Research On Disease Association And Patient Migration Using The Home Page Of Tumour Patient Records

Home page of inpatient medical record is the main basis for inpatient medical record registration,disease classification and examination and China has made the requirement for its writing standard very early.It contains a lot of valuable information.New analytical methods can be used to obtain more in-depth and valuable information,so in this paper,data mining is carried out from two aspects:disease association and patient migration.The research in this paper covers the following four main areas.(1)In this paper,we propose an improved FP-Growth algorithm,and use it to construct disease networks to study the correlation between diseases.The frequent itemset mining algorithm is used to get disease pairs that often appear simultaneously.Aiming at the frequent itemset mining algorithm FP-Growth,this paper makes some improvement on the establishment process of FP-Tree,and the algorithm speeds up when the minimum support threshold is low.There is a step to judge whether the inserted node already exists in the child node in the insertion tree.This paper uses the method of space for time to build a new variable CLF,and uses bit operation instead of cyclic comparison operation.The experiment shows that in the data set of this paper,when the minimum support is set to 100,the running speed of the algorithm is increased by 12%,but when the minimum support is set to 500,the performance is poor and the running speed is reduced by 5%.Then the correlation analysis was performed for the disease pairs,and relative risk(RR)was calculated.The disease pairs with RR>1 were stored in the graph database.A disease network with 2604 nodes and 34020 edges is constructed.Through the analysis of the disease network and the calculation of confidence and co-occurrence,association rules between diseases were obtained.For example,kidney fat can lead to intestinal adhesion(confidence level 0.887),and cirrhosis and liver cancer often occur simultaneously(co-occurrence level 0.56).(2)In this paper,multiple regression models are used to model the number of cancer patients who migrate between cities,and the important influencing factors of cancer patients’migration are analyzed.Three regression models were used to model the effects of medical,economic,population and spatial barriers on the migration of cancer patients.The results showed that the R~2 value of the model built by Cat Boost was the highest,reaching 0.894.Finally,the model parameters were used to analyze the factors related to the migration of cancer patients,and it was found that the largest factor affecting the migration of cancer patients was the number of Third-class hospital at the destination,while the indicators at the starting point had little influence.(3)This paper constructs an analysis system of the home page of medical records.In addition to the basic function of storing and managing the first page of medical records,the system also provides the function of analyzing and displaying the relationship between diseases and patient migration studied in this paper.A visual interface is designed for it,which provides a function to inquire about the relationship between diseases,which can be used by researchers and ordinary users.