Cui Yuyu, Xu Zhening, Cui Zhaoshu, Guo Yuanyuan, Wu Peiwei, Zhou Xiaoyan
School of Medicine, Yan'an University, Yan'an, 716000, China.
Sci Rep. 2025 Jan 30;15(1):3814. doi: 10.1038/s41598-025-88275-y.
The study aims to develop and validate an effective model for predicting frailty risk in individuals with mild cognitive impairment (MCI). The cross-sectional analysis employed nationally representative data from CHARLS 2013-2015. The sample was randomly divided into training (70%) and validation sets (30%). The least absolute shrinkage and selection operator (LASSO) and multivariable logistic regression model were used to identify independent predictors and establish a nomogram to predict the occurrence of frailty. The receiver operating characteristic (ROC) curve, the calibration curve, and the decision curve analysis (DCA) were used to evaluate the performance of the nomogram. A total of 3,196 MCI patients were analyzed, and 803 (25.1%) exhibited symptoms of frailty. Multivariate logistic regression analysis revealed that age, activities of daily living (ADL) score, depression score, grip strength, cardiovascular disease (CVD), liver disease, pain, hearing, and vision were associated factors for frailty in MCI patients. The nomogram based on these factors achieved AUC values of 0.810 (95% CI 0.780, 0.840) in the training set and 0.791 (95% CI 0.760, 0.820) in the validation set. Calibration curves showed good agreement between the nomogram and the observed values. The Hosmer-Lemeshow test results for the training and validation sets were P = 0.396 and P = 0.518, respectively. The ROC curve and decision curve analysis further validated the robust predictive ability of the nomogram. The application of this model may facilitate early clinical interventions, thereby potentially reducing the incidence of frailty among patients with MCI and significantly enhancing their long-term health outcomes.
该研究旨在开发并验证一种有效的模型，用于预测轻度认知障碍（MCI）个体的衰弱风险。横断面分析采用了2013 - 2015年中国健康与养老追踪调查（CHARLS）具有全国代表性的数据。样本被随机分为训练集（70%）和验证集（30%）。使用最小绝对收缩和选择算子（LASSO）及多变量逻辑回归模型来识别独立预测因子，并建立列线图以预测衰弱的发生。采用受试者工作特征（ROC）曲线、校准曲线和决策曲线分析（DCA）来评估列线图的性能。共分析了3196例MCI患者，其中803例（25.1%）表现出衰弱症状。多变量逻辑回归分析显示，年龄、日常生活活动（ADL）评分、抑郁评分、握力、心血管疾病（CVD）、肝脏疾病、疼痛、听力和视力是MCI患者衰弱的相关因素。基于这些因素的列线图在训练集中的AUC值为0.810（95%CI 0.780, 0.840），在验证集中为0.791（95%CI 0.760, 0.820）。校准曲线显示列线图与观察值之间具有良好的一致性。训练集和验证集的Hosmer-Lemeshow检验结果分别为P = 0.396和P = 0.518。ROC曲线和决策曲线分析进一步验证了列线图强大的预测能力。该模型的应用可能有助于早期临床干预，从而有可能降低MCI患者衰弱的发生率，并显著改善他们的长期健康结局。
