Please wait a minute...
Frontiers of Medicine

ISSN 2095-0217

ISSN 2095-0225(Online)

CN 11-5983/R

Postal Subscription Code 80-967

2018 Impact Factor: 1.847

Front. Med.    2019, Vol. 13 Issue (1) : 112-120    https://doi.org/10.1007/s11684-017-0582-z
RESEARCH ARTICLE
Characterization of hidden rules linking symptoms and selection of acupoint using an artificial neural network model
Won-Mo Jung1, In-Soo Park2, Ye-Seul Lee1,2, Chang-Eop Kim3, Hyangsook Lee1,2, Dae-Hyun Hahm4, Hi-Joon Park1,2, Bo-Hyoung Jang5, Younbyoung Chae1,2()
1. Department of Science in Korean Medicine, Graduate School, Kyung Hee University, Seoul 130-701, Republic of Korea
2. Acupuncture & Meridian Science Research Center, College of Korean Medicine, Kyung Hee University, Seoul 130-701, Republic of Korea
3. Department of Physiology, College of Korean Medicine, Gachon University, Seoul 131-120, Republic of Korea
4. Department of Physiology, School of Medicine, Kyung Hee University, Seoul 130-701, Republic of Korea
5. Department of Preventive Medicine, College of Korean Medicine, Kyung Hee University, Seoul 130-701, Republic of Korea
 Download: PDF(2160 KB)   HTML
 Export: BibTeX | EndNote | Reference Manager | ProCite | RefWorks
Abstract

Comprehension of the medical diagnoses of doctors and treatment of diseases is important to understand the underlying principle in selecting appropriate acupoints. The pattern recognition process that pertains to symptoms and diseases and informs acupuncture treatment in a clinical setting was explored. A total of 232 clinical records were collected using a Charting Language program. The relationship between symptom information and selected acupoints was trained using an artificial neural network (ANN). A total of 11 hidden nodes with the highest average precision score were selected through a tenfold cross-validation. Our ANN model could predict the selected acupoints based on symptom and disease information with an average precision score of 0.865 (precision, 0.911; recall, 0.811). This model is a useful tool for diagnostic classification or pattern recognition and for the prediction and modeling of acupuncture treatment based on clinical data obtained in a real-world setting. The relationship between symptoms and selected acupoints could be systematically characterized through knowledge discovery processes, such as pattern identification.

Keywords acupuncture      indication      neural network      pattern identification      prediction     
Corresponding Author(s): Younbyoung Chae   
Just Accepted Date: 27 December 2017   Online First Date: 12 April 2018    Issue Date: 12 March 2019
 Cite this article:   
Won-Mo Jung,In-Soo Park,Ye-Seul Lee, et al. Characterization of hidden rules linking symptoms and selection of acupoint using an artificial neural network model[J]. Front. Med., 2019, 13(1): 112-120.
 URL:  
https://academic.hep.com.cn/fmd/EN/10.1007/s11684-017-0582-z
https://academic.hep.com.cn/fmd/EN/Y2019/V13/I1/112
Fig.1  (A) “Charting Language” program. This system uses specific syntax and medical terms unconstrained by the structure inherent to a defined language. A total of 232 clinical records were collected for neural network analysis. (B) ANN model. A feed-forward network that consists of input, hidden, and output layers was employed as an ANN model to learn the acupuncture treatment patterns in Korean medicine. A total of 87 nodes in the input layer corresponded to 87 symptoms present among the medical records (layer with blue circles). The number of hidden nodes was 11 based on the highly predictive performance achieved through a tenfold cross validation (layer with green circles). A total of 77 nodes in the output layer correspond to 77 acupoints identified from the medical records (layer with red circles).
