军事知识服务平台
基于神经电生理指标构建面神经损伤患者的预后预测模型
Prognostic prediction model for facial nerve injury patients based on neuroelectrophysiological parameters
摘要 目的 构建基于神经电生理指标的面神经损伤患者预后预测模型,为临床早期评估预后、制定个体化治疗方案提供客观参考依据.方法 回顾性选取 2022 年 1 月—2024 年 11 月大连理工大学附属中心医院/大连市中心医院收治的面神经损伤患者 146 例为研究对象,患者均接受标准治疗方案和眼部护理,治疗 6 个月后根据 House-Brackmann(HB)面神经功能分级系统的预后评估分为预后良好组 116 例和预后不良组 30 例.所有患者进行面神经电图(ENoG)、瞬目反射(BR)、表面肌电图(sEMG)检测,获取神经电生理核心指标面神经复合肌肉动作电位(CMAP)波幅及潜伏期、R1 波波幅及潜伏期、F 波潜伏期及出现率、静息态均方根值(RMS)、收缩态积分肌电值(iEMG);采用Pearson 或 Spearman 法分析神经电生理指标与基线 HB 分级Ⅳ~Ⅵ级、基线 Sunnybrook 评分的相关性;多因素 Logistic回归分析面神经损伤患者预后不良的影响因素;结合回归结果绘制列线图预测模型,通过 Bootstrap 自抽样法(抽样次数=1 000)进行内部验证,采用校准曲线、受试者工作特征(ROC)曲线、决策曲线(DCA)分析评估模型的区分度、校准度及临床实用性.结果 预后不良组 CMAP 潜伏期、R1 波潜伏期、F 波潜伏期长于预后良好组,静息态 RMS 高于预后良好组,CMAP 波幅、R1 波波幅、F 波出现率、收缩态 iEMG 低于预后良好组(t/P=13.539/<0.001、13.464/<0.001、4.659/<0.001、14.476/<0.001、12.431/<0.001、9.604/<0.001、14.931/<0.001、7.567/<0.001);面神经损伤患者 CMAP 波幅、R1波波幅、F 波出现率、收缩态 iEMG 与基线 HB 分级均呈显著负相关(rs/P=-0.829/<0.001、-0.728/<0.001、-0.860/<0.001、-0.675/<0.001),与基线 Sunnybrook 评分呈正相关(r/P=0.791/<0.001、0.696/<0.001、0.832/<0.001、0.643/<0.001);CMAP 潜伏期、R1 波潜伏期、F 波潜伏期、静息态 RMS 与基线 HB 分级呈正相关(rs/P=0.847/<0.001、0.842/<0.001、0.516/<0.001、0.812/<0.001),与基线 Sunnybrook 评分呈负相关(r/P=-0.815/<0.001、-0.808/<0.001、-0.478/<0.001、-0.776/<0.001);CMAP 潜伏期长、R1 潜伏期长、静息态 RMS 高是面神经损伤患者预后不良的独立危险因素[OR(95%CI)=1.421(1.115~1.813)、1.505(1.047~2.163)、1.088(1.018~1.164)];基于回归结果构建的列线图预测模型经验证,模型预测值与实际预后结局一致性良好(Hosmer-Lemeshow χ2=5.236,P=0.703),预测面神经损伤患者预后效能显著(曲线下面积为 0.856,敏感度 0.800,特异度 0.781),在广域范围内(0.1~0.6)模型临床净收益率均较高.结论 基于 CMAP 潜伏期、R1 潜伏期、静息态 RMS 构建的列线图预测模型,对面神经损伤患者预后具有良好的预测效能,且临床实用性强,可为临床医生早期识别预后不良高危人群、优化治疗方案提供科学依据.
Abstract:
Objective To construct a prognostic prediction model for patients with facial nerve injury based on neu-roelectrophysiological parameters,providing an objective reference for early clinical prognosis assessment and personalized treatment planning.Methods A total of 146 patients with facial nerve injury admitted to the Central Hospital Affiliated to Dalian University of Technology(Dalian Central Hospital)from January 2022 to November 2024 were enrolled.All patients received standardized treatment and eye care.Clinical characteristics were assessed before treatment.Electroneurography(ENoG),blink reflex(BR),and surface electromyography(sEMG)were performed to obtain key neuroelectrophysiological in-dicators,including compound muscle action potential(CMAP)amplitude and latency,R1 wave latency and amplitude,F-wave latency and occurrence rate,resting root mean square(RMS)value,and integrated electromyography(iEMG)during voluntary contraction.At six months after treatment,the House-Brackmann(HB)grading system was used to evaluate prognosis.Pearson or Spearman correlation analysis was applied to examine the correlations between neuroelectrophysiological indicators and baseline HB grade Ⅳ–Ⅵ as well as baseline Sunnybrook score.Multivariate logistic regression analysis was used to identify risk factors for poor prognosis.Based on the regression results,a nomogram prediction model was constructed.Inter-nal validation was performed using the bootstrap method(resampling times=1,000).The discrimination,calibration,and clin-ical utility of the model were assessed by calibration curve,receiver operating characteristic(ROC)curve,and decision curve analysis(DCA).Results At 6 months after onset,116 patients(79.45%)had a good prognosis and 30(20.55%)had a poor prognosis.Compared with the good prognosis group,the poor prognosis group had significantly longer CMAP latency,R1 wave latency,and F-wave latency,higher resting RMS,and significantly lower CMAP amplitude,R1 wave amplitude,F-wave occurrence rate,and iEMG during contraction(t/P=13.539/<0.001,13.464/<0.001,4.659/<0.001,14.476/<0.001,12.431/<0.001,9.604/<0.001,14.931/<0.001,7.567/<0.001).CMAP amplitude,R1 wave amplitude,F-wave occurrence rate,and iEMG during contraction were significantly negatively correlated with baseline HB grade(rs=-0.860 to-0.675,all P<0.001)and significantly positively correlated with baseline Sunnybrook score(r=0.643 to 0.832,all P<0.001).CMAP latency,R1 wave latency,F-wave latency,and resting RMS were significantly positively correlated with baseline HB grade(rs=0.516 to 0.847,all P<0.001)and significantly negatively correlated with baseline Sunnybrook score(r=-0.815 to-0.478,all P<0.001).Logis-tic regression analysis showed that CMAP latency,R1 wave latency,and resting RMS were independent risk factors for poor prognosis in patients with facial nerve injury(all P<0.05).Conclusion The nomogram prediction model based on CMAP latency,R1 wave latency,and resting RMS demonstrates excellent predictive performance and strong clinical utility for prog-nostic assessment in patients with facial nerve injury.It provides a scientific basis for early identification of high-risk patients and optimization of treatment strategies.
刊 名 疑难病杂志 ISTIC
年,卷(期): 2026, 25(10)
分类号: R745.1+2
机标分类号: R745.1 R246 R445.2
在线出版日期: 2026年10月29日