| Objective:1.Discussions and prescriptions on the diagnosis and treatment of low back pain in classical Chinese medicine books were systematically sorted and compiled.Digging out the medication rule of Qianjin Fang in the treatment of low back pain and analyzing the characteristics of prescription before Tang Dynasty in the treatment of low back pain.It provides resources for clinical staff to read classics purposefully and provides new ideas for clinical treatment of low back pain.It also lays a data foundation for the construction of artificial intelligence prescription platform for the diagnosis and treatment of low back pain.2.Based on the knowledge of Traditional Chinese Medicine(TCM),the artificial intelligence prescription platform for low back pain was preliminarily constructed through Monte Carlo method and other technologies,which contributed to the development of TCM in the era of "intelligent medical treatment".Methods:1.The basic work is to excavate,collect,classify,collate and summarize the information of the representative TCM classical ancient books on low back pain before the Republic of China.Focus on reading the discourse and prescriptions on the diagnosis and treatment of low back pain in TCM classics and proofread with the point-correction edition edited and published by modern editors,and summarize after checking.Data mining and analysis were carried out on the prescriptions of Qianjin Fang for the treatment of "low back pain" using the TCM inheritance auxiliary platform,so as to obtain the results of drug frequency,prescription rule,new prescription analysis and so on,and to explore the characteristics of the prescriptions for the treatment of low back pain.2.Construction of low back pain prescription platform of artificial intelligence based on the classic of TCM knowledge: Sort out the classics of TCM extract to the content of the "low back pain",form TCM classics knowledge network,generate from the "random prescriptions" to "virtual prescriptions",and further training to "virtual prescriptions" through the real prescription.An artificial intelligence prescription platform for low back pain has been initially formed.Results:The first part : Classical carding of TCM low back pain and data mining of "Qianjin Fang"1.Huang Di Nei Jing,Treatise on Exogenous Febrile Disease,Qianjin Fang have a lot of discussion on the etiology and pathogenesis,principles and methods of treatment,and prescription medication of low back pain,which are effective references for the treatment of low back pain by modern Chinese medicine.According to the data collected,classified and summarized,the prescriptions related to low back pain were retrieved as follows.Including one prescription from Shanghan Lun;one prescription from Jinkui Yao Lue;29 prescriptions from Valuable Prescriptions for Emergencies;4prescriptions from Qianjin Yi Fang;40 prescriptions from Wai Tai Mi Yao Fang;4prescriptions from Pu Ji Ben Shi Fang;12 prescriptions from Sanyin Ji Yi Bingzheng Fang Lun;7 prescriptions from Danxi Xin Fa;15 prescriptions from Jingyue Quanshu;10 prescriptions from Zheng Zhi Hui Bu;7 prescriptions from Yizong Jinjian;one prescription from Sisheng Xin Yuan.2.Study on the medication rules of Qianjin Fang for the treatment of low back pain based on data mining technology,a total of 38 prescriptions were adopted.Developed by Institute of Chinese Materia Medica of China Academy of Chinese Medical Sciences,the version is "Traditional Chinese Medicine Inheritance Assistant System Software(V2.1)".The results showed that,according to the statistics of the drug frequency in the38 low back pain prescriptions,the top 10 high-frequency drugs were successively:cinnamon,liquorice,eucommia ulmoides,poria cocos,angelica,chuanxiong,dried ginger,ginseng,Atractylodes atractylodes and aconite.Four Qi and Five Flavours and their return to meridian: Among the 38 prescriptions for the treatment of low back pain in Qianjin Prescriptions,the frequency of attribution of Four Qi of traditional Chinese medicine ranged from the highest frequency of warm medicine 101 times to the lowest frequency of cool medicine 16 times.Rank from large to small,we can get: the first: warm medicine treatment of low back pain 101 times;Second: 59 times of low back pain were treated with plain drugs.The third place:thermal medicine for low back pain 38 times;Fourth place: cold drug treatment of low back pain 22 times;The fifth place: cold drug treatment of low back pain 16 times.Five tastes of drugs in accordance with the order from high to low frequency,the result is:sweet(139),pungent(106),bitter(84),acid(18),salt(11),flat taste(10).By statistics according to the analysis,in 38 prescriptions in Qianjin Fang for the treatment of low back pain,access to 687 statistics,the result is: the kidney meridian occupy 127 times;Spleen meridian occupied 122 times;Liver meridian occupied 118 times;The heart meridian occupied 111 times;Lung meridian occupied 85 times;The stomach meridian occupied 63 times;Bladder meridian occupied 19 times;Large intestine meridian occupied 15 times;Gallbladder meridian occupied 14 times;The pericardial meridian was the least,occupying 13 times.Study on the rule of