The Data Analysis Of Hyperthyoid Disease Data Based On Multidimensional Time-series

Hyperthyroidism is an illness seriously harmful to human health with long course and many complications, its incidence trends to raise in recent years. When hyperthyroid patients are provided medical services in hospitals, a large number of clinical trial data are generated over time, which is significant for analyze the state of hyperthyroid patients. It will be an effective work to study the hyperthyroidism’s generation and development for individual health plan by the combine of disease data and big data analysis technology.In the treatment of hyperthyroid patients, doctors will determine the condition of individual hyperthyroid patient by the following eight clinical examination indicators:T3(Triiodothyronine),T4(Thyroxine), FT3(Free Triiodothyronine),FT4(Free Thyroxine),TRAB(Thyrotropin Receptor Antibody),TSH(Thyroid Stimulating Hormone),TGAB(Thyroglobulin Antibodies),TPOAB(Thyroid Peroxidase Antibody) and so on. Then specific the rapeutic regimen will be prescribed accordingly. According to all the clinical examination indicators of individual hyperthyroid patient, we regard the individual patient as the basic unit ordering in the examination time points of indicators. Then the multidimensional time series of the clinical examination indicators for the hyperthyroid patient can be generated, on the basic of which, we can deeply research the cluster analysis of hyperthyroidism’s multidimensional clinical examination indicators time series. We believe similar patients with the same trend of clinical examination indicators can be found for doctors to have a deeper insight in the relation between examination indicators and hyperthyroidism. It’s helpful to provide a more accurate diagnosis and more targeted treatment for the patients.Based on the authentic clinical examination indicators of hyperthyroid patients, this paper design a clinical examination data analysis system based on multivariate time series and effectuate it. The following three modules are included.1) Data preprocessing module, working in the pretreatment of the source data and the synchronized treatment of structural data. On account of the authentic clinical examination indicators from hospitals including unstructured data with much noise, complex unstructured data is preprocessed firstly in this paper. Then a regularization algorithm for time series is presented, effectuating the synchronization of different time series dimension and time point.2) Multidimensional time series clustering analysis module, working in deep mining and analysis of clinical examination indicators time series’ synchronization. Based on it, the clustering algorithm DBScan is improved by bringing in a user defined parameter, the noise points share NoisePro. And multidimensional asynchronous clinical examination indicators time series clustering algorithm LabTS-CLU which is also presented based on density dividing thought.3) Parallel processing module, working in improving the extraction efficiency of overall system and dealing with huge amount of data. With a view to the clinical examination indicators of hyperthyroid patients from each organization accumulating through long-term and generating large amount of data, parallel computing framework, MapReduce is used to analysis and process the clinical examination indicators time series of hyperthyroid patients, and deploy it on open source distributed platform Hadoop in this paper.In the end, clinical trial data set from one top three hospital about over one hundred thousand hyperthyroid patients of the past ten years is used in experience, showing that the arithmetic in this system is superior to the existing mainstream of time series analysis algorithm on whether execution efficiency or accuracy rate. And after processed by the parallel computing platforms Hadoop, execution efficiency of the system analyzing have been improved greatly. The experimental results show the effectiveness of this system, which can provide technical support in some degree for doctors to have a comprehensive understanding of Hyperthyroidism.