Heart Disease Clustering Modeling Using a Combination of the K-Means Clustering Algorithm and the Elbow Method

Purpose: Heart disease is the leading cause of death throughout the world, especially in developing countries like Indonesia. Modern approaches for diagnosing and managing heart disease rely on machine learning due to the complexity of medical data. Among the biggest challenges in using machine learning is clustering heart disease patients. This study aims to develop a machine-learning model using K-means clustering to determine the severity and level of patient emergencies. The specific objective of the model is to find the optimal number of clusters using the Elbow Method. Methods/Study design/approach: Model development using a dataset from the Kaggle Repository consisting of 303 patient data. Each data point consists of the attributes age, gender, type of chest pain, blood pressure, serum cholesterol level, blood sugar, electrocardiography results, maximum heart rate, angina, ST depression, and segment slope ST. The combination of the K-means clustering algorithm and the Elbow Method is expected to find the optimal number of clusters in the developed model. Result/Findings: The results of building a machine learning model show that k-means clustering is quite effective in clustering heart disease patients. For the model built with 303 data points, the elbow method successfully identified the optimal number of clusters, resulting in two clusters (k=2), where the elbow point on the graph shows a significant decrease in the Sum of Squared Errors (SSE). Novelty/Originality/Value: This study combines the k-means clustering algorithm and the elbow method to determine the severity and level of patient emergencies. The clustering model produces specific risk clusters that help doctors determine more appropriate interventions.