Deep learning-based recognition of stained tongue coating images

Objective To build a dataset encompassing a large number of stained tongue coating images and process it using deep learning to automatically recognize stained tongue coating images.Methods A total of 1001 images of stained tongue coating from healthy students at Hunan University of Chinese Medicine and 1007 images of pathological(non-stained)tongue coat-ing from hospitalized patients at The First Hospital of Hunan University of Chinese Medicine withlungcancer;diabetes;andhypertensionwerecollected.Thetongueimageswererandomi-zed into the training;validation;and testing datasets in a 7:2:1 ratio.A deep learning model was constructed using the ResNet50 for recognizing stained tongue coating in the training and validation datasets.The training period was 90 epochs.The model’s performance was evaluated by its accuracy;loss curve;recall;F1 score;confusion matrix;receiver operating characteristic(ROC)curve;and precision-recall(PR)curve in the tasks of predicting stained tongue coating images in the testing dataset.The accuracy of the deep learning model was compared with that of attending physicians of traditional Chinese medicine(TCM).Results The training results showed that after 90 epochs;the model presented an excellent classification performance.The loss curve and accuracy were stable;showing no signs of overfitting.The model achieved an accuracy;recall;and F1 score of 92%;91%;and 92%;re-spectively.The confusion matrix revealed an accuracy of 92%for the model and 69%for TCM practitioners.The areas under the ROC and PR curves were 0.97 and 0.95;respectively.Conclusion The deep learning model constructed using ResNet50 can effectively recognize stained coating images with greater accuracy than visual inspection of TCM practitioners.This model has the potential to assist doctors in identifying false tongue coating and prevent-ing misdiagnosis.

胃癌脾虚证唾液蛋白指纹图谱分子诊断模型研究

目的研究建立胃癌脾虚证唾液蛋白质指纹图谱新型分子诊断模型。方法采集未经手术和放化疗治疗的57例胃癌患者(脾虚证组40例,非脾虚证组17例)、45例健康志愿者(正常对照组)的唾液标本,用弱阳离子交换型(WCX)纳米磁珠联合基质辅助激光解析离子化飞行时间质谱(MALDI-TOF MS)技术进行检测,获得各组标本的蛋白指纹图谱。比较各组唾液蛋白质质谱数据,找出胃癌脾虚证组与正常对照组、胃癌脾虚证组与胃癌非脾虚证组之间具有统计学意义的差异表达蛋白质峰;采用决策树算法计算出多个变量即质荷比(m/z)变化对两组样本的判别价值,确定最优化的诊断模型。结果胃癌脾虚证组与正常对照组唾液蛋白质质谱比较共有106个差异蛋白质峰有统计学意义,胃癌脾虚证组与胃癌非脾虚证组唾液蛋白质质谱比较共有6个差异蛋白质峰有统计学意义(P<0.05或P<0.01)。筛选建立了由m/z为5439.67、2411.67、3619.18共3个主要差异蛋白峰组成的胃癌脾虚证诊断模型;临床回代检验该模型的灵敏度和特异度分别达100%和93%;交叉验证法进一步验证诊断模型的灵敏度和特异度分别为87%和89%。结论基于WCX纳米磁珠结合MALDI-TOF-MS技术建立的胃癌脾虚证唾液蛋白质组无创性分子诊断模型敏感性高、特异性强,对胃癌脾虚证的诊断具有良好效率。

A Network Pharmacology Approach to Uncover the Molecular Targets and Associated Potential Pathways of Lycii Fructus for the Treatment of Retinitis Pigmentosa

Objective To explore the molecular targets and associated potential pathways of Lycii Fructus(LF,Gou Qi Zi,枸杞子)in the treatment of retinitis pigmentosa(RP)by the approaches of network pharmacology and bioinformatics.Methods The potential blood-entry active ingredients and targets of LF were retrieved by Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform(TCMSP).RP-related gene targets were retrieved through disease comprehensive databases.Protein-protein interaction(PPI)network of LF component-targets and RP disease-targets was constructed by STRING,and the intersection of the 2 networks was extracted.Gene Ontology and Kyoto Encyclopedia of Genes and Genomes(KEGG)pathway analysis of theintersection network were conducted by Database for Annotation,Visualization and Integrated Discovery(DAVID).CytoHubba was used to screen the key targets.Results A total of 188 chemical constituents related to LF was retrieved from TCMSP database.45 active ingredients were screened according to pharmacokinetic parameters oral bioavailability(OB)and drug similarity(DL).36 active ingredients were further screened and 201 targets related to these constituents were obtained.206 target genes directly related to RP were obtained from the disease comprehensive databases,and 89 genes were obtained from the intersection of componenttarget and disease-target PPI network.These genes were mainly involved in intracellular signal transduction,GTPase activity regulation,cell morphology regulation,and other biological processes.Molecular functions were mainly related to Rho guanine nucleotide exchange factor activity,GTPase activator activity,receptor signal protein serine/threonine kinase activity and so on.They were enriched in the cytoplasm,cell membrane,Golgi apparatus,and other regions.The mechanism was related to cell cycle pathways,neurotrophin signaling pathways,Ras signaling pathways,and so on.10 key gene targets of LF in the treatment of RP were screened.Conclusions The material basis for LF to exert its pharmacodynamic effect is 36 active ingredients such as cycloartenol,mandenol,and so on.The key targets of LF in the treatment of RP include 10 genes,such as Rho,PAK,and so on.The main mechanism is related to the regulation of the Ras signaling pathway,neurotrophin signaling pathway,cell cycle related pathway,and other signaling networks.

