Abstract:

Objective To construct a risk predictive model of intraoperative hypothermia for patients undergoing laparoscopic surgery and to verify the predictive effect of the model. Methods 1043 patients who underwent laparoscopic surgery and met the inclusion and exclusion criteria were selected in our hospital from June to October 2020,using the convenience sampling method. They were randomly assigned to a modeling group and a verification group at a ratio of 7 ∶ 3. The influencing factors of patients with intraoperative hypothermia（n=407） and patients without intraoperative hypothermia（n=323） in the modeling group were compared,which is conducive to the random forest algorithm to sort the influencing factors and build the prediction model. Results The incidence of intraoperative hypothermia was 55.75% in the modeling group and 54.95% in the validation group. In the importance score of random forest algorithm variables,basic body temperature,operating room temperature,BMI,operation time and other indicators have a high contribution to the model classification,with clinical significance. The area under the receiver operating characteristic curve of the predictive model is 0.797;the sensitivity is 78.74%;the specificity is 64.03%;the accuracy is 72.20%. Conclusion The prediction model based on random forest algorithm is effective,which is of great significance to identify the key factors of intraoperative hypothermia in patients undergoing laparoscopic surgery and intervene timely and effectively.

Key words: Intraoperative Hypothermia, Correlative Factor, Risk Prediction, Random Forest Algorithm, Nursing Care

摘要：

目的 构建腹腔镜手术患者术中低体温风险预测模型并验证模型的预测效果。方法 采用便利抽样法,选取2020年6月—10月在山东省某三级甲等医院接受腹腔镜手术并符合纳入标准的患者1 043例,按照7 ∶ 3的比例随机分配至建模组和验证组。将建模组发生术中低体温患者（407例）和未发生术中低体温患者（323例）的各项影响因素进行对比,利用随机森林算法对各项影响因素进行排序并构建预测模型。利用验证组数据绘制受试者操作特征曲线,并以曲线下面积检验模型的预测效果。结果 患者术中低体温发生率建模组为55.75%,验证组为54.95%。随机森林算法变量重要性评分中,基础体温、手术间温度、BMI、手术时长等指标对模型分类的贡献度较高,且具有临床意义。预测模型的受试者操作特征曲线下面积为0.797,灵敏度为78.74%,特异度为64.03%,准确率为72.20%。结论 该研究基于随机森林算法构建的预测模型具有良好的预测效能,对识别腹腔镜手术患者术中低体温的关键因素具有重要意义。

关键词: 术中低体温, 相关因素, 风险预测, 随机森林算法, 护理