关键词: 老年人, 认知障碍, 社区, 列线图, 预测模型

Abstract: Background With the further aging of the population, the incidence of cognitive impairment is increasing, and there is a lack of effective treatment methods, constructing an accurate risk prediction model, with a view to helping community healthcare workers to identify, warn and intervene in the early stage of potential patients, and to reduce the social healthcare pressure. Objective In this study, we proposed to construct a prediction model for the risk of cognitive impairment in community-dwelling older adults, analyze the influencing factors of cognitive impairment in older adults, and provide empirical references for the development of targeted interventions. Methods In this study, data from the China Health and Elderly Care Tracking Survey Database 2020 were used, and the data were randomly divided into a training set (n=5133) and a validation set (n=2201) in a ratio of 7:3. Lasso regression with ten-fold cross-validation was used to screen for the best predictor variables, and the Logistic Regression Model was used to analyze the factors affecting cognitive impairment in the elderly, and to construct a column chart, and the ROC curve was constructed using the area under the curve of the ROC curve. The predictive performance of the column-line diagrams was assessed using the area under the curve of the ROC curve, and calibration curve, and other analysis. Results A total of 7334 older adults were screened, of which 1062 were cognitively impaired. Lasso regression screened 9 potential predictor variables, including age, type of residence, marital status, gender, education, exercise, socialization, ability to perform activities of daily living, and depression. Multifactorial logistic regression analysis showed that age [70-79 years OR=1.238, 95% CI=1.109～1.504; ≥80 years OR=2.231, 95% CI=1.546～3.222], type of residence (OR=2.144, 95% CI=1.617～2.842), marital status (OR= 0.691, 95% CI=0.562～0.851), educational attainment [elementary school and below OR=0.209, 95% CI=0.173～0.254, junior high school OR=0.059, 95% CI=0.038～ 0.090, high school or vocational high school OR=0.043, 95% CI=0.021～0.089, college and above OR=0.038, 95% CI=0.005 ～0.280], socialization (OR=0.746, 95% CI=0.624～0.892), ability to perform activities of daily living (OR=1.529, 95% CI=1.171～1.997), and depression (OR=1.580, 95% CI=1.319～1.891) were the independent influences of cognitive impairment (P<0.05), and a column-line graph prediction model was established based on the seven predictor variables screened by multifactor regression analysis. The areas under the ROC curves of the prediction model in the training and validation sets were 0.821 (95% CI=0.805～0.836) and 0.839 (95% CI=0.817～0.861), respectively; the Hosmer-Lemeshow〖 χ〗^2=5.022、P=0.755, and〖 χ〗^2=3.963、P=0.860; and the calibration curves showed that there was significant consistency between the predicted and actual values. significant agreement between predicted and actual values. Conclusion In this study, we established a cognitive impairment risk prediction model for community-dwelling older adults with seven indicators, including age, type of residence, etc. The prediction model was accurate and discriminative, and can be used to identify the risk of cognitive impairment in older adults.

Key words: Older adults, Cognitive impairment, Communities, Columnar graphs, Predictive modeling