| Purposes and content:To explore the solutions for preliminary data preparation for the ancient prescription data mining related problems, this paper analyzes the situations of composition factors of prescriptions from the ancient Chinese medical literature. The paper is from the following point of view:Firstly, after reviewing of the background of data mining, the main process of data mining, the subjects related, the characters of the data be used, and the theory of data mining, the paper targets the research at the methods of preliminary data preparation for the data mining of ancient prescriptions.Secondly, by retracing the methods which be used in the researching of Chinese medicine, the paper find the methods feature of data mining. It is hard to study the message from the ancient Chinese medical literature by statistics method. It is surely limited by human conditions if we use documentary method. So data mining is most valuable in the study of the ancient Chinese medical literature.Thirdly, by reviewing both the areas of Chinese medicine used data mining and the data preparation for the ancient prescription data mining, it is believed that the theory of Chinese medicine should be adapted into the progress of the data mining to find ?new? knowledge in Traditional Chinese Medicine. And there are lots of problems need solutions during the data preparation courses.Fourthly, the purpose of ancient prescription data mining is to find out the relationships between drugs and diseases, between drugs, dose-effect relationships, and so on, which are never been point out clearly by anyone. There are special situations during the mining course, mainly refers to the uneasy standardizing of the data, the complex relationships of the components in one prescription, the data of diseases and drugs should be completed from the information of other resources, the data of ancient prescription are not the daily transaction style as usually data mining used. Due to the purpose and the special situations, the Chinese medicine knowledge and documentary method should be added into the preliminary data preparation course. Fifthly, thinking about the problems in the preliminary data preparation course, the paper lists the views, conclusions, or the results below:1. There are differences in the ancient prescription documents belonging to different styles or times, which demand us to choose and use them accordingly. For the purpose of saving the diverse information about prescriptions in these documents, the database should be designed based on the contents.2. The minimum conditions to form a record of prescription are disease, drugs and correspondence between them. And any changes in diseases, drugs, dose, drug preparation, dosage form, and so on, can make it necessary to derive a new record.3. Since there are diverse disease description methods in different documents, the database should be designed to keep different kinds of information.4. The disease description words in ancient documents should be divided to separate disease units out. And the disease units should be standardized with appropriate manner. The traditional documentary methods should be valued during the course, and the information of disease unit weighting in the whole disease and the information of treatment should be properly handled.5. The standardization of drug name must be carried on after the database?s accomplishment. Based on the standard drug name list, making sure of the drug's standard name, the name should be marked anew.6. If a prescription be used as a drug in another prescription, the first prescription should be divided into drugs, and join the later prescription by its drug form.7. Information from herbal literature should be saved in drug database after the drug name standardization. The prescription database and the drug database can be seamlessly jointed by the shared standard drug names. The drug database should save the information of the drug?s indications, nature of cold or hot, tastes, toxicity, and relationships with other drugs, at the same time, the information source can be traced.8. The information of drug preparation should be saved and standardized too.9. Some of the names of dosage form in ancient prescription have different meanings with nowadays, for example, the"powder"made by concentrating, the Dan by chemical reaction, the"Wine Agent"brewed from drugs. These dosage forms should have special new names. 10. The liquid for decoction, or taking pill and powder, the Yao Yinzi of decoction, the binder of pill, the base of ointment, and so on are special materials. Their attributes should be marked out.11. In practice, some drug could be in a form different with the dosage form of its prescription, so, the dosage forms of drugs should be saved independently.12. To facilitate data processing, the paper divided the changing progress of the Chinese medicine use metrology system into three phases, Han and Tang period, Song and Jin period, and Yuan, Ming and Qing period.13. To explain the conversion relationship between the units of the Chinese medicine used metrology system in history and international metrology system nowadays, the paper makes reference tables for data processing.14. After analyzing the meanings of Qian, Qianbi, Zi, and Zibi, the paper ascertains that one Fangcunbi in the times of Tao Hongjing equals 5ml by different methods at the first time. And the volumes of some natural objects and Qianbi, Qianwubi, etc. are figured out too.15. After analyzing Fen, the unit of weight in ancient prescription documents, by its changes of meaning and value, the paper proved the medicine use weight unit Fen?s meaning change from ?big Fen in Zhu system? to ?Fen in Qian system? happened in Yuan period, not in Song period as History of Song mentioned.16. In the process of ancient