Research on Text Mining of Syndrome Element Syndrome Differentiation by Natural Language Processing

Objective Natural language processing (NLP) was used to excavate and visualize the core content of syndrome element syndrome differentiation (SESD). Methods The first step was to build a text mining and analysis environment based on Python language, and built a corpus based on the core chapters of SESD. The second step was to digitalize the corpus. The main steps included word segmentation, information cleaning and merging, document-entry matrix, dictionary compilation and information conversion. The third step was to mine and display the internal information of SESD corpus by means of word cloud, keyword extraction and visualization. Results NLP played a positive role in computer recognition and comprehension of SESD. Different chapters had different keywords and weights. Deficiency syndrome elements were an important component of SESD, such as "Qi deficiency""Yang deficiency" and "Yin deficiency". The important syndrome elements of substantiality included "Blood stasis""Qi stagnation", etc. Core syndrome elements were closely related. Conclusions Syndrome differentiation and treatment was the core of SESD. Using NLP to excavate syndromes differentiation could help reveal the internal relationship between syndromes differentiation and provide basis for artificial intelligence to learn syndromes differentiation.

Visualization analysis of the research status of FU Qingzhu’s Obstetrics and Gynecology based on bibliometrics

Objective This study examined the research status and development process of FU Qingzhu’s Obstetrics and Gynecology(Fu Qing Zhu Nv Ke,《傅青主女科》,FQZNK)in the past 40 years with bibliometrics and visual analysis.Methods Retrieved all related literature in the research field of FQZNK from the domestic and foreign databases:China National Knowledge Infrastructure(CNKI),China Science and Technology Journal Database(VIP),Wanfang Database,and Web of Science(WOS)core database,including Science Citation Index Expanded(SCIE),Social Sciences Citation Index(SSCI),and Arts&Humanities Citation Index(A&HCI).The search range was from January 1,1980 to March 10,2021.In addition,bibliometrics and CiteSpace 5.7.R2 software were used to analyze literature types,published journals,cited literature,the number of author publica-tions,co-author networks,co-institution networks,keyword co-occurrence networks,keyword clusters,and keyword bursts.Results A total of 678 valid records were included in the final dataset.Literature types,high publication journals,highly cited literature,high-yield institutions,high-yield research teams,and high-productivity scholars in this research field were found through bibliometrics.Liter-ature types can be divided into four categories,among which 451 are theoretical studies on academic thoughts of FQZNK,accounting for 66.5%of the included journals.The Journal of Shanxi Traditional Chinese Medicine had the largest volume of published articles(61),ac-counting for 9.0%of the total number of the included journals.The most cited literature was ZHOU Mingxin’s article“Using the quantitative method to discuss author’s authenticity and formula characteristics of FU Qingzhu’s Obstetrics and Gynecology”,which was cited 94 times.Hunan University of Chinese Medicine,the institution with the most publications,published 45 articles,and YOU Zhaoling,the most published author,published 33 articles.Moreover,it was found that most high-yield researchers came from high-yield institutions and that Hun-an University of Chinese Medicine had the most research on FQZNK.Keyword co-occur-rence analysis revealed that the keyword“FQZNK”had the highest frequency(597 times)and the highest centrality(1.00).Keyword cluster analysis used the Log-Likelihood Ratio(LLR)al-gorithm to form eleven important clusters:#0 treatment aiming at its root causes,#1 gynecopathy,#2 Siwu Decoction(四物汤),#3 FU Qingzhu,#4 post-partum,#5 infertility,#6 dysmenorrhea,#7 sterility,#8 coordinate the heart and kidney,#9 Danggui Buxue Decoction(当归补血汤),and#10 treatment.It was found that the prescriptions of FQZNK were studied mainly before 2000,the theoretical studies were mainly conducted before 2010,and its clinic-al application was mainly explored from 2010 until now.Diseases such as dysmenorrhea,morbid vaginal discharge,infertility,metrorrhagia,and polycystic ovarian syndrome(PCOS)have recently become popular topics in this field.Conclusion The current study provides more scientific,accurate,and comprehensive sci-entific support for further research and development of traditional Chinese medicine(TCM)in FQZNK.With this foundation,people can use burst detection to ascertain the current hot-spots in research,get their development trends,and forecast future research directions.In ad-dition,infertility,morbid vaginal discharge,flooding,and PCOS treatments based on TCM syndrome differentiation are currently popular research topics for FQZNK.

