Analysis of Abnormality Diagnosis in Emergency Medicine by Integrating K-means and Decision Trees-a Case Study of Dongyang People's Hospital in China

The performance of a triage system can facilitate patient classification in an emergency department, enabling patients in critical condition to receive better medical care; therefore, more perfect allocation and use of resources of emergency medical treatment are required. The correctness of nurses and doctors is related to triage medical care quality, patient satisfaction, and life safety. Hence, how to effectively extract experience by data mining and triage in the background of continuously increasing numbers of emergency patients is an issue worth exploring. Based on the case of Dongyang People’s Hospital in China, this study established a triage prediction model from process construction, parameter selection, and sampling, and randomly generated 501 samples of patients from the emergency database for cluster analysis (Ward’s method and K-means) and decision trees analysis upon data mining. The findings of this study show that the triage categorization of nurses is higher than that of doctors and most abnormal diagnoses occur to patients not examined on the date of admittance. The vital signs of pulse and temperature are more discerning. According to the confidence and support proportion, this study proposed seven association rules.