prescription data mining, it is an inevitable demand to turn all the non-weight units into weight units. But it?s not proper to do so when it is hard to define the ancient drugs? quality and standards. ?Dengfen? could be treated according to Tao Hongjing?s explanation.17. For the ancient prescription data mining, different drug doses should be in one standard form. The form should be daily dose or each dose figured out according to the original document, and expressed in grams.18. Based on the analysis of problems above, the database tables and the Entity Relationship Diagram for the ancient prescription data mining are designed.Sixthly, the elements relationships inside one prescription record are studied. On this basis, a series of problems are investigated, including determining a drug?s direct treatment function to certain disease, dose-effect relationship, detecting customary combination of drug, and identifying the principal, the assistant, the complement, and the guide in one prescription.Conclusion:1. The ancient prescription data mining means, a whole process of standing on Chinese medicine knowledge, using data mining technology, collecting and integrating prescription data from ancient Chinese medicine documents, mining the data for new Chinese medicine knowledge, and expressing the new knowledge in Chinese medicine language.2. The ancient prescription data have the following characteristics: a. They are from diverse sources. b. They are example data. c. Some of their attribute data are commonly missing. d. The original words which they use are commonly not standard. e. There are complex relationships between the data from same attribute or different attributes.3. The ancient prescription data processing should follow the following principles: a. Ensure that the original words in documents can be found via data records correspondingly. b. Once there is any change among the attributes of a prescription record, the possibility should be considered to build a new record. c. Wholly standardize the non-standard terms used in related attributes. d. There is necessity to make full use of known knowledge on Chinese medicine, and to value the involvement of relevant disciplines.4. The terms of some attributes should be wholly standardized. Disease terms, symptoms terms, pathogenesis terms, treat terms, and drug indications terms should be treated together. Drug name should be standardized mainly from the prescriptions, and unified drug names should be used in different tables. The standardizing working should be carried on among the terms of drug preparation, special use of drug, document information, so on. The common standardizing working steps are below: a. Collect all the relevant terms. b. Analysis the meaning items of all the terms. c. Merge same meaning items and give the standard name. d. Mark the standard name by studying the original term, meaning item and standard name.5. Ancient prescription dose data processing:The standard ancient prescription dose form should be daily dose and each dose expressed in grams, with certain condition of dosage form, preparation, and method of use.The paper issues tables of conversion relationship between the units of the Chinese medicine used metrology system and International metrology system for data processing.For the very first time, the paper ascertains that one Fangcubi in the times of Bencaojingjizhu equals 5ml by different methods, and accordingly systematically obtained the volume of Qianbi, Zibi, Fangcunbi, etc. in different times.For the first time, the paper ascertains the values of the weight unit Fen in different times in gram.The paper believes that it?s not proper to turn all the non-weight units into weight units at present time. The temporary processing method of the non-weight units should be kept the original words but no other way.6. The paper believes that there are some kinds of possibility for the relationships between diseases, between drugs, and between disease and drugs in one prescription record. The direct treatment function of one drug to one disease could be supported only after the elimination of other kinds of relationships. Besides data mining, we can affirm the function by the prescription data with one drug, drug addition or subtraction related to certain symptom, and the drug treatment function data.The paper also believes that during the course of ancient prescription data mining, determining dose-effect relationship, detecting customary combination of drug, and discovering the identifying patterns of the principal, the assistant, the complement, and the guide components, are all based on the works of preparing the data reasonably and ascertaining the key relationships above.7. As an inevitable link of ancient prescription data mining, the preliminary data preparation work shoud be accurate and reasonable. And the work also needs not only systematic standardization based on original documents as possible, but also the consideration of the relationships between prescription elements.In this process, there are some common problems need attentions, such as: the use of Chinese medicine theory or conclusion, standardization of the various properties, attention to analyzing and processing all the details. It is also important to consider other traditional methods, such as: traditional Chinese medicine way of thinking, textual research, cultural study, the logical method, mathematical methods, measuring methods and experimental methods.As to the preliminary data preparation for ancient prescription data mining, it is necessary to combine rigorous treatment of various details with specific mining purpose. Both accurate and reliable data and operability should all be considered. For this is a relatively detailed study, some aspects can be appropriately simplified.
|
PDF Full Text Request