Research on Herb Pairs of Classical Formulae of ZHANG ZhongJing Using Big Data Technology

Objective To analyze various herbal combinations in Treatise on Exogenous Febrile Diseases(Shang Han Lun,《伤寒论》)and Synopsis of Prescriptions of the Golden Chamber(Jin Gui Yao Lve,《金匮要略》),seeking to identify fundamental rules dictating the selection of herbal combinations through probability models and big data technology.Methods A total of 252 formulae were collected from Treatise on Exogenous Febrile Diseases(Shang Han Lun,《伤寒论》)and Synopsis of Prescriptions of the Golden Chamber(Jin Gui Yao Lve,《金匮要略》)by ZHANG Zhong-Jing.Formulae were then preprocessed with all herb names standardized.The concepts of candidate herb pair and candidate herb pair probability were proposed to analyze the rules of combinations in classical formulae based on probability statistics.MapReduce parallel computing framework of distributed big data technology was adopted to analyze large data samples combined with inverted index algorithm.Results The results showed that the core herbs were Glycyrrhizae Radix Rhizoma(Gan Cao,甘草),Cinnamomi Ramulus(Gui Zhi,桂枝),Zingiberis Rhizoma Recens(Sheng Jiang,生姜),Jujubae Fructus(Da Zao,大枣),Paeoniae Radix Alba(Bai Shao,白芍),etc.43 high-frequency pairs co-occurring 10 times or above were extracted,and 35 of these combinations were recognized as traditional herb pairs,such as Cinnamomi Ramulus(Gui Zhi,桂枝)-Glycyrrhizae Radix Rhizoma(Gan Cao,甘草),Zingiberis Rhizoma Recens(Sheng Jiang,生姜)-Jujubae Fructus(Da Zao,大枣),and Cinnamomi Ramulus(Gui Zhi,桂枝)-Ginseng Radix Et Rhizoma(Ren Shen,人参).The other 8 pairs of combinations,such as Paeoniae Radix Alba(Bai Shao,白芍)-Zingiberis Rhizoma Recens(Sheng Jiang,生姜),Paeoniae Radix Alba(Bai Shao,白芍)-Jujubae Fructus(Da Zao,大枣),and Zingiberis Rhizoma Recens(Sheng Jiang,生姜)-Ginseng Radix Et Rhizoma(Ren Shen,人参),were not defined traditionally,but in connection with commonly used herbs.Classical formulae took the core herbs as principles,focusing on tonifying deficiency,strengthening the spleen and the stomach,strengthening the healthy Qi,and eliminating pathogenic factors.The compatibility pattern of properties involved was mainly acrid and sweet,which reflected the compatibility laws of benefiting Qi and tonifying Yang,replenishing Qi and nourishing blood,etc.Conclusions The research of classical formulae provides common understanding of some basic rules that have been adopted to tackle common illnesses/diseases using herbal medicine.The results help to reinforce theoretical understanding and development of traditional Chinese medicine(TCM),and revealing the hidden rules of combination in TCM data.Analyzing wider data samples of various herbal combinations through computation and big data technology can further optimize the use of TCM.

Thoughts About Paths for Digitization of Traditional Chinese Medicine-Refined Data Elements

For digitalization of traditional Chinese medicine(TCM),research is being conducted on objectivization of diagnosis and treatment,mathematical models of TCM theories,and application of modern information technology to digitize the vast amounts of existing information.However,the author believes that TCM practitioners should first conduct a systematic and comprehensive refined analysis on the knowledge of TCM and unify data elements used in computer intelligence to avoid ambiguity.Thus,we must overcome the epistemological constraints and carefully analyze the relationship among data elements to achieve systematic results and administer TCM appropriately.

Design of a WeChat Learning Platform for Syndrome Differentiation

Objective To design a WeChat mini program called Chinese Syndrome Differentiation Learning Platform(CSDLP)on smartphone to improve health literacy.Methods The developer tools of WeChat(Version:v1.01.170925)were used for designing and debugging the mini program.SPSS17.0 was used for statistical purposes.“View container”“Basic content”“Form component”“Navigation”and“Media components”were used for the development of the WeChat mini program.The detailed method was referred to https://mp.weixin.qq.com/debug/wxadoc/dev/.Results A WeChat mini program called CSDLP was developed.This program has three major functions which are WeChat reading,WeChat class and WeChat syndrome differentiation.The official test report showed that there were no functionality errors for the seven android smartphones(referred to as A,B,C,D,E,F and G)that CSDLP was tested on.Statistical analysis results showed that the average memory in D,E,F and G was lower than in A,B and C.Average ratio was the highest in F and the lowest in G.The average loading time was the same for all smartphones.The audio database for diagnostics using traditional Chinese medicine(TCM)and a lecture video database were based on diagnostic textbook.Our team built a syndrome differentiation database which included 51 diseases.Conclusion CSDLP can improve knowledge visualization,studying process,and information sharing in terms of the training and development of new techniques for syndrome differentiation and treatment in TCM,and it can provide a better illustration for people to understand TCM.