A Computer Rule-based Expert System Model Investigation For Correlations Between The Zang-fu Organs

The Five Elements Theory (the five elements being metal, wood, water, fire and earth) in Chinese Medicine was used by the ancient Chinese medical doctors, for a very long period of time, to explain various physiological, pathological and pathogenesis phenomena in human diseases. It is becoming realized that the rigid frame-work proposed in the Five Elements Theory does not totally explain the complexity of the pathogenesis among the Zang-fu organs. Professor Deng Tietao, a monumental figure in TCM education proposed the used of Five-viscera Correlation Theory to replace the Five Elements Theory in Chinese Medicine. In a paper written in1988in Guanzhou College of Chinese Medicine Journal(issue2) by the title "Brief discussion on Five-viscera Correlation to replace The Five Elements Theory", Professor Deng formally put forward the hypothesis or doctrine. Although the title suggested a replacement, but the actual intension of Professor Deng is to use the Five-viscera Correlation to augment the Five Elements Theory so as to further explain the complex correlations among the Zang-fu organs. Since then, the Five-viscera Theory have been used in various researches in Chinese medicine including basic theoretical studies and clinical investigations. Various studies which used mathematical techniques and computer technologies on the Five-viscera Theory have been tried; these include statistical methods, structural equation modelling and data-mining. The shortfall of these methods is that they commonly used a black-box approach which treat relationships among the Zang-fu organs as a set of co-eff icients; the higher the co-efficient values the more closely related one Zang or Fu is with the other Zang or Fu. This thesis suggests a different approach by using Rule-based expert system.ObjectiveIt is the intension of this study to find out ways to scrutinize in details the relationship among the Zang-fu organs when disease happens. In the Yellow emperor’s Classic of Internal Medicine, it is mentioned that the "· five-viscera are all related and the qi moves from one visceral to another in an orderly manner…".This gives the author the inspiration that in order to really understand the correlations among the five-viscera, it is necessary to follow the pathogenesis of the disease as its effects transform from one Zang or Fu to another Zang or Fu. The search for a technique which can do just that started and fruitfully succeeded when the author came across an old medical diagnostic system called MYCIN.MYCIN was an expert system, developed in the early1970’s. It used artificial intelligence to identify bacteria causing severe infections like bacteremia and meningitis. The expert system name MYCIN is derived from the name of antibiotics, since many antibiotics have the suffix "-mycin". The MYCIN expert system was also used for the diagnosis of blood clotting diseases. It’s capability includes antibiotics dosage recommendation for each patient, taking into account of the patient’s body weight. MYCIN was originally developed at Stanford University, USA, over a period of five or six years. The system was written in Lisp (A list-based expert system computer language), it was a product of the PhD thesis of Edward Shortliffe who is a student of Bruce Buchanan and Stanley N. Cohen. MYCIN was never actually used in clinical applications, yet research indicated that the proposed therapy is accurate in about69%of cases. If judged using the same criteria, the performance of medical experts were no better than that of the MYCIN system.This thesis proposes to use Rule-based expert system as a means to realize the Five-viscera Correlation Theory from a new perspective; that is to use the explanation facility of a Rule-based expert system to follow through the pathogenesis of a disease as it transforms from one visceral to another. This will avoid the previous attempts using black-box approach which all resulted in a serious of co-efficient values. The values showed how closed related the Zang-fu organs are, but they cannot show in what ways these organs are related, nor do they show the sequence of events that had happened to these organs during the onset of a disease that caused Zang-fu organs dysfunctions.Methods(1) Data collection:Collect a total of120syndromes and their533related symptoms from books of "Syndrome differentiation of Zang-fu organs" and from "Clinic Terminology of Traditional Chinese Medical Diagnosis Treatment" These include50single Zang or Fu syndromes, and12two-two Zang-Fu related syndromes from the book of "Explanation of TCM Diagnosis through pictures and charts" and59two-two Zang-Fu related syndromes from the book of "Terminology of Traditional Chinese Medicine for Clinical Diagnosis"(2) Data pre-processing:In TCM, there is a certain degree of fuzziness in the terminology used in syndrome names and symptom names. Many different terms may mean the same symptom. Also in Chinese characters, the swapping of the order of two characters may still mean the same symptom. But in computer analysis, every piece of information needs to be exact so that the computer can do exact matching for accurate analysis. Therefore, it is necessary to pre-process or clean up the information first before they can be put to use in the Rule-based expert system. The pre-processing includes:①Standardizing the syndrome names. In TCM one syndrome could have many different names. For example,"Liver despondence and qi sluggishness"’(肝郁气滞)can also be called "Liver qi stagnation."