贵州农村老年人轻度认知功能损害患病风险评估模型研究

Evaluating model for the risk factors of mild cognitive impairment among rural elderly in Guizhou Province

目的通过机器学习分析脑健康生活方式相关因素和人口统计学特征,实现对贵州农村老年人轻度认知功能损害患病风险的评估效果研究。方法 2019年7—8月,采用多阶段整群抽样方法,选取贵州省1 235名60岁及以上的农村老年人为研究对象,进行问卷调查和体格检查。采用简易智能状态量表(mini-mental state examination,MMSE)进行认知功能检测,并基于12项脑健康生活方式相关因素和4项人口统计学特征进行重要特征及最优特征个数的选择。采用逻辑回归和随机森林算法构建贵州农村老年人轻度认知功能损害的患病风险评估模型,并使用精确率、准确率、灵敏度、特异度、F1分数和受试者工作特征曲线下面积(area under curve,AUC)综合评估模型效能,采用Delong法检验两模型间AUC值的差异。结果共检出轻度认知功能损害291例,总检出率为23.56%(291/1 235)。逻辑回归和随机森林模型评估贵州农村老年人轻度认知功能损害患病风险的AUC值分别为0.758和0.820,差异有统计学意义(均P<0.05)。其中随机森林模型的评估效果更佳,精确率为0.823、准确率为0.805、灵敏度为0.874、特异度为0.767、F1分数为0.838,各指标均优于逻辑回归模型;且经过10折交叉验证,随机森林模型也更具稳定性。结论脑健康生活方式相关因素结合人口学特征因素能较准确地评估贵州农村老年人轻度认知功能损害患病风险,且随机森林模型评估效能优于逻辑回归模型。

Objective To analyze the lifestyle for brain health related factors and demographic characteristics through machine learning to achieve the assessing effect of mild cognitive impairement prevalence risk among rural elderly people in Guizhou.Methods From July to August 2019,a multi-stage cluster random sampling method was used to select 1235 rural elderly people aged 60 years and above in Guizhou Province as the subjects,and the investigation was performed with questionnaire and physical examination.The Mini-Mental State Examination(MMSE)was used to assess cognitive function,and the important features and optimal number of features based on 12 LIBRA factors and 4 demographic characteristics were selected.Logistic regression and random forest algorithm were used to establish a evaluation model for the risk of mild cognitive impairment in the elderly.The evaluation efficacy of the model was also assessed using a combination of precision,accuracy,sensitivity,specificity,F1 score and area under curve(AUC)of receiver operating characteristic curve,and the Delong method was used to check the difference of AUC values between the two models.Results A total of 291 subjects were diagnosed with mild cognitive impairment,with a detection rate of 23.56%(291/1235).The AUC values of logistic regression and random forest models evaluating the risk of mild cognitive impairment in the rural elderly were 0.758 and 0.820,respectively,and the differences were statistically significant(both P<0.05).The random forest model had better evaluations with an accuracy of 0.823,precision of 0.805,sensitivity of 0.874,specificity of 0.767 and F1 score of 0.838,all of which were better than those of the logistic regression model.And the random forest model was also more stable after 10-fold cross-validation.Conclusion The lifestyle for brain health related factors combined with demographic characteristics can more accurately evaluate the risk of mild cognitive impairment among rural elderly people in Guizhou.The random forest model is better than the logistic regression model.