Prognosis Analysis and Data Mining for Emergency Patients after Arriving Hospital

With the increasing of the population of elderly people in recent years, the need of emergency medical services is also growing. Emergency medical service (EMS) is the first defense line when people encounter diseases and injuries. The purpose of EMS is to reduce patients' secondary damage in patients and increase the survival rate or have better prognosis. The EMS procedures must be closely and appropriately integrated in the courses at the accident place, on route to the emergency department, and the medical care after arriving emergency department. In the past four years, we committed to the study of pre-hospital emergency medical decision Model (NSC 99-2221-E-038-009-MY2 01-2221-E-038-005- and 102-2221-E-038 -002 -). We have established an emergency decision-making model by using 11 factors of Emergency Medical Services (EMS). The AHP (Analytic Hierarchy Process, AHP) was used to calculate the weighting of the factors to complete the model’s objective function. The model can be used as a reference when make a decision to transport the emergency patients. In addition, we have also successfully connected the emergency medical services database from Taipei Fire Department to the Health Insurance database from the Department of Health on the bases of individual patient. We have used the integrated database to do some statistical analyses to know the prognosis of the emergency patients. Prehospital Emergency Medical Services and the quality of medical care after arriving emergency room are the two most important services for the patients when the patients get injured, encountered diseases and other emergency situations. The life support before patients arrive hospital and the treatment after arrive hospital must be in close combine, to enhance the prognosis of patients effectively. Those two sets of data can’t be linked because of the data are owned by different government organizations, fire department have the data of emergency medical services, and the medical record is owned by different hospitals. These two organizations didn’t share the data to each other before. At present, only a few cases (include: AMI, stroke for example), the hospital is required to transfer back patient’s situation to the fire department. Therefore, it is the first time that the emergency medical services database and the Health Insurance database are linked successfully under a very careful data security examination. Following our previous results, in this two-year project, we will use the integrated database to do some data mining and rules regulation. First of all, we will continuously analyze the related features of the decision model for pre-hospital emergency care using current emergency medical service data and geographic model. Secondly, we will use variety of algorithms such as association rule, decision tree, and artificial neural network to analyze the relationships between patient’s situation and the prognosis after they are sent to hospitals with different degrees. Rules are expected to be generated to explain which kind of patients should be sent to which level of hospitals. We expect the results of this project will give a good reference for EMT training and policy decision-making for Taipei City Fire Department and Ministry of Health and Welfare.