Abstract
In recent years, the massive growth of healthcare data, combined with advances in artificial intelligence, has opened up exciting new possibilities for improving disease detection and patient care. This thesis explores how machine learning can be used to create an intelligent system that helps detect potential health issues by analyzing patient data and symptoms. The goal is to assist both medical professionals and the general public in quickly and accurately identifying possible health concerns.To achieve this, we apply a variety of machine learning techniques, such as Decision Trees, K-Nearest Neighbors (KNN), Naïve Bayes, Logistic Regression, and Support Vector Machines (SVM). These algorithms analyze different types of data—structured data like test results and vital signs, as well as unstructured data such as doctors’ notes or symptom descriptions. By combining these approaches, the system becomes more accurate in diagnosing a wider range of real-world health problems. Additionally, the proposed system offers a unique feature: a Hospital Queuing Recommendation (HQR) model. This model helps predict patient wait times during various stages of treatment, which can ease congestion and improve the flow of patients through the hospital. Using real hospital data, we test the system with a focus on chronic conditions like cerebral infarction, aiming to demonstrate its effectiveness. With an impressive prediction accuracy of 94.8%, the system shows promising results. This tool is designed not only as a decision-support system for healthcare providers but also as a self-assessment resource for patients—especially those living in areas with limited medical access. Our ultimate goal is to provide smarter, faster, and more accessible healthcare by combining machine learning with real-world clinical data, leading to better early detection and more effective treatment planning for all.