The Comparison of applications of Logistic Regression and Random Forests for developing triaging Acute Coronary Syndromes (ACS) Patients Presented by the Ambulance and themselves / 比較邏輯斯回歸和隨機森林建構判斷透過救護車以及自行到院的急性冠狀動脈症候群(ACS)患者之模型成效

碩士 / 元智大學 / 工業工程與管理學系 / 105 / Background: Acute Coronary Syndrome(ACS)is the most acute and severe disease in the all coronary artery diseases, the shorten treatment time is a key to reduce the incidence of mortality. When the patients with suspected ACS who needed to be transferred to the hospital in a great urgent situation via an ambulance, it is possible to reduce the implementation of the relevant detection process time if the patients were effectively and promptly to be identified as a ACS before arrival to the hospital, and thus to achieve the shorten D2B time. Purpose: This study aimed at developing a judge tools for the patients who was transferred by ambulance or walk-in to the hospital in the effectively triaging possible suspected ACS. Methods: 1,125 clinical cases were collected from the Far Eastern Memorial Hospital between February 2016 and February 2017, were used to develop and test the ACS judge tools model by using cluster analysis, stepwise logistic regression, decision tree and random Forests. Results: The ACS judge tool models can separated into two ways, one of the model is triaging the patients who was transferred to the hospital by ambulance, the another one is triaging the patients who arrival to the hospital by themselves. The first model called ambulance ACS model based on shock, sex and radiation pain, the second model called walk-in ACS model based on shock, sex, hypertension and chest pain. Compare to the Pei-Li Chung (2016) ACS model had better performance (Patients through the ambulance: discriminability index d’ = 0.5, compare to 0.39; Patients through the walk-in: discriminability index d’ = 0.61, compare to 0.43). Conclusion: This study suggested that the contents of the medical records should be more comprehensive and the scope of the data should be expanded to increase the effective and authenticity of the judgment model. Furthermore, the ACS judge model can be implemented to rapid and effective analysis the suspected ACS patients on the ambulance to reduce ultimate D2B times.

Identiferoai:union.ndltd.org:TW/105YZU05031031
Date January 2017
CreatorsShih-Yu Chung, 鍾世佑
ContributorsJui-Feng Li, 林瑞豐
Source SetsNational Digital Library of Theses and Dissertations in Taiwan
Languagezh-TW
Detected LanguageEnglish
Type學位論文 ; thesis
Format125

Page generated in 0.0105 seconds