Page 100 - 《南京医科大学学报》2026年第1期
P. 100
南京医科大学学报(自然科学版) 第46卷第1期
· 94 · Journal of Nanjing Medical University(Natural Sciences) 2026年1月
·临床研究·
癫痫患者认知功能障碍的影响因素及模型构建——基于
NHANES数据库分析
葛怀廷 ,王 刚 ,付美芝 ,魏 伟 ,曲立新 ,聊娜娜 2*
1
1
1
1
1
山东大学齐鲁医院德州医院神经内科,山东 德州 253052;德州市妇幼保健院儿童保健科,山东 德州 253052
1 2
[摘 要] 目的:基于Lasso回归构建癫痫患者认知功能障碍预测模型,并分析该模型的预测效能。方法:采用横断面研究,纳
入美国国家健康与营养调查(National Health and Nutrition Examination Survey,NHANES)数据库的癫痫患者,依据数字符号替
换测试(digit symbol substitution test,DSST)总分将参与者分为认知功能障碍组(DSST<34分)和认知功能正常组(DSST≥34分)。
收集人口学资料、社会经济学特征、体力活动情况、疾病资料以及25⁃羟基维生素D[25⁃hydroxyvitamin D,25(OH)D]、神经丝轻
链蛋白(neurofilament light chain,NfL)等实验室检测指标,应用Lasso回归筛选非0系数变量,构建多因素Logistic回归模型分析
癫痫患者认知功能障碍的影响因素并构建列线图预测模型,应用受试者工作特征(receiver operating characteristic,ROC)曲线、
Bootstrap 校准曲线和 Hosmer⁃Lemeshow 拟合优度检验评价列线图模型对癫痫患者认知功能障碍的预测能力。结果:共纳入
282 例研究对象,癫痫患者认知功能障碍发生率为 32.62%。经 Lasso 降维处理和 Logistic 回归共筛选得到血尿酸、NfL、血清
25(OH)D和教育程度等癫痫患者认知功能障碍的影响因素。基于上述影响因素构建列线图预测模型,ROC曲线显示模型的
曲线下面积为 0.833(95% CI:0.780~0.886),特异度为 0.804,灵敏度为 0.781。Bootstrap 校准曲线显示实测概率与列线图预测
概率之间具有较高的一致性;Hosmer⁃Lemeshow 拟合优度检验结果也表明列线图模型的拟合程度较好(χ =7.781,P=0.455)。
2
结论:癫痫患者认知功能障碍的发生受到血尿酸、NfL、血清25(OH)D和教育程度等因素的影响,基于Lasso回归构建的癫痫患
者认知功能障碍预测模型显示出良好的预测效能,为临床评估和干预提供了重要依据,有助于提高癫痫患者的预后管理效果。
[关键词] Lasso回归;癫痫;认知功能障碍;列线图
[中图分类号] R742.1 [文献标志码] A [文章编号] 1007⁃4368(2026)01⁃94⁃09
doi:10.7655/NYDXBNSN250481
The influencing factors and model construction of cognitive dysfunction in epilepsy
patients——analysis based on NHANES database
1
1
1
1
1
GE Huaiting ,WANG Gang ,FU Meizhi ,WEI Wei ,QU Lixin ,LIAO Nana 2*
1 Department of Neurology,Qilu Hospital of Shandong University,Dezhou 253052;Department of Child Healthcare,
2
Maternity & Child care Center of Dezhou,De zhou 253052,China
[Abstract] Objective:A predictive model for cognitive dysfunction in epileptic patients was constructed using Lasso regression,and
the predictive efficacy of this model was analyzed. Methods:A cross⁃sectional study was conducted,including epileptic patients from
the National Health and Nutrition Examination Survey(NHANES)database. Participants were categorized into a cognitive dysfunction
(DSST<34)group and a normal cognitive function group(DSST≥34)based on the total score of the digit symbol substitution test
(DSST). Data on demographics,socioeconomic characteristics,physical activity,medical history,as well as laboratory indicators such
as 25⁃hydroxyvitamin D[25(OH)D]and neurofilament light chain(NfL)were collected. Lasso regression was used to select non⁃zero
coefficient variables,and a multivariate logistic regression model was constructed to analyze factors affecting cognitive dysfunction in
epileptic patients and to develop a nomogram prediction model. The model’s predictive ability was evaluated using receiver operating
characteristic(ROC)curves,Bootstrap calibration curves,and Hosmer ⁃ Lemeshow goodness ⁃ of ⁃ fit tests. Results:A total of 282
subjects were included,with a cognitive dysfunction incidence of 32.62% among the epileptic patients. Factors identified through
[基金项目] 山东省医药卫生科技发展计划项目(2017WS765)
通信作者(Corresponding author),E⁃mail:liaonana01@163.com(ORCID:0009⁃0003⁃0660⁃2379)
∗

