目的：评估微血管减压术（microvascular decompression，MVD）治疗面肌痉挛（hemifacial spasm，HFS） 后迟发性面瘫（delayed facial palsy，DFP）的发生率，确定对DFP发生最重要的影响因素，并建立DFP发生的 预测模型。 方法：2015年9月至2020年6月，我院共收治HFS并行MVD术的HFS患者400例。出现DFP 48例（12%），收集患者的临床资料，应用LASSO-Logistic回归分析建立一个预测模型。使用C指数、ROC （受试者工作曲线）、calibration校正曲线来评估预测模型的鉴别、校准和预测性能。 结果：预测列线图中包 含的预测因素包括年龄、性别、有无面神经压迹、症状侧别、高血压和糖尿病。对模型进行内部验证发现，C 指数为 0.869（95%CI 0.805～0.933），说明模型具有良好的区分度；建模组和验证组的校准曲线与标准曲线 均极为接近，具有良好的校准度。结论：高血压和面神经压迹可能是MVD术后发生DFP的危险因素。基 于列线图的MVD术后DFP发生风险决策模型具有良好的区分度和准确度。

To investigate the incidence of delayed facial palsy (DFP) after microvascular decompression (MVD) for hemifacial spasm (HFS), identify the most important factors affecting the occurrence of DFP, and establish a prediction model for DFP occurrence. Methods: A total of 400 patients with HFS were admitted to our hospital and underwent MVD between September 2015 and June 2020. Among them, 48 patients (12% ) developed DFP. The clinical data of the patients were collected, and a prediction model was established by LASSO-Logistic regression analysis. The discrimination, calibration, and predictive performance of the prediction model were assessed using C index, receiver operating characteristic (ROC) curve, and calibration curve. Results: The predictive factors in the prediction nomogram included age, sex, presence of facial nerve pressure trace, symptom laterality, hypertension, and diabetes mellitus. Internal validation of the model revealed a C index of 0.869 (95% CI 0.805～0.933), indicating good discrimination of the model; the calibration curves of both the modeling and validation groups were extremely close to the standard curves, indicating good calibration performance. Conclusion: Hypertension and facial nerve pressure trace may be risk factors for the occurrence of DFP after MVD. The nomogram-based risk prediction model of DFP after MVD has good discrimination and accuracy.