| Summary: | 碩士 === 國立成功大學 === 醫學資訊研究所 === 98 === In recent years, chronic diseases accounts for the 60% proportion in cause of death, it going to be up to 73% dead because of chronic diseases in 2020. According to literatures, obesity increase the morbidity of chronic diseases and Regular exercise reduce the incidence of chronic disease.
The purpose of this research is aimed to use single 3 axis accelerometer which embedded in smart phone to develop an activity recognition system. It can detected and take down the 8 types of human daily activity and their duration. Besides, the user will be rewarded the record as a diet basis. It can be a self-exercise manager platform, moreover, can be used by medical organization. Through this system, doctor can provide user appropriate exercise suggestion which will help user to form exercise habit, and lower the morbidity of chronic disease.
On methodology, the research transformed raw data to feature vector by information gain, developed a tree augmented na?ve bayes classifier in recognition system.
Finally, this research evaluated the performance of the classifier in practical experiment. Our result have successfully validated: 1) the effectiveness of feature selection 2) recognition capability of the classifier and achieve satisfactory performance for human activity recognition.
|
|---|
