A Research Of Practical Case Entity Recommending Framework Based On Related Information Mining

With the developing of information,large numbers of medical data are accumulated in information system in medical area.These data contain valuable medical information,which can be utilized to assist the medical processes.So it is of great importance to dig into these data.In the diagnosis,doctors make treatments based on patients' condition and it depends mainly on their own experience and medical knowledge but sometimes the information is incomplete and decisions are uncertain.So making use of association mining of existed medical data can get to know the relation among data,the useful information and therefore can recommend similar medical cases,which can provide similar diagnosis and treatment for doctors.So the hospitals have a demand for medical case recommending.According to the demand of case recommending,we propose a practical case entity recommending framework oriented to related information mining.It aims to match the input case with existing cases and find the similar cases so it can provides guidelines and advice to doctors.The doctors can know the treatment of this disease and helps with diagnosis.The research includes:1.Propose a practical case entity recommending framework oriented to related information miningWe propose a practical case entity recommending framework.The architecture has five layers,including application layer,data management layer,association mining layer,source data management layer,data storage layer and the visiting interface above the layers.The application layer provides case input and recommendation output.The data processing layer is responsible for transforming the input data.The association mining layer is responsible for mining related data.The source data management layer is responsible for managing the source data.The data storage layer provides useful medical data storage.2.Design the associated data conversion moduleAccording to the characteristics of medical data used in this article,define transforming rules to transform the database of structured data into associated data so as to match them according to the structure.At the same time,the input information transformation module is designed to transform the input unstructured case information into association data.3.Propose a case matching algorithm based on association dataFor the recommendation process,a matching algorithm for two case ontologies is proposed.Two ontologies are input and the similarity is output.The case matching process includes an association data matching algorithm and a temporal sequence-matching algorithm.4.Verify the architectureIn this paper,the data of gastric cancer,colorectal cancer,gastritis and enteritis were selected as the basic data,and the database of valid information is constructed.On the basis of medical data,we construct a case recommendation platform.Through a detailed description of the construction process,the framework shows the feasibility and usability.