基于Lasso回归的老年卒中后疲劳风险预测模型的构建与验证-现代医学

基于Lasso回归的老年卒中后疲劳风险预测模型的构建与验证

单位：1. 广西中医药大学护理学院, 广西南宁 530200;
2. 广西中医药大学第一附属医院脑病科, 广西南宁 530023

分类号：R743.3

出版年·卷·期（页码）：2024·52·第四期（530-536）

摘要：

目的：分析老年缺血性脑卒中(IS)患者急性期发生卒中后疲劳(PSF)的影响因素，构建老年IS患者急性期发生PSF的风险预测模型，并进行内外部验证评价模型预测效果。方法：选取2022年10月至2023年10月于南宁市某2家三甲医院住院治疗的老年IS患者共438例(训练集330例，验证集108例)。应用Lasso回归筛选预测变量，并根据筛选的预测变量构建多因素Logistic回归模型，绘制模型的列线图。模型的验证和评估主要基于受试者工作特征曲线下面积(AUC)、Hosmer-Lemeshow检验(H-L检验)，并根据决策曲线分析(DCA)法评价该模型的实际临床净收益。结果：基于Lasso回归惩罚收缩方法筛选出年龄、糖尿病史、既往卒中史、规律锻炼4个预测变量，进一步构建Logistic 回归模型显示，高龄、糖尿病史、既往卒中史是发生PSF的危险因素，规律锻炼是发生PSF的保护因素(P<0.05)。内外部验证显示该模型预测效能较优，训练集AUC为0.836(95%CI 0.792~0.880)，灵敏度为73.0%，特异度为83.6%，约登指数为0.566；验证集AUC为0.798(95%CI 0.712~0.884)，灵敏度为82.0%，特异度为68.1%，约登指数为0.501；H-L检验(训练集χ²=10.452,P=0.235；验证集χ²=7.042,P=0.425)均显示该模型具有较好的一致性和拟合度，DCA结果显示当训练集和验证集的阈概率分别为0.10~0.94和0.18~0.99时，该模型临床有效性较高。结论:本研究构建的模型预测效果较好，可为临床护理人员识别老年IS患者急性期发生PSF及实施预防性护理提供借鉴。

Objective: To analyze the influencing factors of post-stroke fatigue(PSF) in the acute phase of elderly patients with ischemic stroke(IS), to construct a risk predictive model for PSF in the acute phase of elderly IS patients, and to evaluate the prediction effect of the model by internal and external verification. Methods: A total of 438 elderly IS patients(330 in the training set and 108 in the validation set) were selected from October 2022 to October 2023 in two tertiary hospitals in Nanning. Lasso regression was used to screen the predictive variables, and a multivariate Logistic regression model was constructed based on the screened predictive variables, and the Nomogram of the model was drawn. The validation and evaluation of the model were mainly based on the area under curve(AUC) of the receiver operating characteristic curve, the Hosmer-Lemeshow test(H-L test), and the actual clinical net benefit of the model was evaluated according to the decision curve analysis(DCA). Results: Based on the Lasso regression penalty contraction method, four predictors of age, diabetes history, previous stroke history, and regular exercise were selected. Further construction of the Logistic regression model showed that higher age, diabetes history, and previous stroke history were risk factors for PSF, and regular exercise was a protective factor for PSF(P<0.05). The internal and external validation showed that the model had better predictive performance. The AUC of the training set was 0.836(95%CI 0.792-0.880), the sensitivity was 73.0%, the specificity was 83.6%, and the Youden index was 0.566. The AUC of the validation set was 0.798(95%CI 0.712-0.884), the sensitivity was 82.0%, the specificity was 68.1%, and the Youden index was 0.501. The H-L test(training set χ²=10.452, P=0.235, validation set χ²=7.042,P=0.425,) showed that the model had good consistency and goodness of fit. The DCA results showed that when the threshold probabilities of training set and validation set were 0.10~0.94 and 0.18~0.99, respectively, the clinical effectiveness of the model was higher. Conclusion: The model constructed in this study has a good prediction effect, which can provide reference for clinical nurses to identify PSF in the acute phase of elderly IS patients and implement preventive nursing.

