Identification of Risk Group for Root Caries and Analysis of Associated Factors in Older Adults Using Unsupervised Machine Learning Clustering

• DOI: 10.2147/cia.s520229
• Publication Date: 2025-04-24T09:55:11Z
• Authors: Linxin Jiang; Shaohong Huang; Daniel Reissmann; Gerhard Schmalz; Jianbo Li
• Keywords: Identification

This study aimed to identify the high-risk group for root caries using unsupervised machine learning and explore associated factors. cross-sectional included 423 adults aged 65 74 years, surveyed in 2021. Clusters representing risk were identified k-prototypes clustering, with optimal number of clusters determined by maximum silhouette index. The confusion matrix alluvial diagram used visualize predictive accuracy composition clustering results. Binary logistic regression models further analyzed factors, while ROC (receiver operating characteristic) curves random forest model visualized performance most important Two identified: cluster 1, low (21.5% 78.5% without caries), 2, high (83.7% 16.3% caries). results predicted an 0.81, sensitivity 0.79, specificity 0.83. Overlapping from binary indicated that older age, more periodontal pockets, attachment loss, female, a history systemic diseases, presence xerostomia, unrestored tooth loss positively 2. Brushing ≥2 times per day level oral health knowledge negatively curve showed AUC (area under curve) 0.84. Individuals who are older, poorer status, suboptimal hygiene behaviors, lower levels likely be as group. revealed through learning, can facilitate personalized prevention management strategies adults.