Research And Design Of Data Mining Platform For Massive Medicaltreatment Data

With the rapid development of medical information, the traditional digitalinformation systems and electronic health records systems of hospitals can not meetthe massive medical data storage and medical data services requirements. Aftermassive medical data generation and collection, how to efficiently storeheterogeneous, massive, real-time, a variety of data in order to achieve fast andaccurate response to large-scale complex health data queries; while applying datamining and knowledge discovery theory of medical history data modeling andanalysis, how to mine key physiological characteristics from medical data, andreliable, fast and efficiently detect early disease and predict health risks for users toprovide valuable medical services, which need to be solved in the current healthareas.Based on the characteristics of mass medical data, and the current lack ofexisting massive medical data mining applications, the massive medical data miningdesign platform is proposed, which includes massive medical data mining platformmodel, rapid statistical query algorithm of medical data, as well as medical relevanceof data mining. In this paper, the specific contents are as follows:(1) Based on the characteristics of massive medical data and medical servicerequirements, massive medical data mining platform model is proposed, includingdesign principles of data mining platform, massive medical data storage strategy, aswell as massive medical data processing. The platform model has the flexibility,scalability, reusability and so on.(2) Based on fast query response requirements of service built by data statisticsof massive medical data and strong association characteristics of medical data, a faststatistical query algorithm of massive medical data based on statistical tree andincremental computing is proposed, which optimizes the storage model, improvequery efficiency, and efficiently support data statistical based medical service, such ashealth status reasoning.(3) For personalized health care plans and concurrent disease forecast demand, the association mining algorithm for medical data is proposed, including health statusreasoning services based on Bayesian network and concurrent disease forecastingservices based on association rules, which can provide early warning and diseaseprevention recommendations to individual users, and can be used for analysis ofclinical decision support and disease concurrent in medical institutions.In this paper, on the proposed massive medical data mining platform, massive,diverse medical data is efficiently stored in order to achieve fast and accurateresponse to large-scale complex health data queries; through the analysis of existingdata sets mining, the correlation between the various test results and physicalcondition and resulting illnesses related individuals, as well as the associationbetween disease and illness, are studied, which is used to provide the service of healthstatus reasoning for patient.