بهبود برنامه ریزی توسط پزشکان اورژانس با استفاده از تجزیه و تحلیل داده کاوی
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|22136||2009||10 صفحه PDF||سفارش دهید||7287 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Expert Systems with Applications, Volume 36, Issue 2, Part 2, March 2009, Pages 3378–3387
Emergency departments are the first line in hospitals to face emergency patients. As a major function of emergency medicine, when a patient comes to the emergency department, the emergency medical personnel will first perform a triage procedure and then transfer the patient to associated departments for treatment. Due to the utilization pattern of the Taiwanese people in medicine, the emergency departments in most major hospitals are always overcrowded. The arrangement of manpower or the distribution of resources to handle patients’ demands can affect disease outcomes and quality of medical treatment. Therefore, the prediction of demands of physician manpower certainly will affect the quality and cost in medical treatment, and has significant impact on patients’ life and satisfaction. This study used data mining, classification and a decision tree to analyze the prediction model of patients’ demand in the Emergency department from real treatment situations. The result was the accuracy of shift anticipation improved from 22% to 50%. This study also used anticipant performance evaluation matrix integrated with loss function to evaluate the performance between the anticipation of demand established by mining and the original arrangement. It helped to save the cost of the medical personnel by 37%. In the end it combined the DMAIC action procedure from 6-Sigma and developed an anticipation model that can be suitable in different departments to dispatch medical personnel. It provided a reference of the decision maker of the hospital.
Along with the development of the Internet and the maturation of database techniques, the methods of data and information storage in different businesses and professions have become more diverse and simpler today. In order to combine the applications of the Internet, many people are promoting data digitalization. Large quantity data collection techniques, high performance multi-processor computer structure and the maturation of data mining mathematical algorithms are the three most important elements for the prosperous development of data mining today. They have been extensively used in different business and professions (Kdnuggets, 2007). Generally medical institutes only use general statistics skill instead of data mining techniques to properly utilize this information. We need to think how to collect useful information from this enormous database and find valued knowledge for medical or hospital management; how to utilize this information effectively to improve the efficiency of management, quality and costs in the hospital; understand patients’ demand for medical treatment and make a proper medical service strategy. But in fact, most of hospitals are not doing this at present. Currently the dispatch of emergency manpower is always made by each department from experiences and emergency triage. However, patients’ sickness distribution and variation in the patient numbers in peak and non-peak hours are not fully considered, so the emergency department might do jobs in a hasty and disorderly manner or might be over-crowded. While emergency medical procedures are heavy and complicated and emergency departments are always filled with patients, how can we establish a medical manpower deployment system to enhance medical quality and reduce medical disputes? (Academia Sinica, 1998 and Yang and Yang, 2004) The main purpose of manpower resources planning is to reduce uncertainties. By clarifying the environmental uncertainties and planning before it happens, we expect to reduce the impact (Dessler, 2006 and Richard, 1988). If medical manpower can be appropriately deployed, hospitals not only can provide medical care that is cost effective and also satisfying to the patient (Chou, 2003, LaMar et al., 1997 and Yeh et al., 2003). In the past most studies in physicians’ shift arrangement emphasized outpatient physicians’ shift arrangement (Liu & Wang, 2005); however characteristics of the demand on emergency physicians are a little bit different. There are two major differences between emergency services and outpatient services. The first is the urgency of physicians’ diagnosis. Generally patients in the emergency departments need urgent treatment in a very short time. Therefore, emergency departments have a triage system in accordance with patients’ condition (Tsan, 2006). The second major difference is that the emergency service is provided 24 h a day, while the outpatient physicians have demand-led shifts. In this research, hospitals divide outpatients into 6 departments, including internal medicine, surgery, traumatology, pediatrics, obstetrics and gynecology and dental departments. The rule of department assignment does not depend on patient numbers but because of the hospital’s policy and the payment system of the National Health Insurance. Also because of hospital organizational structure and demands from medical education, emergency physicians do not take all the shifts in the emergency department; generally the above mentioned six departments assign a resident physician to perform the first line services in the emergency department each month. Therefore, the emergency department is like an epitome of a small hospital. The study takes the largest department in the emergency department, the internal medicine department, as the research object and makes some suggestions to improve the arrangement of the physicians’ shift schedule. In this research, the objective of human resources utilization is that the numbers of physicians should be considered by patients’ requirement in order to promote the quality of emergency service and arrange proper number of personnel. It compares current internal medicine physicians’ shift arrangement in the emergency department and selects the variable. In the emergency medicine of the case hospital, demand of the patients’ side is to triage and provide emergency treatment to the patient in the first place. On the supply side it should include the allocation of personnel from different fields such as physicians, nursing staff, technicians, social workers, first aid technicians, administrative staff, janitors and volunteers. This study focuses on direct manpower supply of emergency physicians; accredited physicians from other departments and other indirect personnel are not considered. To conclude from the above, according to actual demands for physicians, it is planned to use triage technique as a preliminary analysis tool to analyze the emergency patient inspection and shift arrangement data. After developing a forecast model, a performance evaluation matrix combining Taguchi’s Loss Function is the final performance evaluation method. This procedure is also used to build a 6-Sigma’s DMAIC action procedure (George, Rowlands, Price, & Maxey, 2005) to achieve a continuing improvement administrative circulation for future references, shown in Fig. 1.
نتیجه گیری انگلیسی
First of all, this study uses the 6-Sigma SD action procedures to establish the research steps, then through DMAIC measure to define and clarify our problems, causes, sample drawing measurement standard and measures to standardize predictors. By using numbers of physicians needed, which is converted from number of patients in the case hospital in 2004, we forecast number of emergency physicians needed in 2005 for the case hospital through data mining decision tree classification. We also take realistic conditions into consideration and make appropriate solutions. Besides, we also construct the forecast performance evaluation matrix with ideas from Taguchi’s Loss Function to estimate the cost loss and also use the performance evaluation matrix to control and continuously improve the shift arrangement situation. By data mining decision tree classification, we not only forecast the physician demand in the case hospital, but also allocate emergency medicine manpower, obtain the decision tree model for physician demands and forecast, and present a new shift arrangement method. The accuracy rate is increased to 50% from the original 22%. Finally we use the forecast performance evaluation matrix to evaluate and quantify the performance of the data mining classification forecast model and the performance of the original shift arrangement; the total cost is reduced from the original’s NT$540,065 dollars to NT$341,880 dollars, showing a 37% significant decrease in cost loss from shift arrangement. The adopted data mining forecast model presents smaller values in both total estimated distance and total forecast costs. By combining the control line diagram ideas and the performance evaluation matrix, we focus on samples that are not in the range of control lines and obtain an understanding on each single sample. Therefore the forecast model, even if a few sample points are out of control line range, has better shift arrangement than the original shift arrangement, and actually obtains improvements. It also uses non-symmetrical Taguchi’s Loss Function to estimate and evaluate the possible cost loss. It successfully constructs a performance evaluation model to evaluate our forecast. Hopefully through the completion of this study, it cannot only solve the existing problem in this case hospital, but also through the action procedures in this study to introduce in a different medical treatment environment.