prescription formulation: Among 38 prescriptions,there are 31 commonly used drug combinations matching each other in the treatment of low back pain by Qianjin Fang.The top 10 drug combinations or drug pairs ranked in order of their frequency of occurrence were: Licorice-cinnamon,liquorice-angelica,licorice-poria cocos,rhizoma ligustici wallichii-licorice,rhizoma ligustici wallichii-angelica,rhizoma ligustici wallichii-cinnamon,ginseng-licorice,angelica-cinnamon,cinnamon-eucommia bark,cinnamon-lateral root of aconite.Rule analysis revealed a total of 42 combinations of drugs.Current party drugs or medicine to appear in the formula,according to the drug or drug combination confidence from high to low arrangement,the top ten: angelica,cinnamon-> licorice;aconite->cinnamon;rhizoma ligustici wallichii,angelica-> licorice;rhizoma ligustici wallichii,liquorice-> angelica;rhizoma ligustici wallichii,cinnamon-> licorice;rhizoma ligustici wallichii,liquorice-> cinnamon;angelica,cinnamon-> rhizoma ligustici wallichii;ginseng,licorice-> poria cocos;licorice,angelica,cinnamon-> rhizoma ligustici wallichii;rhizoma ligustici wallichii,licorice,cinnamon-> angelica.According to the network statistical results of associated drugs,the top 15 drugs in the list of frequency of use were cinnamon,licorice,eucommia ulmoides,poria cocos,angelica sinensis,chuangxiong,dried ginger,ginseng,Atractylodes macrocephala,aconite,cortex peony,rehmannia glutinosa,ginger,duhuo,and paraffin.The network included 13 medicines,including aconite,dried ginger,paraffin,eucommia ulmoides,cinnamon,rhizoma chuanxiong,ractylodes atractylodes,licorice,poria,angelica,ginseng,ginger,and dozuo.According to the statistics of drug frequency,it can be seen that the drugs in the core network are all included in the 15 drugs with the most drug frequency.The correlation analysis of drugs showed that there were 16 drug pairs in the prescriptions of Qianjin Fang for the treatment of low back pain,and the correlation degree of these drug pairs was more than 0.055.The analysis of drug core combinations showed that according to the mutual constraint principle of correlation degree and punishment degree,the constraint of correlation degree of 8 and punishment degree of 2were set.A total of 14 groups of drug combinations containing three to four core Chinese medicines in each group were counted in the prescription of Qianjin Fang for the treatment of low back pain.The new prescription analysis showed that 7 new prescriptions appeared according to the algorithm after the core drug combination was statistically obtained.The second part: the construction of artificial intelligence prescription platform for low back pain based on TCM knowledge element1.Extract and sort out the contents of "low back pain" from classic ancient books of Chinese medicine:First,the scope of TCM classics was determined.By consulting experts and referring to textbooks such as Internal Medicine of TCM and Formulas of Chinese medicine,this study finally determined 15 TCM classics that mainly need to be sorted out.Including Suwen,Lingshu Classic,Synopsis on Febrile Diseases,Synopsis on Golden Synopsis,Bijie Qianjin Yaofang,Qianjin Yi Fang,Jingyue Quanshu,and Danxi Xin Fa,etc.On this basis,through searching the keywords of "low back pain","back pain","waist and knee" and so on,the theoretical analysis and treatment of low back pain in the classic Chinese medicine is sorted out.Then manual verification was carried out to remove the prescriptions that were too little associated with the main treatment and low back pain,or the treatment was external treatment,and the prescriptions for the treatment of pregnancy,menstrual,postpartum and traumatic low back pain were retained.According to the classic contents of TCM low back pain after screening,proofread the literature,search and calibrate the rare words,rare words and missing words in the classic items one by one,and modify and supplement them one by one.To unify the alias of traditional Chinese medicine for low back pain,and to supplement the omissions in traditional Chinese medicine prescriptions,etc.At the same time,to unify and standardize the names of traditional Chinese medicine in classic ancient books of TCM,for example,"thick park" unified to "magnolia bark","cassia heart" unified to "cinnamon","peony" unified to "radix paeoniae alba",and so on.2.Forming the network of TCM classical "knowledge element"Based on the fact that the online prescription platform is dominated by modern Chinese medicine decoction pieces and the principle of "knowledge element",a total of 195 Chinese medicine sets were formed.According to the Pharmacopoeia of the People’s Republic of China(2015 edition),the standard dose of drugs that can be used for low back pain is determined and supplemented in the collection of traditional Chinese medicine.The standardized clinical manifestations were summarized and classified to obtain a standardized clinical manifestation set containing a total of 51 major symptoms,and then the 51 clinical manifestations were classified according to the order and content of the ten questions,so as to meet the follow-up needs of artificial intelligence for symptom inquiry and