training phase. Research results demonstrate outstanding performance with an accuracy rate of 94.5% and a sensitivity (recall) reaching 95.8% in accurately detecting stroke cases. The system provides an average prediction value of 0.89 in critical conditions, triggering automatic recommendations for immediate thrombolysis treatment within the golden period.