Design Of Heart Disease’s Classification And Diagnosis System Based On Data-Driven

The incidence of heart disease is very high all over the world,and it is the number one killer that threatens human life.In the traditional medical industry,doctors relying on their knowledge and experience diagnose diseases for their patients.In order to reduce the risk of misjudgment due to insufficient experience of doctors in small and medium-sized hospitals,classification algorithms of machine learning can be used to assist doctors in judging the patient’s condition.The classification model supported by medical big data can make corresponding judgments on whether the patient is ill or not,after a comprehensive analysis of the patient’s examination data.Although the classification algorithm of machine learning can achieve a high level of classification accuracy,it cannot completely replace the doctors.Doctors can use the classification results given by machine learning as reference opinions and make more accurate diagnosis and treatment judgments for patients based on their own experience.It is a key step for smart medical care that combines the massive data with machine learning to build an auxiliary medical system which can bring innovation and change to the medical field.The research purpose of this paper is to construct a heart disease diagnosis model with high level of classification accuracy and strong generalization ability,through a series of algorithm optimization and model improvement.The main research contents are as follows:(1)Aiming at the problem that the classification level of RBF kernel support vector machine is largely limited by selection of parameter,the differential evolution algorithm is used to replace the traditional grid search method to find the parameter combination that fits the model.After expanding the parameter solution space from the discrete domain to the continuous domain,it is indeed possible to find a parameter combination that is more suitable for the model,which makes the model has a better performance.(2)Due to the limitation of the convergence speed and the population size,differential evolution algorithm may be trapped in a local optimal solution.In order to find the global optimal solution in a larger continuous solution space without take too much time overhead,the mutation strategy is changed by introducing mutation operators,and the population diversity is maintained as much as possible in the early stage,which is beneficial for fully searching the solution space.And according to the population characteristics of each generation,the scaling factor F and the crossover probability CR are dynamically adjusted to balance the problem of global optimal solution and convergence speed.And through experiments,it is proved that the RBF kernel support vector machine classification model optimized by the improved differential evolution algorithm does have a further improvement in a series of evaluation indicators such as classification accuracy and AUC value.(3)In order to make the model have a ability of self-evolving,the self-evolution process of the model is simulated by collecting a large number of medical literatures and then adding relevant attributes to the data set after manual screening and judgment.The model selfevolution simulation experiment can prove that with the expansion of medical knowledge in the industry and the update of cardiac examination equipment,we can use its conclusions to expand the dimension of the data set,and complete the self-evolution function of the heart disease classification model by rebuilding the training model,which can improve the model’s ability to classify and judge the condition of the disease.The experimental results show that,after preprocessing the data set,the RBF kernel support vector machine optimized by the improved differential evolution algorithm has a good performance level and generalization ability in heart disease classification and diagnosis.