No. Most frequently observed symptoms Most frequently used acupoints
1 Back pain (14.8%) LI11 (7.9%)
2 Shoulder pain (9.5%) LU8 (5.3%)
3 Strained muscle (4.5%) SP3 (4.8%)
4 Knee pain (4.1%) LR4 (4.5%)
5 Lower limb pain (4.1%) BL23 (4.4%)
6 Upper limb pain (3.7%) BL24 (4.4%)
7 Edema (3.6%) BL26 (4.3%)
8 Neck pain (3.6%) GB43 (4.3%)
9 Dyspepsia (2.9%) ST36 (4.1%)
10 Ankle pain (2.8%) KI10 (3.9%)
11 Finger joint pain (2.6%) LU9 (3.9%)
12 Scapula pain (2.6%) LR8 (3.8%)
13 Pressure pain on BL24 (2.1%) TE6 (2.5%)
14 Arthralgia (1.8%) TE15 (2.4%)
15 Foot pain (1.8%) LI1 (2.1%)
16 Stiff neck (1.8%) GB41 (2.0%)
17 Wrist pain (1.8%) SI14 (2.0%)
18 Insomnia (1.7%) SP9 (1.7%)
19 Headache (1.6%) ST43 (1.7%)
20 Pressure pain on BL26 (1.6%) LI5 (1.7%)
Tab.1  Most frequently observed symptoms and acupoints of patients in 232 clinical records
No. Most frequently used acupoint combinations
1 SI12–SI18 (n = 38)
2 SI12–LI4 (n = 38)
3 SI18–LI4 (n = 38)
4 SP10–LI5 (n = 38)
5 TE15–SP10 (n = 38)
6 TE15–LI5 (n = 38)
7 EX-LE4–SP10 (n = 37)
8 SI12–SI2 (n = 32)
9 SI12–SP10 (n = 31)
10 SI18–SI2 (n = 31)
11 SI2–LI4 (n = 31)
12 SI18–SP10 (n = 30)
13 SP10–LI4 (n = 30)
14 SI2–SP10 (n = 29)
15 EX-LE4–TE15 (n = 23)
16 EX-LE4–LI5 (n = 23)
17 BL23–SP10 (n = 22)
18 EX-LE4–SI2 (n = 20)
19 LR2–SP10 (n = 20)
20 LR2–BL23 (n = 20)
Tab.2  Most frequently used acupoint combinations in 232 clinical records
Fig.2  Network analysis of acupoint combinations. The collected data contained 232 acupoint prescriptions. Acupoint combination network is generated with nodes (acupoints) and edges (co-occurrences between two acupoints). Networkx 1.11.0 (Python library) is used to construct the network, and Gephi 0.8.2 is utilized to visualize the network. The investigation of the visual pattern of the network is facilitated by intuitive and comprehensive analyses. The acupoint combination network contained 52 acupoints. The color of the node represents eigenvector centrality (colors closer to red represent higher values). Edge thickness is proportional to the frequency of the co-occurrences among linked nodes.
Fig.3  Selected acupoints for 232 cases were represented in a color-coded matrix. (A) Estimated acupoints from the fully fitted ANN model (average precision score, 0.865; precision, 0.911; recall, 0.811). (B) Actual acupoint selections in 232 cases. Selected acupoints were coded in black in the two matrices. (C) Correct estimations (black), false positive errors (blue), and false negative errors (red).
Fig.4  Network visualization of relationships among symptoms, acupoints, and hidden nodes based on the correlation of activity patterns. The five symptoms and acupoints that show the strongest correlations with the activity patterns of the 11 hidden nodes were extracted. In total, 11 hidden nodes were associated with 22 symptoms and 26 acupoints (5 element acupoints are marked with a red “#.”). Line widths are proportional to the size of the correlation coefficients that describe the relationships among connected nodes.
