醫院手術室是醫院裡相當重要的資產，並且手術室的時間管理，對於病患滿意度、員工滿意度或經濟上的指標有重大的影響力。而在相關研究發現對於使用機器學習方法進行手術室管理具有許多潛力，特別是手術執行時間的預測上。故本研究致力於改善手術執行時間預測之方法研究。 本研究運用支持向量迴歸，以台灣某醫學中心之手術資料，建構各科別手術執行時間的預測模型，結果發現在多數外科，麻醉因素有極高的重要性，而病患是否住院，可以作為所有科別共同的手術執行時間預測因子。另外，在模型比較上，無調整參數的支持向量迴歸模型比多元線性迴歸模型，適用於更多科別，特別是平均手術時間高且手術時間標準差大的科別，其平均誤差時間縮短了3分鐘。據此研究結果，院方可運用此模型當作醫務管理中手術排程的參考或是醫療過程手術室時間管理的基礎。 ;For a hospital, the operating room is a very crucial asset. Time management of the operating room especially has a significant influence on patient satisfaction, employee satisfaction, or economic indicators. Previous researches have indicated that using machine learning methods to improve the management of the operating room has great potentials, especially in the prediction of surgical case duration. So, this research aims at improving the methods of the prediction of surgical case duration. Support vector regression is used in the research to construct a surgical case duration prediction model based on the surgical data from various departments of a medical center in Taiwan. The results show that anesthesia factors is extremely important for most surgical operations, and hospitalization is one of the common predictors to all departments. Besides, in model comparison, the support vector regression model without adjustment of parameters is more suitable than the multiple linear regression model for most departments; especially for those with high average case duration and large standard deviation of case duration; Based on the research, the average error duration is shortened by 3 minutes. According to the results of the research, this model can be applied to surgical scheduling of medical management, or used as the basis for the time management of operating rooms.