参考文献：

[1] 王梅杰,邓雨芳,周翔,等.加拿大《最佳实践建议:卒中后抑郁、认知、疲劳》解读[J].中国全科医学,2021,24(17):2214-2217.
[2] ENGLISH C,SIMPSON D B,BILLINGER S A,et al.A roadmap for research in post-stroke fatigue:consensus-based core recommendations from the third stroke recovery and rehabilitation round table[J].Int J Stroke,2024,38(1):7-18.
[3] 李昱,彭敏,余浩佳,等.脑梗死后早期疲劳的发生率及其相关危险因素[J].医学与哲学,2018,39(6):42-45.
[4] HANWEN Z,GUIDONG L,LI Z,et al.Personalized biomarkers and neuropsychological status can predict post-stroke fatigue[J].Brain Sci,2023,13(2):295.
[5] REN W,WU J,WU Z,et al.Serum uric acid levels at admission could predict the chronic post-stroke fatigue[J].Front Nutr,2022,9:850355.
[6] HILL G,REGAN S,FRANCIS R,et al.Research priorities to improve stroke outcomes[J].Lancet Neurol,2022,21(4):312-313.
[7] 中国老年学和老年医学学会.老年缺血性脑卒中慢病管理指南[J].中西医结合研究,2022,14(6):382-392.
[8] LANCTOT K L,LINDSAY M P,SMITH E E,et al.Canadian stroke best practice recommendations:mood,cognition and fatigue following stroke,6th edition update 2019[J].Int J Stroke,2020,15(6):668-688.
[9] 高雅云,林丽.脑卒中后疲劳列线图预测模型的构建及评估[J].中外医学研究,2023,21(4):139-143.
[10] SU Y,YUKI M,HIRAYAMA K,et al.Development and internal validation of a nomogram to predict post-stroke fatigue after discharge[J].J Stroke Cerebrovasc Dis,2021,30(2):105484.
[11] 中华医学会神经病学分会,中华医学会神经病学分会脑血管病学组.中国各类主要脑血管病诊断要点2019[J].中华神经科杂志,2019,52(9):710-715.
[12] 中华医学会神经病学分会,中华医学会神经病学分会脑血管病学组.中国脑血管病一级预防指南2019[J].中华神经科杂志,2019,52(9):684-709.
[13] 臧爽,崔赢,倪翠萍,等.卒中后疲劳评估工具的范围综述[J].中华护理杂志,2023,58(1):46-54.
[14] HINKLE J L,BECKER K J,KIM J S,et al.Poststroke fatigue:emerging evidence and approaches to management a scientific statement for healthcare professionals from the American Heart Association[J].Stroke,2017,48(7):E159-E170.
[15] 邢晓慧,王明明,姚昱名,等.老年脑卒中后失语患者评估和康复护理的最佳证据总结[J].现代医学,2023,51(10):1482-1488.
[16] 陈丽.卒中后疲劳的发生及可能的危险因素研究[J].中国病案,2023,24(1):80-82.
[17] 张爱琴,余金甜,陈俊杉.脑卒中患者发生急性期疲劳危险因素的Meta分析[J].中国循证医学杂志,2020,20(3):272-280.
[18] ZHANG X,FANG H,MA D,et al.Risk factors and imaging mechanisms of fatigue after mild ischemic stroke:an exploratory study from a single Chinese center[J].Front Neurol,2021,12:649021.
[19] 张东,李治璋,马瑞楠,等.应激性高血糖比值与急性缺血性卒中患者临床预后的相关性研究[J].中国卒中杂志,2022,17(5):483-490.
[20] LARSSON P,BIDONDE J,OLSEN U,et al.Association of post-stroke fatigue with physical activity and physical fitness:a systematic review and Meta-analysis[J].Int J Stroke,2023,18(9):1063-1070.
[21] TAI D,FALCK R S,DAVIS J C,et al.Can exercise training promote better sleep and reduced fatigue in people with chronic stroke? A systematic review[J].J Sleep Res,2022,31(6):e13675.
[22] 曾晓东,熊兰,刘国勤,等.脑卒中患者发生急性期疲劳的危险因素分析[J].实用心脑肺血管病杂志,2020,28(7):54-58.

服务与反馈：

【文章下载】【发表评论】【查看评论】【加入收藏】

提示：您还未登录，请登录！点此登录