1 JEDayhoff, JM DeLeo. Artificial neural networks: opening the black box. Cancer 2001; 91(8 Suppl): 1615–1635
https://doi.org/10.1002/1097-0142(20010415)91:8+<1615::AID-CNCR1175>3.0.CO;2-L pmid: 11309760
2 DMantzaris, G Anastassopoulos, AAdamopoulos. Genetic algorithm pruning of probabilistic neural networks in medical disease estimation. Neural Netw 2011; 24(8): 831–835
https://doi.org/10.1016/j.neunet.2011.06.003 pmid: 21723704
3 BHua, E Abbas, AHayes, PRyan, L Nelson, KO’Brien. Reliability of Chinese medicine diagnostic variables in the examination of patients with osteoarthritis of the knee. J Altern Complement Med 2012; 18(11): 1028–1037
https://doi.org/10.1089/acm.2011.0621 pmid: 22897413
4 CMChang, HT Chu, YHWei, FPChen, SWang, PC Wu, HRYen, TJChen, HHChang. The core pattern analysis on Chinese herbal medicine for Sjögren’s syndrome: a nationwide population-based study. Sci Rep 2015; 5(1): 9541
https://doi.org/10.1038/srep09541 pmid: 25923413
5 HKang, Y Zhao, CLi, YChen, K Tang, LYang, CMa, J Peng, RZhu, QLiu, Y Hu, ZCao. Integrating clinical indexes into four-diagnostic information contributes to the traditional Chinese medicine (TCM) syndrome diagnosis of chronic hepatitis B. Sci Rep 2015; 5(1): 9395
https://doi.org/10.1038/srep09395 pmid: 25797918
6 KAO’Brien, E Abbas, JZhang, ZXGuo, R Luo, ABensoussan, PAKomesaroff. An investigation into the reliability of Chinese medicine diagnosis according to Eight Guiding Principles and Zang-Fu Theory in Australians with hypercholesterolemia. J Altern Complement Med 2009; 15(3): 259–266
https://doi.org/10.1089/acm.2008.0204 pmid: 19249997
7 JFarquhar. Knowing Practice: the Clinical Encounter of Chinese Medicine. Boulder, Colo.: Westview, 1994
8 TJKaptchuk. The Web That Has No Weaver. Chicago, IL: Comtemporary Books, 2000
9 KJSherman, DC Cherkin, CJHogeboom. The diagnosis and treatment of patients with chronic low-back pain by traditional Chinese medical acupuncturists. J Altern Complement Med 2001; 7(6): 641–650
https://doi.org/10.1089/10755530152755199 pmid: 11822612
10 WMJung, T Lee, ISLee, SKim, H Jang, SYKim, HJPark, YChae. Spatial patterns of the indications of acupoints using data mining in classic medical text: a possible visualization of the meridian system. Evid Based Complement Alternat Med 2015; 2015: 457071
https://doi.org/10.1155/2015/457071 pmid: 26539224
11 LHa, T Li, FWang. Exploration and analysis on the “similar-indication acupoints”. Chin Acupunct Mox (Zhongguo Zhen Jiu) 2015; 35(12): 1263–1265 (in Chinese)
pmid: 26964173
12 VNapadow, J Liu, TJKaptchuk. A systematic study of acupuncture practice: acupoint usage in an outpatient setting in Beijing, China. Complement Ther Med 2004; 12(4): 209–216
https://doi.org/10.1016/j.ctim.2004.10.001 pmid: 15649834
13 YYWang, F Lin, ZLJiang. Pattern of acupoint selection based on complex network analysis technique. Chin Acupunct Mox (Zhongguo Zhen Jiu) 2011; 31(1): 85–88 (in Chinese)
pmid: 21355168
14 ANRamesh, C Kambhampati, JRTMonson, PJDrew. Artificial intelligence in medicine. Ann R Coll Surg Engl 2004; 86(5): 334–338
https://doi.org/10.1308/147870804290 pmid: 15333167
15 CCYang, P Veltri. Intelligent healthcare informatics in big data era. Artif Intell Med 2015; 65(2): 75–77
https://doi.org/10.1016/j.artmed.2015.08.002 pmid: 26306669
16 FRosenblatt. The perceptron: a probabilistic model for information storage and organization in the brain. Psychol Rev 1958; 65(6): 386–408
https://doi.org/10.1037/h0042519 pmid: 13602029
17 RDutta, J Aryal, ADas, JBKirkpatrick. Deep cognitive imaging systems enable estimation of continental-scale fire incidence from climate data. Sci Rep 2013; 3(1): 3188
https://doi.org/10.1038/srep03188 pmid: 24220174
18 ERezaei-Darzi, F Farzadfar, AHashemi-Meshkini, INavidi, MMahmoudi, MVarmaghani, PMehdipour, MSoudi Alamdari, BTayefi, SNaderimagham, FSoleymani, AMesdaghinia, ADelavari, KMohammad. Comparison of two data mining techniques in labeling diagnosis to Iranian pharmacy claim dataset: artificial neural network (ANN) versus decision tree model. Arch Iran Med 2014; 17(12): 837–843
pmid: 25481323
19 YJChang, ML Yeh, YCLi, CYHsu, CC Lin, MSHsu, WTChiu. Predicting hospital-acquired infections by scoring system with simple parameters. PLoS One 2011; 6(8): e23137
https://doi.org/10.1371/journal.pone.0023137 pmid: 21887234
