Intelligent Pre Diagnosis Method Based On Personal Information Of Patients

With the rapid development of information technology, biomedical field has entered massive data era. In the field of medical and health data, the hospital’s clinical data accumulated in the course of the year has not b een effectively utilized. Under the traditional doctor-patient mode, there are some bad phenomenon, such as pre-illness prevention, unsatisfactory treatment experience, lack of services after the treatment. Starting from the hospital electronic medical records, the thesis developed an intelligent pre-diagnosis method based on semi-structured electronic medical records. This can not only save patients’ time to determine the department, but also can greatly improve the efficiency of hospital staff to co-ordinate the distribution of health care workers and save medical expenses. Electronic patient records are semi-structured, in which the complaint is relatively short, the personal information of patients like present illness, past history, etc. is not only an important basis for doctors to diagnose patients, and can also serve as an important feature of this research.To solve the problem of pre-diagnosis department, this paper uses a rule-based approach to explore the issue. Finally, support vector machine and the convolution neural network classification algorithm were applied to the multi-classification problem. In data preprocessing, this paper describes construction of medical entity-relationship database, standardized patient departments. For how to effectively use the course record patient information and achieve an accurate diagnosis and treatment purposes, experiments are conducted to examine which medical information will help enhance the pre-diagnosis accuracy.In the process of exploration and research of pre-diagnosis problem, this paper performed comparison experiments to optimize the model, meanwhile carried out various experiments based on different information in order to discover the contribution of each part, with the goal of improving the performance of the pre-diagnosis system. By comparing the results of the experiments and the evaluation system, the intelligent pre-diagnosis method based on the patient’s personal information described performs better than rule-based classification model and support vector machine. Good performances on two different kinds of datasets also show that this model has strong anti-noise ability.