基于随机森林的多模态神经电生理指标预测老年脑卒中患者衰弱风险模型研究

doi:10.3760/cma.j.cn441417-20240910-06018

摘要/Abstract

摘要：

目的　构建基于多模态神经电生理指标的老年脑卒中患者衰弱风险预测模型，探索其在早期干预中的应用价值。方法　采用回顾性分析，选取2021年1月至2023年1月山东第一医科大学附属东平医院收治的265例老年首次急性脑卒中住院患者，男148例、女117例，年龄（70.42±8.63）岁，缺血性卒中220例、出血性卒中45例，收集基线资料及神经电生理指标。采用Fried衰弱表型量表评估衰弱状态，使用独立样本t检验、秩和检验、χ²检验进行单因素分析，logistic回归进行多因素分析。采用随机森林算法构建预测模型并与logistic回归模型进行比较。结果　265例患者分为训练组212例，验证组53例，两组一般特征、临床指标、实验室检查和神经电生理指标等基线资料比较，差异均无统计学意义（均P>0.05）。衰弱组98例、非衰弱组167例，两组患者年龄、美国国立卫生研究院卒中量表（NIHSS）评分、白蛋白、神经电生理指标及功能、认知评分等比较，差异均有统计学意义（均P<0.05）。多因素分析显示，基线NIHSS评分（OR=1.17）、蒙特利尔认知评估量表（MoCA）评分（OR=0.87）、年龄（OR=1.09）、经颅磁刺激（TMS）皮质静息期（OR=1.11）、白蛋白（OR=0.89）、定量脑电图（qEEG）δ波功率（OR=1.05）和老年抑郁评估量表（GDS-15）评分（OR=1.18）是脑卒中后衰弱的独立预测因子（均P<0.05）。随机森林模型受试者操作特征曲线（ROC）的曲线下面积（AUC=0.728）高于多因素logistic回归模型（AUC=0.629），P=0.029。结论　整合多模态神经电生理指标和临床特征可显著提高老年脑卒中后衰弱的预测精度。

关键词:

脑卒中, 老年, 衰弱, 神经电生理, 预测模型, 随机森林, Logistics回归

Abstract:

Objective　To construct a prediction model of frailty risk in elderly patients with stroke based on multimodal neuroelectrophysiological indicators, and to explore its application value in early intervention. Methods　This was retrospective study. Two hundred and sixty-five elderly patients with first acute stroke treated at Dongping Hospital, Shandong First Medical University from January 2021 to January 2023 were selected, including 148 males and 117 females. They were (70.42±8.63) years old. There were 220 cases of ischemic stroke and 45 cases of hemorrhagic stroke. Their baseline data and neurophysiological indicators were collected. The fried frailty phenotype was used to evaluate their frailty status. Independent-sample t test, rank sum test, and χ² test were used for the univariate analysis. The logistic regression was used for the multivariate analysis. The random forest algorithm was used to construct a prediction model and compared with the logistic regression model. Results　Among the 265 patients, there were 212 cases in the training group and 53 cases in the testing group; there were no statistical differences in the general features, clinical data, laboratory examination, and neuroelectrophysiological indicators between the two groups (all P>0.05). There were 98 cases in the frailty group and 167 cases in the non-frailty group; there were statistical differences in the age, score of National Institute of Health stroke scale (NIHSS), albumin, neuroelectrophysiological indicators, and scores of function and cognition between the two groups (all P<0.05). The multivariate analysis showed that the baseline score of NIHSS (OR=1.17), score of Montreal Cognitive Assessment (MoCA) (OR=0.87), age (OR=1.09), transcranial magnetic stimulation (TMS) cortical resting period (OR=1.11), albumin (OR=0.89), quantitative electroencephalography (qEEG) δ wave power (OR=1.05), and score of Geriatric Depression Scale (GDS-15) (OR=1.18 ) were the independent predictors of frailty after stroke (all P<0.05). The area under the receiver operating characteristic curve (ROC) of the random forest model was significantly higher than that of the multivariate logistic regression model (0.728 vs. 0.629, P=0.029). Conclusion　The integration of multimodal neuroelectrophysiological indicators and clinical features can significantly improve the prediction accuracy of post-stroke fraility.

