Study Of Cirrhosis Disease Prediction Based On Machine Learning

The hospital produces a large amount of data every moment,including clinical laboratory data,disease diagnostic data,biochemical test data,and instrumental inspection data.Exploring potentially valuable data and knowledge in medical data based on machine learning is used to predict the formation mechanism and development trend of diseases,and to provide data support for improving disease diagnostic efficiency and accuracy.At present,most of the various types of data in hospitals are only used for inquiries,and the in-depth analysis and utilization of large amounts of historical medical data needs to be strengthened.The paper reviewed and analyzed 4033 CT examination reports of real chronic liver diseases,and normalized the textual attributes of CT reports that had been structured.Usually,when a doctor describes a CT report text,it is difficult to avoid subjective judgment.After structured processing of the CT report text,part of the description features will have a synonymy of different words.Doctors describe different terms of the illness and actually express the same medical concept."No exception" and "normal" are concepts.Therefore,constructing a set of machine learning samples requires normalization.The paper uses the Spark Mllib machine learning algorithm to quantify the machine learning sample set to support the Spark Mllib machine learning algorithm data type.The paper uses Spark MLlib decision tree machine learning algorithm,random forest algorithm,support vector machine algorithm and naive Bayes algorithm to establish a predictive model of chronic cirrhosis disease,and carries out prediction assessment experimental study,and compares the accuracy of the four algorithm prediction models.The recall rate,F-Measure value,and AUC value were selected.After the comprehensive evaluation,the decision tree and support vector machine learning model were selected as the predictive model for chronic cirrhosis.The main research contents and contributions of this paper include the following aspects:The paper runs Spark's machine learning algorithm,predicts,analyzes and evaluates real experimental data,and constructs a classification model of liver cirrhosis,which has certain practical application value.Due to the limitation of the experimental research data collected by the paper,the machine learning prediction model can only predict two types of diseases:liver cirrhosis and non-cirrhosis,and the prediction accuracy rate reaches 98%,which has certain practical application value.