Medical Big Data Multi-dimension Mining Based On Hana

With the aging population problem is becoming serious, the medical expenses of our country is increasing the proportion of GDP, and the medical big data field will generate trillions of market value, the medical field has begun to embrace the medical big data, which is used to apply in the practice of medical research, medical clinical and medical management. The data generated during the diagnosis and treatment of the patients in medical institutions are referred to as medical data, including patients’ personal information, medical records, medical reports, medical imaging data and indicators examination data, etc. Taking Shanghai Ruijin Hospital as an example, tens of thousands of patients come to visit it every day. So much data has pulling together, years of growth has tested and verified the storage capacity of the hospital’s database. With effective data management in a long time, data mining on these medical big data then enables it to help the hospital managers to carry out the operation decision is worth of research.In the process of medical industry’s electronic information development, it has encountered the challenge of mass data and unstructured data. Especially with the prevalence of "big data" concept in recent years, many big companies have come up with a lot of solutions for medical big data. In view of this situation, it is significant to carry out the medical big data research work for the hospital information development.With the research of online analytical processing system(OLAP), the paper designs a set of visualization system for supporting multi-dimension complex query based on the Hana platform. Then related work of data mining has also been proceeded on the basis of multi-dimension data sets, thereby to find out the potential clinical medical unknown knowledge among big data.Firstly, the data generated in the online application system of Shanghai Ruijin Hospital is needed to be pretreated. The business description of multiple database table and the correlation relationship are given in the paper. It also describes OLAP model based on the clinical data. On this basis, this paper analyzes the system feasibility and functional requirements, and then designs a multi-dimension visualization system of clinical medical data, four major modules of this system are introduced respectively: data preprocessing module, data modeling module, data mining algorithms module and data visualization module.Then this paper introduces some specialized data analysis on the case of multiple diseases. According to the examination of some important indicators of hyperthyroidism patients, the treatment process can be divided into four stages: clinical remission stage, biochemical remission stage, immune remission stage and immune cure stage. The recurrence to last stage may occur. By analyzing the data of different stages, the cure rate of patients with hyperthyroidism can be obtained at different stages. And diabetes is a long-term high blood sugar symptoms with many detected indicators, but there is a strong correlation between them, which can be analyzed for indicators of diabetes data to obtain a potential relationship, so as to provide data support for reducing the useless indicators examination.Finally, in order to verify the validity of the results of the data analysis, this paper also uses a variety of data mining algorithms(association rules algorithm, clustering classification algorithms, etc.) on the patient’s inspection reports, test indicators, drug use and other data, then acquire complementary medical conclusions that could provide some theoretical support for disease analysis in the future.