Li Guoliang, Feng Zhen, Zhang Huiyan, Zou Yongzhou, Xv Hong, Jiang Shunfu
Department of Neurology, Jingdezhen First People's Hospital, Jingdezhen, China.
Front Neurol. 2025 Aug 21;16:1580950. doi: 10.3389/fneur.2025.1580950. eCollection 2025.
Stroke recurrence is an important factor affecting the prognosis of mechanical thrombectomy in patients with middle cerebral artery (MCA) occlusion. This study aims to construct a model for evaluating the degree of stroke recurrence and conduct binary and ternary interaction analysis.
We conducted a retrospective analysis of the clinical data of stroke recurrence patients, collecting demographic data, clinical characteristics, treatment factors, and biochemical indicators. Use XGBoost and RF models to screen features that contribute significantly to the degree of recurrence, and evaluate model performance through indicators such as ROC curve, F1 score, accuracy, and recall. Construct a stroke recurrence evaluation model based on the common features selected from these two models. Use the Andersson model to analyze the binary interaction between the model and other factors. Further analyze the three-way interaction between the model and other factors.
Both XGBoost and RF models perform well. In the multivariate logistic regression analysis, the recurrence model showed that age, smoking history, and infarct size had a significant impact on the degree of stroke recurrence (OR = 1.006, 1.214, 1.167, all < 0.05), and the constructed recurrence model had a significant effect on the degree of stroke recurrence (OR = 1.346, = 0.047). Through binary interaction analysis, it was found that there was a significant antagonistic effect between the recurrence model and age, smoking history, and infarct size. Triple interaction analysis showed that the synergistic effect of the recurrence model with age and smoking history was significant, and the synergistic effect of the recurrence model with smoking history and infarct size was also significant.
Age, smoking history, and infarct size are important influencing factors on the degree of stroke recurrence in MCA occlusion patients after mechanical thrombectomy treatment. The recurrence model performs differently in different patient populations, and the interaction with age, smoking history, and infarct size is of great significance for evaluating the degree of stroke recurrence.
卒中复发是影响大脑中动脉(MCA)闭塞患者机械取栓预后的重要因素。本研究旨在构建一个评估卒中复发程度的模型,并进行二元和三元交互分析。
我们对卒中复发患者的临床资料进行回顾性分析,收集人口统计学数据、临床特征、治疗因素和生化指标。使用XGBoost和随机森林(RF)模型筛选对复发程度有显著贡献的特征,并通过ROC曲线、F1分数、准确率和召回率等指标评估模型性能。基于从这两个模型中选择的共同特征构建卒中复发评估模型。使用安德森模型分析该模型与其他因素之间的二元交互作用。进一步分析该模型与其他因素之间的三元交互作用。
XGBoost和RF模型均表现良好。在多因素逻辑回归分析中,复发模型显示年龄、吸烟史和梗死灶大小对卒中复发程度有显著影响(OR = 1.006、1.214、1.167,均<0.05),且构建的复发模型对卒中复发程度有显著影响(OR = 1.346,P = 0.047)。通过二元交互分析发现,复发模型与年龄、吸烟史和梗死灶大小之间存在显著的拮抗作用。三元交互分析表明,复发模型与年龄和吸烟史的协同作用显著,复发模型与吸烟史和梗死灶大小的协同作用也显著。
年龄、吸烟史和梗死灶大小是MCA闭塞患者机械取栓治疗后卒中复发程度的重要影响因素。复发模型在不同患者群体中的表现不同,其与年龄、吸烟史和梗死灶大小的交互作用对评估卒中复发程度具有重要意义。