Design And Implementation Of Medical Diagnosis Aided System Based On Hadoop

The information system of electronic medical records has been carried out for 10 years. Electronic medical records, laboratory information system, medical image transmission and storage system and other information systems have been introduced into the hospital, which makes the hospital has accumulated a wealth of medical data resources. Taking the Ruijin Hospital Affiliated to Shanghai Jiaotong University School of Medicine as an example, the clinical data emerged is about 60 TB every year. Medical data including clinical diagnosis, test results and other types, has the "4V" features of big data. How to use these data to provide data support for clinical medical diagnosis, so that doctors can better serve patients, is the higher requirement of information construction.Based on the actual project of clinical data platform construction and deep application in Ruijin hospital, the paper designed and implemented the medical diagnosis aided system based on large data processing platform Hadoop. Firstly, the characteristics of medical data are analyzed, and the relevant technologies of data mining are summarized. To study the prescription data of diabetic patients, the drug recommendation based on association rules is studied, and the efficiency of the algorithm is improved. In view of the test data of patients with hyperthyroidism, the complication prediction based on logistic regression, combined with the characteristics of medical data, is analyzed, and the relevant technologies of data cleaning, integration, transformation and prediction are analyzed. Using MapReduce and Spark two models, the paper makes a research on the prediction of drug recommendation and complication of the medical diagnosis aided system, and verifies the validity of the research method and technology. The main work of this paper is as follows.(1) Drug recommendation based on association rules. Using association rule algorithm in data mining to find out the prescription of medical prescription data, and find out the relevant information from the frequent prescription. In order to meet the requirements of the timeliness of drug recommendation, this paper improves the classical Apriori algorithm of association rules, and proposes a Apriori algorithm based on prefixed-itemset.(2) The complication prediction based on logistic regression. The data of various test index data of the patients were combined with the logistic regression algorithm in data mining to establish the classification model. With the help of the model, the effect of assistant diagnosis was achieved. In the realization of forecasting function, this paper uses the technique of to ensure the accuracy of the forecast model, which is based on the "1～99" percentile value method, the numerical discretization based on entropy, and the model evaluation method.(3) The implementation on Hadoop. Using the MapReduce model, the Hadoop platform is used to realize the function of drug recommendation. Through the experiment, the performance of the classical and improved Apriori algorithm is analyzed, and the effectiveness of the improved algorithm is verified. Using the Spark model, the function of the complication prediction is realized, and the function is analyzed in detail, and the function of the realization of the function is analyzed, and the prediction results are given.