대한신장학회

논문분류	춘계학술대회 초록집
제목	Optimal Clustering of Patient Dispositions for Shared Decision-Making in Dialysis Modality Selection
저자	Jiwon Ryu
출판정보	2026; 2026(1):
키워드	Dialysis mordality, Shared Decision Making , patient-centered care, Artificial Intelligence
초록	Objectives: Dialysis modality selection for patients with chronic kidney disease (CKD) has traditionally been physician-driven. With the shift toward patient-centered care, Shared Decision Making (SDM) is expected to improve treatment adherence and satisfaction. This study implemented an SDM framework for CKD patients approaching dialysis and examined patient preferences and dispositions to support personalized decision-making. Methods: This multicenter study included 693 patients from 19 hospitals. An 18-item questionnaire assessed health, lifestyle, and environmental factors, including decision-making self-efficacy, decisional certainty, and self-management. Validated scales measured SDM processes and hospital service satisfaction. Clustering was performed using an iterative weight-sampling approach with a weighted Gower distance matrix. The optimal K-Medoids model was selected by maximizing the Silhouette score across iterations. Results: The five-cluster model (K=5, silhouette score: 0.252) emerged as optimal for classifying patient dispositions. Although K=4 showed higher silhouette score (0.297), it produced redundant patterns with nearly identical clusters, suggesting artificial splitting. K=6 (0.221) resulted in over-segmentation and reduced interpretability. The K=5 model provided distinct, non-overlapping patterns: Cluster 1 (n=108) exhibited high health-consciousness, strong autonomy, and excellent adaptability; Cluster 2 (n=181) showed high health interest but difficulty integrating it into daily life; Cluster 3 (n=93) demonstrated balanced functioning but environmental vulnerability; Cluster 4 (n=187) displayed superior self-regulation and rapid adaptation; Cluster 5 (n=124) represented baseline profiles with lower engagement yet moderate adaptability. Conclusion: Categorizing pre-dialysis patients into five distinct profiles enables tailored educational interventions and optimized SDM. Integrating these patient-centered dispositions into clinical practice is expected to significantly enhance treatment satisfaction and long-term adherence to chosen dialysis therapy.
원문(PDF)	PDF 원문보기

간행물 검색

대한신장학회 회원관리 규정