diagnosis.Under the guidance of the principle of "knowledge element",combined with the above standardized clinical manifestation set,as well as the relevant efficacy,indications and usage attention of traditional Chinese medicine,a "knowledge element" network about the treatment of a symptom or sign of low back pain by a certain traditional Chinese medicine in the treatment of "low back pain" is generated.Based on the traditional Chinese medicine collection for the treatment of low back pain and the standardized clinical manifestation set,a drug-symptom network with195 lines(i.e.195 traditional Chinese medicine items)and 51 lines(i.e.51 standardized clinical manifestations)was finally formed.A total of 9,945 TCM "knowledge element".After expert discussion,we removed some Chinese medicines in the 195 traditional Chinese medicines whose functions are very similar in the collection,or which are rarely used in oral prescriptions in modern clinical practice.In the end,192 traditional Chinese medicines were included in the "virtual medical case",that is,written into the code program.3.Generate from "Random Medical Cases" to "Virtual Medical Cases"Under the guidance of the Monte Carlo method,several clinical manifestations were randomly selected from a standardized clinical manifestation set containing 51 symptoms to form a symptom combination,and it was assumed that this was the information that patients input to the AI platform.Then,a certain number of drugs were randomly selected from the collection of 192 traditional Chinese medicines to form a prescription,and then the prescription was assumed to be able to treat this group of symptoms,forming a "random medical case".On this basis,based on the TCM "knowledge element",the realistic conformity degree of "random medical case" is evaluated,that is,the rationality rate is evaluated,and the "virtual medical case" is formed.Then,the neural network of deep learning trained by "virtual medical case" was used to establish the prediction relationship between "clinical manifestation and traditional Chinese medicine",and the combination of traditional Chinese medicine with the highest probability of "cure" was calculated.In order to avoid the phenomenon of "over-fitting",the neural network is further trained through "real medical cases".Through the test of "real medical cases",the algorithm engine of the artificial intelligence prescription platform is improved and the predicted value of fitting is more accurate.The results of the preliminary construction of the artificial intelligence prescription platform for low back pain are shown in the We Chat mini-program "Classical Knowledge Element",in which patients can obtain prescriptions based on TCM syndrome differentiation by answering the corresponding questions.The third part: Effectiveness evaluation of artificial intelligence prescription platform for low back painAmong the 30 patients with low back pain,21 were female patients,accounting for 70%.There were 9 male patients,accounting for 30%.3 cases were aged 20-30 years old,accounting for 10%;8 patients aged 31-40 years old,accounting for 26.7%;7 cases were aged 41-50 years,accounting for 23.3%;5 cases were aged 51-60,accounting for16.7%;3 patients aged 61-70,accounting for 10%;3 patients aged 71-80,accounting for10%;One case was aged 81-90,accounting for 3.3%.After 5 experts respectively scored the effectiveness of symptoms and prescriptions of 30 patients,the average score distribution was as follows: 1-2 in 4 cases,accounting for 13.3%;2-3 in 3 cases,accounting for 10.0%;3-4 were divided into 10 cases,accounting for 33.3%;4-5 in 13 cases,accounting for 43.3%.The average score above 3 accounted for 76.6%.The overall mean score was 3.607,which was 72.13 points on a percentage scale,with a standard deviation of 0.548.Conclusions:1.The systematic collation and compilation of TCM classical literature related to low back pain facilitates academic research on TCM low back pain,and provides resources and methods for relevant clinical personnel to read classics on purpose and on demand.Based on the inheritance of traditional Chinese medicine auxiliary platform,data mining of Qianjin Fang treatment of low back pain have showed the academic characteristic mainly reflected in the reuse of the wind,make the wine agent,use small prescriptions,use aconite skillfully.The thinking methods and the experience formula on medication,provides more thinking for the clinical treatment of low back pain.2.Through the combination of TCM classics and artificial intelligence,the artificial intelligence prescription platform for low back pain can be preliminarily built,which can adjust the re-visit medication according to the patient’s medical feedback,and plays a beneficial role in inheriting TCM classics and promoting the modernization of TCM.In the trial operation stage,the prescriptions issued by the artificial intelligence prescription platform generally conform to the clinical medication standards.For the clinical applicability of the prescription,the average score of the expert is 3.607,which indicates that the prescription of the platform is basically effective and can be used for clinical reference. |
PDF Full Text Request