脑损伤所致意识障碍患者1年后结局预测列线图的构建：一项回顾性多中心研究

关键词：意识障碍  预测  列线图  脑损伤

Keywords:disorders of consciousness  prediction  nomogram  brain injury

摘要：

      摘要 目的：脑损伤所致意识障碍患者1年后结局预测列线图很少，本研究旨在利用多中心回顾性研究队列建立脑损伤所致意识障碍患者1年后结局预测列线图。 方法：回顾性收集南昌大学第一附属医院和上饶市中医院2018年1月1日至2019年4月1日期间患者的临床资料。对缺失的数据进行了多重插补。数据降维、预测变量选择采用最小绝对收缩和选择算子（the least absolute shrinkage and selection operator, LASSO)回归模型，采用多元Logistic回归分析建立预测模型。纳入了选定的风险因素，并用列线图表示。从区分度与校准度对列线图的性能进行了评估。 结果：共95例符合纳入和排除标准的患者纳入本研究，在LASSO回归模型中从17个潜在预测变量中选择了4个非零系数的预测变量，格拉斯哥昏迷量表评分(Glasgow coma scale，GCS)评分高、白蛋白水平正常、时间短、凝血酶时间（thrombin time，TT）水平高，1年后良好结局可能性较高。该模型具有良好的区分性，曲线下面积(area under the curve，AUC)为0.71(95%CI：0.63—0.77)，表明该模型具有较好的区分度和校准度。 结论：本文构建的意识障碍患者1年后结局预测列线图在内部评估表现出良好的区分度与校准度，但将来需外部验证来进一步证明模型的有效性。

Abstract:

      Abstract Objective：Few nomograms have been previously developed to predict the outcome of patients with disorders of consciousness caused by brain injury after one year. This study aimed to develop a nomogram for predicting the outcome of patients with disorders of consciousness caused by brain injury after one year using multiple-center study cohort. Method：The clinical data in the first affiliated hospital of Nanchang university and Shangrao traditional Chinese medicine hospital from January 1， 2018 to April 1， 2019 were collected retrospectively. We used multiple imputation to deal with the missing data. Lasso regression model was used for data dimension reduction and predictive factors selection. Multivariable logistic regression analysis was used to develop the predicting model. We incorporated selected risk factors and presented them with a nomogram. The performance of the nomogram was assessed by calibration and discrimination. Result：A total of 95 patients who met the inclusion and exclusion criteria were enrolled in this study. 4 out of 17 potential predictors with non-zero coefficients were selected in the LASSO logistic regression model. It is more likely to have a good outcome after 1 year with high GSC score， normal albumin level， short time and high thrombintime level. The model showed good discrimination and calibration， with AUC of 0.71 (95%CI：0.63—0.77). Conclusion： The nomogram developed in this study for predicting the outcome of patients with disorders of consciousness caused by brain injury after one year shows good differentiation and calibration in the internal evaluation， but external verification is still needed in the future study.