20 HYShi, KT Lee, HHLee, WHHo, DP Sun, JJWang, CCChiu. Comparison of artificial neural network and logistic regression models for predicting in-hospital mortality after primary liver cancer surgery. PLoS One 2012; 7(4): e35781
https://doi.org/10.1371/journal.pone.0035781 pmid: 22563399
21 RFéraud, F Clérot. A methodology to explain neural network classification. Neural Netw 2002; 15(2): 237–246
https://doi.org/10.1016/S0893-6080(01)00127-7 pmid: 12022511
22 WMJung, Y Chae, BHJang. Development of markup language for medical record charting: a Charting Language. Stud Health Technol Inform 2015; 216: 879
pmid: 26262181
23 TSchaul, J Bayer, DWierstra, YSun, M Felder, FSehnke, TRuckstieß, JSchmidhuber. PyBrain. J Mach Learn Res 2010; 11: 743–746
24 DTouretzky, D Pomerleau. What’s hidden in the hidden layers. Byte 1989; 14(8): 227–233
25 BPandey, RB Mishra. Knowledge and intelligent computing system in medicine. Comput Biol Med 2009; 39(3): 215–230
https://doi.org/10.1016/j.compbiomed.2008.12.008 pmid: 19201398
26 PUnschuld. Medicine in China: a History of Ideas. Berkeley, Los Angeles and London: University of California Press, 1985
27 TLee, WM Jung, ISLee, YSLee, H Lee, HJPark, NKim, Y Chae. Data mining of acupoint characteristics from the classical medical text: DongUiBoGam of Korean Medicine. Evid Based Complement Alternat Med 2014; 2014: 329563
https://doi.org/10.1155/2014/329563 pmid: 25574179
28 NLZhang, S Yuan, TChen, YWang. Statistical validation of traditional Chinese medicine theories. J Altern Complement Med 2008; 14(5): 583–587
https://doi.org/10.1089/acm.2007.7019 pmid: 18554082
29 NLZhang, S Yuan, TChen, YWang. Latent tree models and diagnosis in traditional Chinese medicine. Artif Intell Med 2008; 42(3): 229–245
https://doi.org/10.1016/j.artmed.2007.10.004 pmid: 18096374
30 WFYeung, KF Chung, NLWZhang, SPZhang, KPYung, PXChen, YYHo. Identification of Chinese medicine syndromes in persistent insomnia associated with major depressive disorder: a latent tree analysis. Chin Med 2016; 11(1): 4
https://doi.org/10.1186/s13020-016-0076-y pmid: 26877762
31 BLiu, X Zhou, YWang, JHu, L He, RZhang, SChen, Y Guo. Data processing and analysis in real-world traditional Chinese medicine clinical data: challenges and approaches. Stat Med 2012; 31(7): 653–660
https://doi.org/10.1002/sim.4417 pmid: 22161304
32 XZhou, S Chen, BLiu, RZhang, YWang, P Li, YGuo, HZhang, ZGao, X Yan. Development of traditional Chinese medicine clinical data warehouse for medical knowledge discovery and decision support. Artif Intell Med 2010; 48(2-3): 139–152
https://doi.org/10.1016/j.artmed.2009.07.012 pmid: 20122820
33 HZhang, CH Tian, BYLiu, XZZhou, YHWang, ZSLiu, . Study of the idea of clinical acupuncture point combination of TCM physician Tian. J Clin Acup Mox (Zhen Jiu Lin Chuang Za Zhi) 2007; 23(2): 36–38 (in Chinese)
34 WMJung, SH Lee, YSLee, YChae. Exploring spatial patterns of acupoint indications from clinical data: A STROBE-compliant article. Medicine (Baltimore) 2017; 96(17): e6768
https://doi.org/10.1097/MD.0000000000006768 pmid: 28445309
35 PMamoshina, A Vieira, EPutin, AZhavoronkov. Applications of deep learning in biomedicine. Mol Pharm 2016; 13(5): 1445– 1454
https://doi.org/10.1021/acs.molpharmaceut.5b00982 pmid: 27007977
36 MLiang, , Z Li, TChen, JZeng. Integrative data analysis of multi-platform cancer data with a multimodal deep learning approach. IEEE/ACM Trans Comput Biol Bioinform, 2015. 12(4): 928–937
37 CHShi, XJ Wang, JXChen, RQLiu, YH Zhao, HJYang. Study on the drug selection law for treatment of chronic gastritis with spleen deficiency and stomach dryness by complex system entropy cluster. J Tradit Chin Med 2010; 30(4): 294–298
https://doi.org/10.1016/S0254-6272(10)60059-3 pmid: 21287788
38 ASFerreira, AJ Lopes. Chinese medicine pattern differentiation and its implications for clinical practice. Chin J Integr Med 2011; 17(11): 818–823
https://doi.org/10.1007/s11655-011-0892-y pmid: 22057410
39 CABerle, D Cobbin, NSmith, CZaslawski. A novel approach to evaluate traditional Chinese medicine treatment outcomes using pattern identification. J Altern Complement Med 2010; 16(4): 357–367
https://doi.org/10.1089/acm.2009.0367 pmid: 20374102
[1] Jiechao Ma, Yang Song, Xi Tian, Yiting Hua, Rongguo Zhang, Jianlin Wu. Survey on deep learning for pulmonary medical imaging[J]. Front. Med., 2020, 14(4): 450-469.
