Research On Medical Data Mining Based On Convolutional Neural Network And Apriori Algorithms

With the increasing demand for medical services,the amount of data in the medical field is explosive growth,and has an upward trend year by year.These medical data truly record the patient’s visiting information and the whole process of doctor’s diagnosis and treatment,and can reflect the real situation in the medical field.At present,most hospital managers use simple statistical methods to process medical data.The utilization value of medical data resources is limited.In order to improve the utilization value of medical data and explore the hidden rules and links between medical data,this paper uses data mining technology to study the data mining of medical data in a three-A hospital.According to the experimental data obtained in this paper,a medical data warehouse is designed and constructed,and then three different mining methods are used to study the data.The main work is as follows:(1)Evaluation of doctor service quality based on K-means clustering algorithmsIn order to evaluate the quality of doctor’s service,the doctor’s visiting volume and patient satisfaction score in a period of time were selected as the main research data in the data warehouse.Three central points are selected to calculate the distance from all data to the central point,and all data are divided into three clusters according to the distance.After adjusting the centers of each cluster,the distance to the new centers is calculated and grouped,so that the results converge.Based on the clustering algorithm,all doctors are divided into three groups.Through the analysis,we can see that the results reflect the quality of service of the three groups of doctors: the new force,the main backbone and experts.The results can also provide decision-making basis for bonus cash,title promotion and special training.(2)Research on medical data mining based on convolutional neural networkIn this chapter,disease classification model and outpatient pharmacy flow forecasting model are proposed.The information of patients with coronary heart disease and pulmonary disease in data warehouse is trained and tested by disease classification model.The results show that the classification accuracy of patients with coronary heart disease is higher,and the three indicators of calculation accuracy P,recall R and accuracy A are calculated.The results show that the classification effect of the model is better.The first 15 days of outpatient pharmacy traffic data were input into the prediction model,and the next 5 days of outpatient pharmacy traffic were predicted.The results show that the first 3 days of outpatient pharmacy traffic forecast effect is better,and the second two days of outpatient pharmacy traffic forecast effect is general.(3)Hospitalization information mining based on association rulesIn order to explore the law of geriatric onset and the potential relationship between medical data,we extract hospitalization information of geriatric patients in data warehouse as research data,and use Apriori algorithm in association rules to data mining.After continuous pruning and connection,10 rules are finally obtained.Through the analysis of the data in the results,it is concluded that the number of hospitalized patients aged between 60 and 80 years old with heart disease and cerebrovascular disease is the largest,the department receives the largest number of urban patients and the medical insurance reimbursement expense accounts for about 53% of the total hospitalization expenses.