Research Methods About Data Mining Technology In The Study Of Formula Compatibility Rules

1. Research purposes:This research firstly aimed at reviewing and summarizing common methods and its application in prescription compatibility law research under the guidance of TCM prescription compatibility theory and data mining theory, secondly at approaching strategies of prescription compatibility law research based on data mining technology by pointing out series problems from data acquisition to sorting, mining and analysis. Then descript the application of the foregoing strategies and methods through clinical study on late gastric cancer treated by prescription, for example, through data collection, sorting, excavation and analysis.2. Research methods:This research took reviewing papers about prescription compatibility law study based on data mining technologies as the starting point, and systematically summarized application of common data mining methods in it from theory overview, basic analysis process, application status of method and some other aspects. Then finally paper will show the application of strategies and methods through gastric cancer prescription data mining analysis instance based on the combination of self designed data acquisition-pretreatment system and open source data mining platform.3. Research results:① In terms of prescription database application, although many online prescription databases contains data from the traditional to the experimental level, but the completeness of data, sharing, openness, integration and terminology coding standards, logical structure of data tables, computer-aided design of tools needs to be further improved or refined. Subject to the above problem, currently in prescription compatibility law research based on data mining, data acquisition and pre-processing still takes more time;② The application of methods in data mining, such as frequencies, association analysis, clustering analysis has been more common in the compatibility rule study by data mining. But problems such as method usually be used singly, narrow perspectives, incomplete analysis results evaluation system still exist in the analysis of perspective literature;③ In terms of data mining and analysis software, although there are many visualization of open-source software at present, and many problems above can be sorted esaly, but it’s mostly designed for professional data analysts, the threshold for Chinese researchers are still high.4. Research conclusions:① Currently the application of prescription database and data mining techniques still exist many problems in the study. So, I think in research process we should pay attention to two big research strategy, "unified data structures standard, established data collection platform" and "reasonable organization mining process, integration custom analysis software", do our good job in database construction and custom mining software oriented TCM professional; In terms of method application, we should joint use frequency statistics, associated analysis, clustering analysis and compatibility network from herb, effect and more multiple angle on prescription compatibility law data mining research;② Compatibility rule is one of the core problems in pharmacology, understanding and generalization of the prescription at higher level. Data mining techniques can provide reliable methods for the analysis the compatibility rule of prescription from frequency of herb, effect and more multiple angles and levels. The advanced gastric cancer treatment prescription data mining case study showed that one can complete the prescription data mining analysis and draw a right conclusion quickly and efficiently under the guidance of those strategies and methods, and rational use of related tools.