基于CV-SVM方法的手术时间估计

Surgery Time Duration Estimation Based on the CV-SVM Method

手术时间估计是进行科学手术排程的前提和依据,为了能够准确地估计手术时间从而为手术排程提供有效信息,采用交叉验证(CV)方法优化支持向量机(SVM)参数,构建基于交叉验证的支持向量机模型对手术时间进行估计。为了验证模型的性能,将CV-SVM模型与径向基(RBF)神经网络模型相对比,通过某医院眼科角膜移植手术时间估计进行实例验证。结果表明,相比RBF模型,基于CV-SVM模型的手术时间估计结果平均绝对百分误差在11%以内,相对误差在23%以内,验证了模型的有效性,为手术时间估计提供了一种有效的方法。

Surgery time duration estimation is the premise and basis of operation room scheduling. In order to estimate the surgery duration time accurately and provide effective information for operation room scheduling, by using the cross validation method to optimize the support vector machine parameter, a cross validation and support vector machine estimation model （CV-SVM） is built. In order to verify the performance of the model, the model is compared with the radial basis function （RBF） neural networks model, and a case of eye cornea transplant of a hospital is used to validate the model. Results show that, in comparison with the RBF model, the mean absolute percentage error of CV-SVM is below 11% and the relative error is below 23%, confirming the effectiveness of the model and providing an effective method to estimate surgery time duration.