【摘要】 目的:发作性前庭综合征(episodic vestibular syndrome, EVS)临床诊断复杂，部分患者存在顽固性复发情况，本研究尝试使用简单的临床因素构建复发的风险预测模型，以利基层医疗机构诊疗及医疗资源分配。方法:本研究基于2017年1月至2023年12月医院眩晕门诊非良性阵发性位置性眩晕(BPPV)的EVS患者登记数据库，根据复发情况将患者分为顽固组(反复发作≥3次或病程>1年且治疗无效)和对照组(复发<3次或病程<1年)。通过病例对照研究设计，采用Lasso回归筛选预测因子，结合logistic回归构建复发风险预测模型并以列线图可视化，同时进行内部验证。结果:共纳入1 710例患者，随机分为训练集(1 197例)和测试集(513例),最终筛选出9个预测因子(失衡感、偏头痛、高脂血症、发作预感、颈肩痛、家族史、视物旋转、自身旋转感、漂浮感)构建模型。模型在训练集和测试集中的C-index分别为0.810(95%CI 0.786～0.835)和0.833(95%CI 0.796～0.871),校准曲线显示预测概率与实际概率一致性良好。结论:该模型简单易行，可助基层医疗机构在EVS的诊疗中快速识别高复发风险患者，优化转诊策略并节约医疗资源。更多还原

【Abstract】 Objective: The clinical diagnosis of episodic vestibular syndrome(EVS) is complex, and some patients have persistent recurrent conditions. This study attempts to construct a risk prediction model for recurrence using simple clinical factors to facilitate diagnosis and treatment in primary medical institutions and the allocation of medical resources. Methods: This study was based on a database of patients diagnosed with EVS, excluding benign paroxysmal positional vertigo(BPPV), at the hospital’s vertigo clinic between January 2017 and December 2023. Patients were categorized into a refractory group(recurrent attacks ≥ 3 times or disease duration > 1 year with ineffective treatment) and a control group(recurrence < 3 times or disease duration < 1 year) according to their recurrence status. A case-control study design was employed, and Lasso regression was applied to screen for predictive factors. A logistic regression model was subsequently constructed to predict recurrence risk and visualized using a nomogram, followed by internal validation. Results: A total of 1 710 patients were included and randomly divided into a training set(1 197 cases) and a test set(513 cases). Nine predictive factors-imbalance, migraine, hyperlipidemia, premonitory symptoms, neck and shoulder pain, family history, visual rotation, self-rotation sensation, and floating sensation-were ultimately selected for model construction. The C-indexes of the training set and test set were 0.810(95%CI 0.786-0.835) and 0.833(95%CI 0.796-0.871), respectively. The calibration curve showed good consistency between predicted probabilities and actual outcomes.Conclusion: This simple and practical model may assist primary care institutions in rapidly identifying EVS patients at high risk of recurrence, optimizing referral strategies, and conserving medical resources.更多还原