Machine Learning Based Intelligent Healthcare Assistance System

• Main Authors: Ke, Yen-Ting, 柯彥廷
• Other Authors: Huang, Chung-Chi
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/tw738u

碩士 === 遠東科技大學 === 機械工程研究所 === 105 === It is proposed to develop a machine learning based intelligent healthcare assistance system in the thesis. It works with quantitative measurement and intelligent analysis in the cloud system for somatic fitness. Automatic measurement will reduce labor costs and enhance the customization of the system for web-based healthcare assistance. The automatic measurement of the system is consist of electrocardiography (ECG), electroencephalography (EEG), electro meridian analysis system (EMAS). And it measures for the physiological, psychological and meridian energy state by sensors. The website of the cloud system with records, assessment, prediction, diagnosis, and prescription function was built by using ASP.NET tool and machine learning methods. By experimental results, The accuracy of intelligent healthcare assistance system using machine learning method is pretty good. The accuracy of the assessment system using multiclass decision forest method is over 88%. The accuracy of the prediction system using decision forest regression method is over 95%. The accuracy of the diagnosis system using multiclass decision jungle method is over 92%. And the accuracy of the prescription system using boosted decision tree regression method is over 100%. These experimental results show the good performance of the machine learning based intelligent healthcare assistance system.