Research And Application Of Medical Data Visualization Analysis

With the rise of information technology and its wide application in medical industry,the volume of medical data is growing exponentially,and it also has the characteristics of heterogeneous and high dimension,which is inefficient or even difficult for doctors to obtain effective diagnosis and treatment information through traditional medical data visualization methods to explore the potential laws of specific diseases.Integrating machine learning into the analysis and processing of medical data can not only reduce the high-dimensional problem of medical data,but also facilitate the extraction of the main feature information for analysis,which can effectively reduce the computational load,shorten the time complexity and improve the accuracy of medical data analysis.Therefore,this paper applies machine learning method to the visualization analysis of medical data,and proposes the T-SNE-DBSCAN algorithm for the similarity analysis of medical data,and constructs the corresponding visualization analysis platform.Aiming at the high dimensional problem of medical data,firstly this paper uses machine learning method to analyze the importance of eigenvalues of medical data.Using lung cancer detection data in the MIMIC-III medical database as the foundation,through pretreatment,using KNN algorithm,support vector machine(SVM)algorithm and random forest algorithm which is typical machine learning methods,classifying the data training,from the aspects such as sensitivity,specificity and accuracy to build the classification performance evaluation methods,so as to get the optimal algorithm,and uses it to calculate the eigenvalue of importance value,and use it as a data preprocessing method of data dimension reduction processing.It provides a similarity analysis tool for medical data.Based on the calculation and processing of the importance of eigenvalues and the idea of dimensionality reduction,then the paper presents the T-SNE-DBSCAN algorithm.During the similarity analysis of MIMIC-III lung tumor detection data,dimension reduction was achieved by calculating the importance of characteristic values of detection data.The machine learning method with the best similarity analysis effect was obtained through comparative analysis.On this basis,the T-SNE-DBSCAN similarity analysis algorithm was established.Experiments show that T-SNE-DBSAN algorithm has better judgment accuracy than the traditional single method machine learning algorithm.Based on the above research,the paper finally built a general medical data visualization analysis platform,through a variety of data visualization methods can realize the analysis of medical data processing,and helps the doctors to mine,present and predict the potential laws and risks in the medical big data,and conduct relevant medical experiment analysis and scientific research.