新疆医科大学学报

目的 探讨贝尔面瘫治疗效果的影响因素，利用数据构建列线图预测模型并进行验证。方法 选择2020年1月-2023年12月在北京中医药大学东方医院就诊的贝尔面瘫323例患者作为研究对象。按照7∶3的比例将患者随机分为建模组(n=226)以及验证组(n=97)。对建模人群进行单因素及多因素Logistic回归分析，筛选出影响疗效的因素并建立列线图预测模型。分别对两组人群预测模型采用受试者工作特征(Receiver operating characteristic,ROC)曲线下面积、GiViTI校准曲线及Hosmer-Lemeshow检验、决策分析曲线进行区分度、校准度、临床有效性的验证。结果 建模组226名贝尔面瘫患者中，治愈125例(55.31%),未治愈101例(44.69%)。单因素分析结果显示，治疗效果不同患者的年龄、就诊时的House-Brackman等级、激素治疗时间、针灸治疗时机、针灸治疗方案的数据差异有统计学意义(P<0.05)。Logistic多因素分析结果显示，年龄、就诊时的House-Brackman等级、激素治疗时间、针灸治疗时机、针灸治疗方案是贝尔面瘫经针灸治疗后效果的独立影响因素(P<0.05)。建模组和验证组ROC曲线下面积分别为0.977、0.981(P均>0.05),提示模型区分度较好；采用Bootstrap自抽样法，重复自抽样1 000次，获得建模组及验证组的校准曲线，两组校准曲线观测值和预测值的重合度较好，提示预测值与实际值具有良好一致性；经Hosmer-Lemeshow检验，建模组(χ²=7.190,P=0.636)与验证组(χ²=6.236,P=0.801)比较差异无统计学意义，提示模型校准度较高；建模组及验证组的决策分析曲线均显示净收益较高，提示模型具有临床有效性。结论 贝尔面瘫患者经过治疗后仍然存在无法痊愈的可能，患者的年龄、就诊时的House-Brackman等级、激素治疗时间、针灸治疗时机、针灸治疗方案是贝尔面瘫治疗后效果的独立影响因素，基于此创建的列线图预测模型能够准确预测患者的治疗效果。

Objective To explore the factors influencing the effect of Bell facial palsy after the treatment, and to construct a clinical prediction model with data and implement validation. Methods From January 2020 to December 2023, 323 patients with Bell facial palsy who were treated in the hospital were selected as study objects. All patients were randomly assigned to modeling group(n=226) and validation group(n=97) in a 7∶3 ratio. To described the effect distribution of Bell facial paralysis after acupuncture treatment, Univariate and multivariate Logistic regression analysis was performed on the model population to screen out the factors affecting the curative effect and establish a nomogram prediction model. The area under receiver operating characteristic(ROC) curve, GiViTI calibration curve, HosmerLemeshow test and decision analysis curve were used to verify the differentiation, calibration and clinical effectiveness of the 2 groups of population prediction models. Results 226 patients with Bell′s facial palsy in the modeling group, 125(55.31%) were cured and 101(44.69%) were not cured. Univariate analysis showed that there were significant differences in age, House-Brackman grade, hormone therapy time, acupuncture treatment time and acupuncture treatment plan among the different patients(P<0.05). Logistic multivariate analysis showed that age, House-Brackman grade at the time of the treatment, hormone treatment time, acupuncture treatment time and acupuncture treatment plan were independent influencing factors for the effect of acupuncture treatment for Bell facial palsy(P<0.05). It had been proved that the area under the ROC curve in the modeling and validation groups was 0.977 and 0.981(both P>0.05), respectively, indicating that the model was better differentiated. The Bootstrap self-sampling method was used for model verification, and the self-sampling was repeated 1 000 times to obtain the calibration curve of the modeling group and the validation group. The coincidence degree of the observed value and the predicted value of the 2 calibration curves was good, indicating a good agreement between the predicted value and the actual value. Hosmer-Lemeshow test on modeling group and validation group: modeling group(χ²=7.190, P=0.636), validation group(χ²=6.236, P=0.801), there was no statistically significant difference, indicating high calibration of the model. The decision analysis curve of both modeling group and validation group showed high net benefit, suggesting that the model had clinical validity. Conclusion Patients with Bell′s facial palsy still have the possibility of not being cured after the treatment. Patients′ age, House-Brackman grade at the time of the treatment, hormone treatment time, acupuncture treatment time and acupuncture treatment plan are independent influencing factors for the treatment effect of Bell′s facial palsy, and the clinical prediction model based on this model can accurately predict the treatment effect of the patients.