Key words:

Stroke, Elderly, Frailty, Neuroelectrophysiology, Prediction model, Random forest, Logistic regression

张晓雯秦丽云陈文倩孙毅.

基于随机森林的多模态神经电生理指标预测老年脑卒中患者衰弱风险模型研究 [J]. 国际医药卫生导报, 2025, 31(6): 973-978.

Zhang Xiaowen, Qin Liyun, Chen Wenqian, Sun Yi.

A random forest-based multimodal neural electrophysiological index model for predicting frailty risk in elderly patients with stroke [J]. International Medicine and Health Guidance News, 2025, 31(6): 973-978.

参考文献

[1] 庞媛月, 曹童童, 董雪. 老年脑卒中患者衰弱相关研究进展[J]. 老年医学研究, 2024, 5(1): 56-61. DOI:10.3969/j.issn.2096-9058.2024.01.010.
[2] Vazquez S, Stadlan Z, Lapow JM, et al. Frailty and outcomes in lacunar stroke[J]. J Stroke Cerebrovasc Dis, 2023, 32(2): 106942. DOI:10.1016/j.jstrokecere- brovasdis. 2022.106942
[3] 黄亚楠, 王鑫钰, 陈燕茹, 等. 脑卒中患者合并衰弱的研究进展[J]. 中国护理管理, 2023, 23(2): 282-286. DOI:10.3969/j.issn.1672-1756.2023.02.025.
[4] 阳佳家, 李凤, 刘光维, 等. 脑卒中病人衰弱评估的研究进展[J]. 护理研究, 2023, 37(21): 3878-3885. DOI:10.12102/j.issn.1009-6493.2023.21.015.
[5] 林伶姿, 叶慧玲, 吴洁媛, 等. 老年脑卒中患者衰弱与营养风险的相关性研究[J]. 国际医药卫生导报, 2021, 27(20): 3243-3247. DOI:10.3760/cma.j.issn.1007-1245.2021. 20.034.
[6] 陈张静, 孔娴波, 王国嫔, 等. 老年急性缺血性脑卒中患者治疗后发生老年衰弱综合征预测模型的建立及验证[J]. 中华老年心脑血管病杂志, 2023, 25(12): 1336-1339. DOI:10.3969/j.issn.1009-0126.2023.12.025.
[7] 张敏, 赵学明, 段虎斌. 神经电生理监测在三叉神经痛诊疗中的应用进展[J]. 中国疼痛医学杂志, 2020, 26(10): 769-773. DOI:10.3969/j.issn.1006-9852.2020.10.009.
[8] 张鑫宇, 张磊, 隋汝波. 基于Logistic回归和人工神经网络构建老年脑卒中患者衰弱预测模型[J]. 军事护理, 2023, 40(2): 10-14,19. DOI:10.3969/j.issn.2097-1826. 2023.02.003.
[9] 路毅, 邓文冲. 不同运动方式对大脑结构及认知功能的调节作用及差异[J]. 中国组织工程研究, 2021, 25(20): 3252-3258.
[10] 刘田田, 刘恒方. 神经重症多模态脑监测的研究进展[J]. 国际神经病学神经外科学杂志, 2022, 49(6): 92-96. DOI:10.16636/j.cnki.jinn.1673-2642.2022.06.018.
[11] 汤翔嵘, 王晓刚, 陈桃林, 等. 精神障碍的神经电生理循证医学证据[J]. 生物化学与生物物理进展, 2021, 48(10): 1157-1176.DOI:10.16476/j.pibb.2020.0399.
[12] Ferreira LO, De Souza RD, Teixeira LL, et al. The GPER1 agonist G1 reduces brain injury and improves the qEEG and behavioral outcome of experimental ischemic stroke[J]. J Neuropathol Exp Neurol, 2023, 82(9): 787-797. DOI:10.1093/jnen/nlad061
[13] 刘秀颖, 蓝瑞芳. 急性缺血性脑卒中定量脑电图特征与CT灌注成像参数的相关性[J]. 上海交通大学学报(医学版), 2021, 41(1): 62-65. DOI:10.3969/j.issn.1674-8115. 2021.01.010
