Research On The Intelligent Diagnostic Technology Based On Medical Information

EMR is a computer-based patient record, also known as a computerized medical record system. It uses computer, health cards and other electronic equipment instead of handwritten paper medical records to store, manage, transmit and reproduce the patient’s medical records. It covers all of the information about the paper medical records. Compared with the traditional handwritten medical records, electronic medical records not only significantly improve the efficiency of doctors, but also realize comprehensive storage and management of different information.Because of the uncertainty of medical diagnosis, it increases the difficulty of diagnostic decision. At the same time, with the development of health care, clinical professional becomes ever more specific, and the department physicians tend to focus on their specialist, which makes their knowledge become more and more narrow. So, it will be very difficult for doctors to analyze all aspects of the patient comprehensively and the diagnostic process has certain risks. Therefore, medical diagnostic decision support systems are becoming increasingly important for medical doctors.The emergence of EMR makes we can collect more convenient to get medical data, while data mining text classification algorithms are matured and developed into effective artificial intelligence algorithms. Text is based on the design and implementation of EMR to get patient’s medical records in hospital, and extract the required description and preliminary diagnosis of cases. Case description as the training set that text classification required, and preliminary diagnosis of case as the classification category train an intelligent classification model, and then the new case description which doctor input use this model to classify, and find disease category, making this model can support intelligent diagnostic.The main work of this thesis has four aspects:(1) the design of an EMR system, which is mainly used to store the information of all patients’ medical records. It is not only convenient for doctor to query patient-related treatment information, but also provides data support for this thesis;(2) use the EMR Railway Group Second Board of the Central Hospital as the data source to obtain medical records, and extract case description related to diagnosis. The core content includes: chief complaint, medical history, physical examination and auxiliary check results. Use the text pretreatment methods to describe consecutive cases as a set of words that best reflect the characteristics of the patient, and to quantify treatment.(3) Selecting 977 hospital records of three kinds of diseases, using the BP neural network algorithm to classify, and get high flexibility mining model. During the process of establishing BP neural network classification model, the number of hidden layers and the learning rate BP neural network is changed. Meanwhile, the effects of various parameters on the classification accuracy and recall of various kinds of diseases are analyzed.(4) Using BP neural networks, support vector machine(SVM) and Naive Bayesian(NB) classification methods to train the processed data, design model, test, compare and analyze the results.