The Application Research Of Constructing Bone Tumor Diagnostic Knowledge Database

Bone tumor is a common bone disease, its category includes benign tumor, malignant tumor, metastasis tumor, cancroid pathological changes, and etc., it's usually seen in juvenile and does many harms. Bone tumor assisted diagnosis system is such a computer programme that applies the design theory and method of expert system, simulates medical expert diagnosis and treatment in bone tumor. The application of computerized diagnosis system in the bone tumor assisted diagnosis could improve diagnostic accuracy in bone tumor, be convenient for medical stuff and be benefit to patients. Computerized diagnosis system is based on knowledge database. The integrality, accuracy and robust of knowledge are related to the accuracy and stability of the system. Knowledge acquisition is a bottleneck in the design and development of expert system. How to construct and evaluate the knowledge in knowledge database plays animportant role in expert system.This research started with establishing diagnosed cases database. After discussing with medical experts, 100 cases were selected from 120 cases inpatients in Institute of Bone Tumors of Chinese PLA, The Second Teaching Hospital, The Fourth Military Medical University as the resource of knowledge database; Clinical checkup knowledge database including image information was set up according to the case content; coding all information of clinical diagnoses, using SQL SERVER 2000 to storage database. The center problem of rough set is to simplify knowledge, which means to get rid of the irrespective attributes and attribute values in database and make the knowledge contained in database appeared. The method of data digging was data collecting to recognize an effective, novel, useful and understandable model. In third part, we established two algorithm: data reduct and MDRBR(Mining Default Rules Based on Rough Set), The objective is to acquire diagnostic knowledge from cases automatically from the diagnosed cases database, in the end established knowledge database that could be used for consequence. And in this part, we also discussed question how to inosculate between acquire knowledge by data mining and experience of clinician, and estimated for knowledge. In fourth part, computing for diagnosed database with above method, acquired 26 and 19 items diagnostic knowledge. In analysis and discussion part, we compared two algorithms, discussed next step for perfecting and improving in the method of data mining and knowledge estimation. It was testified both in theory and practice that this method has a good effect in acquiring medical diagnostic knowledge and it's a good method to solve the bottleneck problem in constructing expert system, andimproving speed of establishing expert system, advancing maturity and veracity. It could provide evidence for posterior consequence using knowledge, and afford material for clinical teaching.There are follow innovative idea: solving the bottleneck problem in constructing medicinal assisted diagnosis system using technology of data mining; starting with classical rough set theory, established two algorithms: data reduct and MDRBR(Mining Default Rules Based on Rough Set), The objective is to acquire diagnostic knowledge from cases automatically from the diagnosed cases database, in the end established knowledge database that could be used for consequence.