目的 探讨重型颅脑损伤患者并发急性胃肠损伤的危险因素，为预防急性胃肠损伤提供借鉴。方法 2021年1月至2023年1月，便利抽样法选取某院收治的重型颅脑损伤患者150例为研究对象，建立基于重型颅脑损伤并发急性胃肠损伤的危险因素的随机森林算法的预测模型。结果 150例重症颅脑损伤患者中，并发急性胃肠损伤患者94例，占62。67％。是否并发急性胃肠道损伤的患者在糖尿病、白蛋白、APACHE-Ⅱ评分、休克指数、液体负平衡、酸中毒、深度镇静、呼吸衰竭方面的差异均有统计学意义(均P＜0。05)。构建重型颅脑损伤并发急性胃肠损伤的随机森林模型，树的数量为103时出现的错误率最低;影响重型颅脑损伤并发急性胃肠损伤的因素重要性排序为糖尿病、液体负平衡、急性生理与慢性健康评分、白蛋白、深度镇静及酸中毒;随机森林模型预测重型颅脑损伤并发急性胃肠损伤的受试者工作特征曲线(receiver operating characteristic，ROC)下面积(area under curve，AUC)为0。798，Logistic回归模型的AUC为0。773。结论 构建的重型颅脑损伤并发急性胃肠损伤的风险预测模型预测效能较高，临床值得推广应用。
Review on Acute Gastrointestinal Injury in Patients with Severe Craniocerebral Injury Based on Random Forest Algorithm and Risk Model Construction
Objective To explore the risk factors of acute gastrointestinal injury in patients with severe craniocere-bral injury,and to provide reference for clinical prevention of acute gastrointestinal injury.Methods From January 2021 to January 2023,150 patients with severe craniocerebral injury admitted to a hospital were selected by conven-ience sampling method.A prediction model based on the risk factors of severe craniocerebral injury complicated with acute gastrointestinal injury was constructed by random forest algorithm Results Among 150 patients with severe craniocerebral injury,94 cases(62.67％)were complicated with acute gastrointestinal injury.There were statistically significant differences in diabetes mellitus,albumin,APACHE-Ⅱ score,shock index,negative fluid balance,acidosis,deep sedation and respiratory failure between patients with or without acute gastrointestinal injury(all P＜0.05).The random forest model for severe craniocerebral injury complicated with gastrointestinal injury was constructed,and the lowest error rate occurred when the number of trees was 103;the factors affecting severe craniocerebral injury compli-cated with acute gastrointestinal injury in the order of importance were diabetes mellitus,negative fluid balance,APACHE-Ⅱ score,albumin,deep sedation,and acidosis;the AUC of ROC curve for prediction of severe craniocere-bral injury complicated with acute gastrointestinal injury by random forest model was 0.798,and that of logistic regression model was 0.773.Conclusions The established risk prediction model for severe craniocerebral injury compli-cated with acute gastrointestinal injury has high predictive efficiency and is worth popularizing.
random forest algorithmsevere craniocerebral injuryacute gastrointestinal injuryrisk model
万方数据
维普
