آمار
| تعداد نشریات | 45 |
| تعداد شمارهها | 1,532 |
| تعداد مقالات | 18,614 |
| تعداد مشاهده مقاله | 61,077,429 |
| تعداد دریافت فایل اصل مقاله | 21,899,067 |
Unsupervised Discovery of Patient Health Patterns Using K-Means Clustering in Healthcare Data | ||
| Computational Methods for Differential Equations | ||
| مقالات آماده انتشار، پذیرفته شده، انتشار آنلاین از تاریخ 13 مرداد 1405 اصل مقاله (9.12 M) | ||
| نوع مقاله: Research Paper | ||
| شناسه دیجیتال (DOI): 10.22034/cmde.2026.73378.3742 | ||
| نویسندگان | ||
| Ali Subhi Alhumaima1؛ Taqwa Salim Mohammed2؛ Bashar Talib Al-Nuaimi3؛ Hussein Alkattan4؛ Mostafa O Abotaleb* 4 | ||
| 1Electronic Computer Centre, University of Diyala, Diyala, Iraq. | ||
| 2College of Medicine, University of Diyala, Diyala, Iraq. | ||
| 3Computer Science Department, University of Diyala, Diyala, Iraq. | ||
| 4Engineering School of Digital Technologies, Yugra State University, Khanty-Mansiysk, Russia. | ||
| چکیده | ||
| The growing availability of healthcare data opens up new opportunities to discover the underlying features of the data that may support patient classification and exploratory analysis for clinical decision-making. In this work, the healthcare data were clustered using a K-Means algorithm to infer valuable patient sub-types based on demographic, physiological and clinical characteristics. The dataset is comprehensive as it involves age, sex, blood pressure, heart rate, cholesterol level, body mass index (BMI), diagnosis and treatment plan, providing a broader picture of patient health status. The data were preprocessed to be ready for clustering, including one hot encoding to convert categorical features into binary variables, imputation for missing values, and scaling to ensure all feature are on the same scale. The Elbow Method was used to determine the optimal number of clusters, which was found to be three. A number of visualization techniques were used to explore a little further the distribution of features in the clustering results, such as box plots, radar charts, and silhouette plots. The results showed three distinct patient clusters in terms of age, cardiovascular indicators, metabolic features, and therapy patterns. One cluster showed higher values for measurements such as blood pressure, heart rate, and BMI, suggesting a high-risk health profile. The remaining clusters showed moderate or mixed characteristics, with one showed distinct clinical and physiological characteristics. Although the distributions from the silhouette scores show some overlap between clusters, the results suggest that K-Means clustering can be used to uncover underlying health profiles in structured data from healthcare environments. | ||
| کلیدواژهها | ||
| K-Means clustering؛ healthcare data؛ unsupervised learning؛ patient profiling؛ clinical decision support | ||
|
آمار تعداد مشاهده مقاله: 4 |
||