Implementation of Deep Learning in Diagnosing Stroke Disease Based on Clinical Data Parameters

Abstract

Stroke is a medical condition that requires rapid and accurate diagnosis to increase the chances of patient recovery and reduce the risk of long-term complications. This research investigates the application of deep learning techniques to diagnose stroke based on clinical data parameters. It develops and applies deep learning models that use various neural network architectures, such as convolutional neural networks (CNN). The dataset used is an open dataset in analyzing the dataset that includes patient clinical information such as hypertension, cardiac history, married status, employment level, body mass index (BMI), smoking and glucose. The model was trained using a dataset consisting of thousands of medical records of patients with stroke and without stroke. Model evaluation was conducted using performance metrics such as accuracy, precision, recall, and F1-score to assess the effectiveness in classification with an accuracy value of 95.05%. The results showed that the deep learning approach significantly improved the accuracy and speed in detecting stroke compared to conventional diagnosis methods. These findings suggest that the integration of deep learning in clinical diagnostic systems can improve early stroke detection and provide a solid basis for better clinical decisions.