Research And Implementation Of Medical Data Analysis System Based On Differential Privacy

The rapid development of medical informatization has given birth to the Age of Medical Big Data.How to use medical big data effectively is a very important research topic.The potential value of medical big data is to be made the best use of by data mining.For example,in the monitoring of infectious diseases,the public health department can conduct comprehensive disease surveillance and rapid detection of infectious diseases through the nationwide electronic medical record database,and analyze the characteristics of disease transmission through data mining.In addition to the value for the society,the impact of medical big data on individuals cannot be ignored either.Let’s take chronic diseases as an example.Patients’ time-tagged data could be collected by wearing wearable devices for a long period of time or performing regular physical examination.Analysis on the time axis can lead to the result of the development of the disease.If the time is used as a marker and combined with the local meteorological data,the onset characteristics of some climate-sensitive diseases can also be analyzed.But for the massive medical data,we are still very inefficient on its utilization rate.On the one hand,major hospitals all have their own hospital information systems,and usually their systems are relatively closed to the outside world.In addition,various hospitals have different data format standards.All of the factors set previously consist in the difficulty of sharing and using data.On the other hand,the high sensitive nature of medical data is a greater difficulty in sharing data.A little misuse can cause very big problems for the credit of individuals and related institutions.This paper presents the design of medical data analysis system based on differential privacy.The system can analyze the medical data in the background of mass data,and the privacy protection mechanism is added in the analysis process to ensure that the privacy of data owners is not leaked.The specific research is as follows:In terms of the rapid growth of medical data in the background of the age of medical big data,a dynamic clustering algorithm that can incrementally mine medical data is proposed in combination with the analysis of streaming data.This algorithm can mine and analysis incremental medical data.As for the high sensitivity of medical data,differential privacy protection mechanism is introduced in the process of data mining to protect the privacy of data owners during the entire process.The introduction of this mechanism can realize the privacy protection requirements during data mining.Based on the theoretical basis of the two points set above,and combining with the actual needs of public health monitoring,this paper designs a medical data analysis system that can be used to detect diseases.The system can dynamically cluster the diseases according to the disease information,and analyze the type and number of diseases.