Clustering Analysis for Anemia Risk Profiling in Hospital Patients Using K-Prototypes Approach
Journal of Electronics, Electromedical Engineering, and Medical Informatics · 24.07.2026 · 8 (3) · 1260-1276
DOI: 10.35882/jeeemi.v8i3.1723 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Electronics, Electromedical Engineering, and Medical Informatics, online erschienen am 24.07.2026. Verfasst von Oki Setiono und 5 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-SA 4.0).
Abstract (Originalsprache)
Anemia in inpatients represent a complex clinical condition often associated with systemic responses such as inflammation and infection. However, conventional univariate approaches based solely on hemoglobin (Hb) levels frequently fail to capture the multidimensional nature of anemia risk, particularly in mixed-type clinical datasets. This study aims to develop an anemia risk stratification framework by integrating hematological and demographic variables using the K-Prototypes clustering algorithm. The dataset consisted of 587 patient records collected from multiple hospitals in Central Java, Indonesia, including hemoglobin, leukocytes, platelets, age, and sex. Data preprocessing involved cleaning, standardization, and mixed-data transformation prior to clustering analysis. Multiple cluster configurations ( = 3, = 4, and = 5) were evaluated using the Elbow Method and clustering validation metrics, including Silhouette Score, Davies–Bouldin Index, and Calinski–Harabasz Index. The results identified = 4 as the optimal clustering configuration, providing the best balance between cluster separation and interpretability. Subsequently, the identified clusters were clinically interpreted into three risk categories, including High Risk, Moderate Risk, and Normal Risk. The High-Risk group exhibited the lowest mean Hb level (9.27 g/dL), while the Moderate Risk group was characterized by elevated leukocyte counts (19.2k/µL), suggesting distinct hematological patterns. The Normal Risk group demonstrated relatively stable hematological profiles and higher mean age. Statistical testing confirmed significant differences among risk profiles for age, hemoglobin, leukocytes, platelets, and gender ( < 0.001). These findings demonstrate that anemia-related risk patterns are influenced by multidimensional interactions among hematological and demographic factors. The proposed framework provides clinically meaningful patient stratification and has potential applications in electronic medical record systems and clinical decision support environments.
Zitieren
APA
Setiono, O., Sugiyanto, Z., Ernawati, D., Kurniadi, A., Pantiawati, I., & Husin, M. N. (2026). Clustering Analysis for Anemia Risk Profiling in Hospital Patients Using K-Prototypes Approach. Journal of Electronics, Electromedical Engineering, and Medical Informatics, 8(3), 1260-1276. https://doi.org/10.35882/jeeemi.v8i3.1723
Vancouver
Setiono O, Sugiyanto Z, Ernawati D, Kurniadi A, Pantiawati I, Husin MN. Clustering Analysis for Anemia Risk Profiling in Hospital Patients Using K-Prototypes Approach. Journal of Electronics, Electromedical Engineering, and Medical Informatics. 2026;8(3):1260-1276. doi:10.35882/jeeemi.v8i3.1723
BibTeX
@article{Setiono2026,
author = {Setiono, Oki and Sugiyanto, Zaenal and Ernawati, Dyah and Kurniadi, Arif and Pantiawati, Ika and Husin, Mohamad Nazri},
title = {Clustering Analysis for Anemia Risk Profiling in Hospital Patients Using K-Prototypes Approach},
journal = {Journal of Electronics, Electromedical Engineering, and Medical Informatics},
year = {2026},
volume = {8},
number = {3},
pages = {1260--1276},
doi = {10.35882/jeeemi.v8i3.1723}
}
EUVIMED