Research On The Data Fusion Technologies In The Medical Care System

Data fusion is the process of integration of multiple data from sensors representing the same real-world object into a consistent, accurate, and useful representation, with certain rules and technologies. At present, most remote medical care systems set alarms based on the preset threshold of patient’s physiological data collected from sensors. These systems always have high false alarm rate, low diagnosis efficiency, and lack of personalized. The introduction of the data fusion technologies gives access to solving the problems above. With data fusion technologies, it can filter and analyze the collected data, and realize intelligent diagnosis of the patient through a certain algorithm and rules, which will significantly improve the level of telemedicine technology, achieve intelligent remote medical diagnosis with high efficiency, low misjudgment, and personalization.This paper proposes an intelligent data fusion system for remote medical care, named DFS-MC system for short. The system can be divided into two parts that are the modeling process and the processing process. The modeling process can set up PSVC and PSVP models via the user’s personalized physical database. Processing process mainly take charge of multi-level denoising of physiological signals, real-time diagnosis based on physiological data fusion engine and PSVC model, and health prediction based on PSVP model. The DFS-MC system runs based on the cooperation and complement of both one.The system involves a variety of technologies, including filtering technology of digital signal based on SE-D, lightweight PSVC modeling technology and PSVP modeling technology based on FIG. The filtering technology based on SE-D with high efficiency and strong robustness is suitable for real-time data processing environment; Lightweight PSVC modeling technology use formula of DT-Fn+MS+SVM to select training samples accurately, which can reduce the amount of samples to label, improve the quality of the training samples, and realize efficient modeling and accurate classification; PSVP modeling technology based on fuzzy information granulation can realize efficient modeling, reduce the sensitivity of the noise signals as well, and expand the scope of prediction as well, which provides strong support for doctor’s later intervention and decision-making.The DFS-MC system mainly realizes the functions of multi-level data denoising, the modeling of PSVC and PSVP, real-time diagnosis, regression fitting and health prediction. The test data is the real collect sign data provided by ZTE Technology co., LTD. The test proves that system recognition rate of noise can reach up to95%with multi-layer filtering method. Two modeling processes take shorter time. The classification accuracy of PSVC model is above96%. The fitting accuracy of PSVP model, according to users’ unique characteristics, can reach more than85%at most time. The efficiency and accuracy of the system have be proved effectively.