Disease Identification And Early Warning Based On Massive Medical Data

In recent years, the aging problem is increasing, and with the huge population base in our country, the problem of medical is becoming more and more obvious. The contradiction between the huge demand and the limited medical resources is inevitable. China’s health care reform is promoting the transformation from the traditional mode of medical treatment to prevention, community health care and family care while optimizing the allocation of medical resources. The application of high and new technologies such as wireless sensor networks, wearable devices and mobile Internet to the health care field is an important part of health care reform. In recent years, wearable health monitoring based on dedicated hardware and mobile Internet has become a hot research topic in the current academic and medical fields. Through specific sensor hardware, the wearable health monitoring system get the user’s physiological parameters, such as the state of motion, heart rate and blood pressure, and analyze and process the medical data, provide a basis for further medical diagnosis and decision making.Parkinson(Parkinson ’disease s, PD) is a neurodegenerative disease of the nervous system, and old people over 60 years of age are more likely to attack. PD patients have obvious motor symptoms, first appeared static tremor symptoms, and the clinical manifestations such as slow motion and gait disturbance appeared after the aggravation of the disease. Therefore, the patient’s motor signs are an important indicator of clinical diagnosis. Early detection of pre-clinical patients has an important role in the prevention, treatment and improves the quality of life of patients.This thesis mainly combined wearable health monitoring and artificial neural network, obtains the user’s body behavior data through a number of triaxial acceleration sensors, then normalization, discretization and de- noising the data, and extracting valuable feature vector, analysis and design an effective early identification and early warning algorithm for Parkinson’s disease. At last, the performance and the recognition rate of the algorithm are analyzed by simulation, and the core algorithm is implemented by Java. The main research work includes the following aspects: first of all, obtain the user’s behavioral data through the sensor hardware and discretization and de-noising the data to get the clean and reliable data; then, use the feature engineering, select the mean value of absolute value, standard deviation, correlation coefficient and AR coefficient, Magnitude Area Signal to form the feature vector, as the input vector of the artificial neural network classifier after normalized treatment; next, select the multilayer feedforward neural networks as the recognition model, using BP algorithm to train the network parameters, design a reasonable neural network identification algorithm; in the end, the pending identification data is input to the trained artificial neural network, the results of the identification are shown in the form of graphs, to analyze the frequency and distribution of symptoms, get the extent of the user’s health and the possibility of Parkinson’s disease, and get the results of Parkinson identification.