[14] Ferreira LO, Mattos BG, Jóia De Mello V, et al. Increased relative delta bandpower and delta indices revealed by continuous qEEG monitoring in a rat model of ischemia-reperfusion[J]. Front Neurol, 2021, 12: 645138. DOI:10.3389/fneur.2021.645138
[15] 黎晋添. 以事件关联电位评价身体活动量对老年人执行控制功能的影响[J]. 中国组织工程研究, 2019, 23(11): 1693-1699. DOI:10.3969/j.issn.2095-4344.1137.
[16] 孙晓琴, 冯英, 肖农. 事件相关电位P300在意识障碍预后评估中的应用进展[J]. 中国康复理论与实践, 2020, 26(7): 784-787.DOI:10.3969/j.issn.1006?9771.2020.07.009.
[17] Hernandez-Pavon JC, Veniero D, Bergmann TO, et al. TMS combined with EEG: recommendations and open issues for data collection and analysis[J]. Brain Stimul, 2023, 16(2): 567-593. DOI:10.1016/j.brs.2023.02.009
[18] Sood I, Injety RJ, Farheen A, et al. Quantitative electroencephalography to assess post-stroke functional disability: a systematic review and meta-analysis[J]. J Stroke Cerebrovasc Dis, 2024, 33(12): 108032. DOI:10.1016/j.jstrokecerebrovasdis.2024.108032
[19] Kesebir S, Yosmaoglu A, Tarhan N. A dimensional approach to affective disorder: the relations between Scl-90 subdimensions and QEEG parameters[J]. Front Psychiatry, 2022, 13: 651008.DOI:10.3389/fpsyt. 2022.651008
[20] 曹凌云, 张萍淑, 元小冬, 等. 卒中后抑郁认知损害和睡眠障碍认知损害脑网络及ERP研究进展(综述)[J]. 中国健康心理学杂志, 2023, 31(10): 1479-1484. DOI:10.13342/j.cnki.cjhp.2023.10.008.
[21] Kumar H, Ganapathy N, Puthankattil SD, et al. Assessment of emotional states in EEG signals using multi-frequency power spectrum and functional connectivity patterns[J]. Annu Int Conf IEEE Eng Med Biol Soc, 2022, 2022: 280-283. DOI:10.1109/EMBC48229.2022.9871510
[22] 杜颖, 岑芷依, 彭欐涵, 等. 重复经颅磁刺激联合计算机辅助治疗对脑卒中后失语病人语言功能的影响研究[J]. 护理研究, 2021, 35(8): 1394-1397. DOI:10.12102/j.issn.1009- 6493.2021.08.013.
[23] 谭娟, 李晓敏, 苟晨, 等. 经颅直流电刺激对脑卒中后中枢性疼痛患者的镇痛效果研究[J]. 川北医学院学报, 2021, 36(6): 753-756. DOI:10.3969/j.issn.1005-3697. 2021.06.018.
[24] 杨珊, 王恺, 刘念, 等. 督脉针刺合舌三针联合康复训练对老年脑卒中失语症患者语言功能、认知水平及炎症因子的影响[J]. 中国老年学杂志, 2023, 43(19): 4630-4633. DOI:10.3969/j.issn.1005-9202.2023.19.007.
[25] 张延玲, 李国丽. QEEG中慢波与颞叶癫痫患者认知功能障碍相关性分析[J]. 中风与神经疾病杂志, 2023, 40(6): 522-525. DOI:10.19845/j.cnki.zfysjjbzz.2023.0123.