Pattern Classification And Early Detection Of Disease Of Wrist Pulse Signals And EEG Based On M-health

Mobile health is an emerging discipline. It combines achievements in a variety of disciplines and technical areas, receiving widespread attention and developing fast in the context of revolution of medical care. From infrastructure of mobile health--medical electronics and communication technology, to its superstructure--intelligent diagnostic algorithms of clinical decision-making and health pattern classification, mobile health technology is increasingly widely used in the field of personal health care, thereby facilitating users informed of physiological state of their own, and making self-identification and diagnosis. However, on the other hand, the in-depth analysis of the physiological signals and feature recognition methods are far from perfect. Especially for a number of chronic diseases, there is no effective mobile medical system and algorithm witch can make ideal detection of disease and health care, therefore, to some extent, it restricts the development of the mobile health.In this thesis, with the pulse signal and EEG as research objects, signal processing and pattern recognition methods are used to provide a paradigm for physiological signal analysis and mobile health solutions. First of all, for a typical class of cardiovascular disease, the stability of the spectral pattern in time domain is verified, and the disease features contained in frequency domain of pulse signals are revealed, thus it realizes a new type of non-invasive diagnosis for cardiovascular disease. In addition, from the perspective of anesthesia monitoring, recording EEG signals are used to evaluate the level of activity of human brain. Parameters in time-frequency domain and nonlinear domain are extracted. By using multiple linear regression and ARX model, the computational model for depth of anesthesia (DOA) is established. Then, the Bispectral Index (BIS) is used as a criterion to adjust the model’s structure and parameters. Simulation and clinical data are used to confirm effectiveness of this method.Finally, this paper presents a mobile health model based on smartphone, with pulse signals and EEG as examples, a platform based on Android smartphone is developed. It achieves physiological signal acquisition, wireless transmission and intelligent analysis. Then, basing on the network communication technique, a self-organizing medical information sharing network is presented. In this mode, the mobile health model not only achieves the collection and transmission of physical signals, but also automatically realizes health condition classification synchronously. This mode effectively contributes to personal applications of mobile health and revolution of medical information svstem.