The use of physical examination and lifestyle data to establish the prediction model of chronic disease

• Main Authors: YANG,HUI-FEI, 楊惠斐
• Other Authors: CHANG,I-CHIU
• Format: Others
• Language: zh-TW
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/32029284457231811289

碩士 === 國立中正大學 === 資訊管理系醫療資訊管理研究所 === 104 === According to the Ministry of Health and Welfare (MOHW), Executive Yuan, R.O.C. (Taiwan), Taiwan's top ten causes of death in 2014 is chronic diseases.So we explore the main factor,we found that hypertension, high cholesterol and diabetes are the main fators,and causing 219.2 people's mortality per 100,000 population, We can’t ignore chronic diseases have become the number one killer of harm to people's health, to take preventive measures, the application of data mining techniques from physical examination data to explore signs of suffering from chronic diseases before, so that we can prevent or delay resulting in chronic diseases. In this study, we use the physical examination data of a region in south-central teaching hospital as a data source.By trying many kind of mining techniques, we proposed a novel framework for discovering health risk patterns and the relation between the health patterns and the target disease from physical examination history data. Results in a random forest (random forest) classification the best performance prediction model constructed, AUC: 76.60%, the best research-based classifiers, In the other hand,we found age and a history of diabetes have a high degree of correlation between suffering from hypertension, high cholesterol, diabetes. The information can build effective prediction model for target disease diagnoses. According to these information provided, the physicians could early provide the health alerting and the medical treatment for people.