Research And Application Of Medical Data Mining Based On Post-Relation Database

Thousands of database systems are under operation all over the world currently, covering almost every aspect of human society, whatever the information systems inside an enterprise, an institution or business process systems in different industries as well as usual information processing or intelligence index. All of them are dependant on database technologies. The healthcare industry features a variety of complicated levels and numerous connections in medical treatment as well as structural multi-dimensions in medical data, which induces some problems that the traditionally relational database unable to simulate so complex data relation and greatly reduce the efficiency of data access.With the rapid development of computer technologies in healthcare industry, the medical data increase explosively as a result of accurate record of all the medical information such as medical history, diagnosis, physiological test, treatment, drug management information and HIS. To mine useful information in massive data and to master the mutual relation and law of development among various diseases so that to summarize the effect of all sorts of therapies is momentous to disease diagnosis, treatment and medical research.Aimed at the data nature in healthcare industry, we create loss scoring function index based on Bayesian Principles with the help of object-oriented characteristics and multi-dimensional framework technology of Caché database as data mining platform to improve medical assessment indicators. Meanwhile, we bring forward the genetic algorithm based on the created function index as fitness function which is to be generated to hybrid algorithm after the combination with decision tree algorithm. Thus the hybrid algorithm is able to redeem the shortages that decision tree algorithm predispose itself to large data sets and Genetic Algorithm need too much training time. Through the data-mining with hybrid algorithm and the modal evaluation with loss scoring function based on Bayesian Principles in physical examination data, we are able to find out the mutual relations and relevant principals among diet structure, habits and life style so as to provide available assistance to reduce probability suffering from hypertension.