本文已被：浏览 3842次 下载 0次
中文摘要: 目的 考虑患者诊疗过程，对急性阑尾炎手术术后住院天数建立预测模型，为医院提高服务效率、医保政策制定以及患者提前了解治疗情况提供参考。方法 随机抽取某三甲医院2016年7月—2017年9月的腹腔镜下急性阑尾切除术患者共243例，利用单因素分析从患者年龄、性别、BMI、是否有合并症和是否有穿孔等因素找出显著影响因素，之后分别利用概率神经网络和多元logistic回归模型的方法构建阑尾炎患者术后住院天数预测模型。结果 影响急性阑尾炎患者术后住院天数的因素主要包括患者年龄、阑尾炎穿孔情况、其他并发症情况、白细胞指数情况、切口恢复天数以及术后发烧天数，概率神经网络模型对测试集的预测精度为72.1%，高于多元logistic回归模型的64.0%。结论 概率神经网络模型拟合优度优于多元logistic回归模型，能够有效预测腹腔镜下急性阑尾切除术患者术后住院天数。
中文关键词: 急性阑尾炎；概率神经网络 术后住院天数 预测模型
Abstract:OBJECTIVE For providing guidance for the efficient management of hospital sickbeds, the reasonable formulation of medical insurance policy and knowing the treatment state for patients, it was important to predict the postoperative length of stay of patients with acute appendicitis. METHODS A total of 243 cases of laparoscopic appendectomy randomly selected in a tertiary hospital between July, 2016 and September, 2017 were reviewed retrospectively. Among the factors of patients’ age, sex, BMI, whether have the complication and whether have the perforation, single factor was applied to find out the significant influencing factors, and to build the prediction model for the patients’ length-of-stay after laparoscopic appendectomy by using probabilistic neural networks and polychotomous logistic regression. RESULTS The results showed that 6 variables, including age, perforation status, other complications, hemoglobin index and the length of fever, were related with the postoperative length of stay. The accuracy of probabilistic neural networks model was 72.1%, which was 64.0% higher than the polychotomous logistic regression model. CONCLUSION Compared with the polychotomous logistic regression model, the proposed model was a better prediction approach for postoperative hospital stay of patients with acute appendicitis.
keywords: acute appendicitis probabilistic neural networks postoperative length of stay prediction model
文章编号：3201907003 中图分类号：R197 文献标志码：
基金项目:
| 作者 | 单位 |
| 梁丽军①#，霍梅亚②，董方岐③ |
| Author Name | Affiliation |
引用文本：