Research On Disease Prediction And Prescription Recommendation Algorithms Based On Medical Data Analysis

In China,the digital hospitals will develop towards the smart hospitals.To this end,artificial intelligence is one of key technologies.At present,the basic information construction of most hospitals has been completed,but the data in the information system is simply saved or queried,and the potential value of the data has not been explored.Therefore,the state has introduced a large number of policies to empower the “ Internet + medical health ”,accelerate the construction of regional medical information sharing platforms,strengthen the integration and sharing of clinical data,and apply emerging technologies such as data mining and artificial intelligence to analyze medical data to mine the potential value of the data,and then provide citizens with high level of medical health services.Based on the excavation of outpatient medical records and prescription information in hospital information system,a k-nearest neighbor(KNN)algorithm based on density-based spatial clustering of applications with noise(DBSCAN)clustering and maximum entropy was proposed to predict patients’ diseases,and a rule,case and mixed prescription recommendation model was proposed.The model is based on disease diagnosis and treatment ontology.The main research contents are:1.Disease prediction method based on electronic medical record analysis.Compared with the prediction model of a specific target,the prediction for the patient similarity group is more universal.The method uses the KNN algorithm to predict the disease of the patient.On the one hand,for the massive medical record data,the DBSCAN clustering algorithm is introduced into the data mining process,the samples with high similarity are cropped,the sample set is reduced,and the algorithm calculation time is reduced.On the other hand,the distance calculation formula using the maximum entropy is used to calculate the distance between the two samples,avoiding the subjective influence of the sample similarity using the weighted Euclidean distance.Finally,the experimental results show that the proposed algorithm has higher precision.2.Prescription recommendation method based on big data analysis.On the basis of the traditional recommendation algorithm,the ontology database of disease diagnosis and treatment field is constructed,and the prescription recommendation is implemented by the rules,cases and mixed recommendation methods based on thedisease diagnosis and treatment ontology.In order to construct the ontology of the disease diagnosis and treatment field,the concepts and connections of diseases,symptoms,and medicines in the electronic medical record are expressed.Based on this,a prescription recommendation model based on rules,cases and mixed reasoning is constructed.Finally,the experimental results show that the proposed method has higher precision.In this thesis,the KNN algorithm based on DBSCAN clustering and maximum entropy is used to predict the disease of patients.The experimental results show that the proposed method is more precise than the traditional tree-based fast search KNN algorithm.The ontology-based rules,cases and mixed reasoning methods are proposed.It is proved by experiments that the ontology-based prescription recommendation method proposed in this paper is more precise than the traditional collaborative filtering based recommendation method.