(肝气郁结).The121syndromes name adopted in this study is mainly based on the specification outlined in the "Clinic Terminology of Traditional Chinese Medical Diagnosis Treatment". A syndrome is fixed to one standard name.②Restricting the order of characters of symptom names. For example,"Many dreams "(多梦)and "Dreams many"(梦多)means the same in Chinese. The rule of thumb is to use which makes more sense in spoken language. In the above case, the term "Many dreams"(多梦)is adopted. And once decided, the whole knowledge-base will use the same terminology.③Excessive characters are removed and terms are defined with the least number of characters. For example "Lack of strength in the four limbs"(四肢乏力),"Tiredness of body and limbs"(肢体疲倦),and "Lack of strength in the body and limbs"(肢体乏力)all have the same meaning. This can be reduced to "Lack of strength"(乏力).④Breaking down complex symptoms into multiple single symptoms. For example "Lateral Lower Abdominal Distension and Pain"(少腹胀痛)can be broken down into "Lateral Lower Abdominal Distension"(少腹胀)and "Lateral Lower Abdominal Pain"(少腹痛).⑤Putting descriptive symptoms into the knowledge-base meta-data knowledge. For example,"Body and eyes become yellowish, colour like orange"(身目发黄,鲜如橘色).The descriptive symptom "colour like orange"(鲜如橘色)will be kept as meta-data that is used to describe the main symptom "Body and eyes become yellowish"(身目发黄).(3) Rule-based expert system model design. The design of the Rule-based expert system model is the heart of this investigation. In this thesis work, the five components of Rule-based expert system data structure are designed.①The Knowledge acquisition facility design②The Knowledge acquisition facility design③The Knowledge base design④The Inference engine design⑤The Explanation facility design⑥The User interface design In a Rule-based expert system, there are three types of knowledge? the Substantive knowledge (Fact), the Strategic knowledge (Rule) and the Meta-data knowledge (additional information to describe Fact).(4) Testing the Rule-based expert system model for Zang-fu organs correlation analysis.ResultsThe Rule-based expert system sucessfully analyzed a case of Myasthenia gravis of which the patient has7symptoms. It is found that the flow of syndrome transfer is from Deficient of Pi qi to Deficient of Gan blood, and from Deficient of Pi qi to Deficient of Pi qi to Deficient of Fei qi.This research analyzed121Zang-fu syndromes and their associates532symptoms. These in turn produces1851symptom facts and151syndrome correlation rules in a Rule-based computer expert system for analyzing the Zang-fu correlations. The system produces1851symptom facts and151syndrome correlation rules. This research achieved the following results:1. The first research to use Rule-based expert system to analyze the correlations among the Zang-fu organs.2. Completed the design of an expert system model for the analysis of Zang-fu syndrome correlations, and use the C language to construct a Rule-based expert for such analysis.3. Breaking through the shortfall of the previous "black-box" approach in Zang-fu correlation analysis, this research provides a method to reveal the direct flow of Zang-fu syndromes after the onset of a disease.Conclusion From the whole research process, one can draw the following conclusions:1. During the research process, it is found that there are non-uniformity in the naming of syndrome and symptoms:these non-uniformity includes positioning swap of Chinese characters in syndrome and symptom names, use of excessive characters in symptoms names, many characters one meaning in symptom descriptions, one word multiple meaning in symptom names, and additional descriptive phrases in symptom names. These non-uniformity has been taken care of in this research.2. It is found that there are not too many cross syndrome rules in the Zang-fu diagnostic method, except the following:(1) Deficient of Pi qi affects all qi deficiencies in all the Zang. Deficient of Pi qi, Deficient of Gan blood and deficient of Xin qi affects one another. Deficient of Pi qi, Deficient of Fei qi and Deficient of Xin qi affects one another.(2) Liver qi stagnation can cause Pi qi deficiency, stomach flaming, stomach deficiency, Phlegm-Fire disturbing mind, systolic blood stasis, heart fire hyperactivity, lung qi deficiency. Liver flaming may cause lungs ying deficiency.(3) Excessive cardiac fire and deficient of kidney ying affects each other.(4) Deficiency of heart-yang and deficiency of kidney-yang are closely related. So as deficiency of spleen-yang and deficiency of kidney-yang.(5) Deficiency of lung-ying and deficiency of kidney-ying are closely related. So as deficiency of liver-ying and deficiency of kidney-ying.3. Jing, Qi, blood and body fluid impact on the Zang-Fu can cause subtle relationships between internal organs.4. By testing the system against the Myasthenia gravis case, the expert system analyzed and found that the patient may have spleen qi deficiency, lung qi deficiency and liver blood deficiency syndromes all at the same time. And that spleen qi deficiency is the cause that lead to lung qi deficiency and liver blood deficiency. Knowing the syndrome flow will improve the accuracy of prescribing Chinese medicine.5. Through the study of Zang-fu correlations, the research raises the clinical practice from single syndrome and main symptoms to many syndrome and many symptoms. The correlations of syndromes and their symptoms flow is useful for clinical diagnostics. The Rule-based expert system for Zang-fu correlation analysis is very meaningful for clinical analysis; not only for Myasthenia gravis, but also for many difficult and complex diseases.