[2] Ke Sheng. Artificial intelligence in radiotherapy: a technological review[J]. Front. Med., 2020, 14(4): 431-449.
[3] Hujie Song, Xiao Chen, Yalan Yu, Ling Zhang. Xingnao Kaiqiao acupuncture combined with Angong Niuhuang Wan for a patient under persistent vegetative state: a case report[J]. Front. Med., 2018, 12(3): 334-339.
[4] Hong Zhao, Fengxia Liang, Yigong Fang, Baoyan Liu. Application of Grading of Recommendations Assessment, Development, and Evaluation (GRADE) to the guideline development for clinical practice with acupuncture and moxibustion[J]. Front. Med., 2017, 11(4): 590-594.
[5] Likun Yang,Yang Wang,Qian Mo,Zhishun Liu. A comparative study of electroacupuncture at Zhongliao (BL33) and other acupoints for overactive bladder symptoms[J]. Front. Med., 2017, 11(1): 129-136.
[6] Juanjuan Xin,Yangshuai Su,Zhaokun Yang,Wei He,Hong Shi,Xiaoyu Wang,Ling Hu,Xiaochun Yu,Xianghong Jing,Bing Zhu. Distinct roles of ASIC3 and TRPV1 receptors in electroacupuncture-induced segmental and systemic analgesia[J]. Front. Med., 2016, 10(4): 465-472.
[7] Jingjing Jiang,Ting Ding,Aiyue Luo,Yunping Lu,Ding Ma,Shixuan Wang. Comparison of surgical indications for hysterectomy by age and approach in 4653 Chinese women[J]. Front. Med., 2014, 8(4): 464-470.
[8] Xu Chen, Xiaomao Xu, Fei Xiao. Heterogeneity of chronic obstructive pulmonary disease: from phenotype to genotype[J]. Front Med, 2013, 7(4): 425-432.
[9] Peijing Rong, Bing Zhu, Yuqing Li, Xinyan Gao, Hui Ben, Yanhua Li, Liang Li, Wei He, Rupeng Liu, Lingling Yu. Mechanism of acupuncture regulating visceral sensation and mobility[J]. Front Med, 2011, 5(2): 151-156.
[10] Jisheng Han, Cailian Cui, Liuzhen Wu. Acupuncture-related techniques for the treatment of opiate addiction: a case of translational medicine[J]. Front Med, 2011, 5(2): 141-150.
[11] Xianze Meng, Shifen Xu, Lixing Lao. Clinical acupuncture research in the West[J]. Front Med, 2011, 5(2): 134-140.
[12] Cuihong ZHENG, Yunxia WU, Guangying HUANG, Wei WANG. Gene silencing efficiency of shRNA expression vectors targeting Cx43 in vitro[J]. Front Med Chin, 2009, 3(2): 130-135.
[13] QIN Lunxiu, TANG Zhaoyou, GUAN Xinyuan, YE Qinghai, JIA Huliang, REN Ning. The predictive value of chromosome 8p deletion for metastasis of hepatocellular carcinoma: a summary of works in 10 years[J]. Front. Med., 2008, 2(3): 211-215.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed