基于Logistic回归对手术分级方法的探索性研究

Exploratory Research Based on Logistic Regression Method of Operation Classification

目的探索通过某些客观指标综合分析实现手术分级的方法。方法选取某院2017年4月1日-2017年9月30日有手术操作共计10 474例的病案首页数据。基于手术级别的影响因素,选取例如手术时间、主刀医师职称、疾病系统等指标与手术级别进行Logistic回归分析,探索手术级别与客观指标的回归模型,从而实现手术分级。结果手术相关数据品质、客观指标删选的专业程度对模型建立影响较大,致使本次研究仅实现了手术级别的无序多分类Logistic回归,其对手术级别预测准确度达到67%。手术持续时间较短、主刀医师职称较低或总费用较少的手术,其手术级别为一级手术可能性较大。手术持续时间越长、主刀医师职称越高或费用越大的手术级别越大。结论手术持续时间、主刀医师职称与费用类指标分别反应手术的难度和消耗资源情况,是确定手术级别的重要因素,完善相关指标提升数据品质后,运用客观指标对手术进行综合分类的方法具有一定的可行性,此方法结合计算机网络技术,能较好的实现手术分级的及时性和准确性。

Objective To explore the method of operation classification by comprehensive analysis of some objective indicators. Methods This paper set about from the influence factors of the operation level, select such as operation time, leading physicians titles, such as disease system logistic regression analysis with age., so as torealize the operation classification. Results Because of the poor quality of medical record data, causes the study only to realize the operation level of disordered multiple Logistic regression and the accuracy of operation level prediction was 67%.Shorter duration of surgery, leading physicians titles is low or the total cost less operation, its level of minor operation more likely. The longer the duration of surgery, the higher the surgeon＇s title, or the higher the cost, the higher the operation level. Conclusions It was concluded that the method of comprehensive classification of operation by objective index was feasible in improving the relevant index system. And this method combined with computer net work technology could better realize the timeliness and accuracy of operation classification.