（2016/8/9）
土木系陳柏華教授之研究團隊與台大醫學院合作，
論文刊登於《IEEE J-BHI》期刊封面

刊登期刊：IEEE Journal of Biomedical and Health Informatics

論文名稱：Demand Forecast Using Data Analytics for the Preallocation of Ambulances

本研究與台大醫學院緊急醫療研究中心及台大急診部教授共同合作，利用實際數據研究緊急救護需求之時空變動。將具緊急救護需求的病人，及時送達適切的醫院，是緊急醫療救護服務運作的重要目標。本研究探討緊急救護案件之需求預測分析方法，並以新北市為研究案例，對分析方法進行驗證。本研究之步驟包含了時間與空間之離散化、模式校估與選擇、及驗證。利用新北市三年約50萬筆到院前緊急醫療之案件資料，對於各離散化之空間進行時間序列之分析。所使用之模式包含支持向量回歸、類神經網路、以及週期回歸。此預測模式可用以提供救護車調度及派遣決策之參考，提升到院前緊急醫療之反應時間。

Prof. Albert Y. Chen’s research team, in collaboration with NTU Medical School, has published an article and featured as the cover article of IEEE Journal of Biomedical and Health Informatics.

Title of article: Demand Forecast Using Data Analytics for the Preallocation of Ambulances

IEEE Journal of Biomedical and Health Informatics

By increasing the operational efficiency of pre-hospital Emergency Medical Service (EMS), the survival rate of patients could potentially be increased. The Geographic Information System (GIS) is introduced in this work to manage and visualize the spatial distribution of demand data and forecasting results. A flexible model is implemented in GIS, through which training data are prepared with user desired sizes for the spatial grid and discretized temporal steps. We applied models such as Moving Average, Artificial Neural Network, Sinusoidal Regression, and Support Vector Regression for the demand forecast of pre-hospital emergency medicine. The results from these approaches, as a reference, could be used for the pre-allocation of ambulances. A case study is conducted for the EMS in New Taipei City, where pre-hospital EMS data has been collected for 3 years. The model selection process has chosen different models with different input features for the forecast of different areas. The best daily mean absolute percentage error during testing of the EMS demand forecast is 23.01%, which is a reasonable forecast based on Lewis’ definition. With the acceptable prediction performance, the proposed approach has its potential to be applied to the current practice.