Analysis on acupoints selection and combination for functional dyspepsia: Data mining of randomized controlled trials ✩

Abstract

Background: The aim of the study was to explore the regularity of acupoints combination in the treatment of functional dyspepsia (FD) by means of data mining technology. Methods: We conducted a systematic search of nine databases from their inception to December 2022 to identify eligible randomized controlled trials (RCTs). The acupoints prescriptions in the treatment of FD were extracted and analyzed based on association rule mining and network analysis. Furthermore, the acupoints prescribed for irritability symptom of FD were quantified by forward inference, reverse inference and Bayes factor (BF) correction. Results: A total of 45 RCTs on 30 acupoints were included, involving 172 times the total application frequency. Analysis of acupoints prescriptions covered 29 meridian acupoints and one extra-ordinary point, disturbing in 10 meridians. The top three acupoints in application frequency were Zusanli (ST36, 19.77%), Neiguan (PC6, 16.86%), Zhongwan (CV12, 12.21%). The most involved meridian was the foot yangming stomach meridian (34 times). The most frequently used acupoints combinations were PC6 and ST36. The core prescriptions were ST36, PC6, Tianshu(ST25), CV12. The specificities of acupoints selection for FD with dysphoria as follows: using the forward inference, PC6, Taichong(LR3), Qimen(LR14), CV12, Danzhong(CV17), ST25, ST36 were selected. Using the reverse inference and BF correction, PC6, CV12, CV17, Sanyinjiao(SP6), ST25, ST36 were selected. Conclusion: ST36, PC6, ST25, CV12 could be used as the main prescriptions in treating FD. For FD with dysphoria the acupoints prescription is PC6, CV12, CV17, ST25 and ST36.

Headnote

ARTICLE INFO

Article history:

Received 5 November 2023

Revised 21 January 2024

Accepted 6 April 2024

Available online 7 June 2023

Keywords:

Functional dyspepsia

Acupoints indication

Acupoints specificity

Data mining

ABSTRACT

Background: The aim of the study was to explore the regularity of acupoints combination in the treatment of functional dyspepsia (FD) by means of data mining technology.

Methods: We conducted a systematic search of nine databases from their inception to December 2022 to identify eligible randomized controlled trials (RCTs). The acupoints prescriptions in the treatment of FD were extracted and analyzed based on association rule mining and network analysis. Furthermore, the acupoints prescribed for irritability symptom of FD were quantified by forward inference, reverse inference and Bayes factor (BF) correction.

Results: A total of 45 RCTs on 30 acupoints were included, involving 172 times the total application frequency. Analysis of acupoints prescriptions covered 29 meridian acupoints and one extra-ordinary point, disturbing in 10 meridians. The top three acupoints in application frequency were Zusanli (ST36, 19.77%), Neiguan (PC6, 16.86%), Zhongwan (CV12, 12.21%). The most involved meridian was the foot yangming stomach meridian (34 times). The most frequently used acupoints combinations were PC6 and ST36. The core prescriptions were ST36, PC6, Tianshu(ST25), CV12. The specificities of acupoints selection for FD with dysphoria as follows: using the forward inference, PC6, Taichong(LR3), Qimen(LR14), CV12, Danzhong(CV17), ST25, ST36 were selected. Using the reverse inference and BF correction, PC6, CV12, CV17, Sanyinjiao(SP6), ST25, ST36 were selected.

Conclusion: ST36, PC6, ST25, CV12 could be used as the main prescriptions in treating FD. For FD with dysphoria the acupoints prescription is PC6, CV12, CV17, ST25 and ST36.

1. Introduction

Functional dyspepsia (FD), affects more than 20% of the general population [1] and is one of the most common medical gastrointestinal conditions in the absence of an organic cause [2]. FD is characterized by the presence of one or more of the specific symptoms: epigastric pain, epigastric burning, postprandial fullness, and early satiation [3]. It has been reported that approximately 10% to 15% of patients with FD worldwide report symptoms of anxiety or depression, leading to worse quality of life, although FD is not associated with increased mortality [4].

Because of the complex pathophysiology, the exact mechanism of FD is still not well understood. Several related mechanisms are thought to contribute to this condition. Some of them include delayed gastric emptying, gastroesophageal reflux, and visceral hypersensitivity alterations in the nervous system. Others included gastroduodenal inflammation, altered sensitivity to duodenal acid or lipids, and impaired barrier function [5]. Currently, FD treatments for symptom management include H2 receptor antagonists (H₂RAs), proton pump inhibitors (PPIs), prokinetic agents, and even antidepressants [6]. However, the effectiveness and safety of these drugs remain controversial [7], Food and Drug Administration (FDA) have issued a warning about the cardiac safety of domperidone, uncommon adverse effects include rebound acid hypersecretion syndrome, fragility fractures, interstitial nephritis, electrolyte derangements, pneumonia, enteric infection and vitamin B12 deficiency.

Acupuncture, a traditional Chinese therapy, effectively improved the recovery of FD related symptoms with fewer side effects, as reported by several systematic reviews [8-11]. Ho RST, et al. concluded that manual acupuncture combined with clebopride has the highest probability of being the most effective treatment for FD related symptoms. Patients who are contraindicated for prokinetics may use manual acupuncture or moxibustion as alternatives [12]. Zhang J, et al. concluded that manual acupuncture alone was the most effective therapy for FD [13]. Acupoints are one of the factors determining the effectiveness of acupuncture therapy in clinical practice [14], and it is essential to identify which acupoints are specifically associated with the curative effect on FD. However, there is still a lack of analyses of the specificity of prescribed acupoints for FD, especially a lack of analyses of the specificities of acupoints selection based on specific symptoms for FD such as dysphoria.

Therefore, this study aimed to identify the specificities of acupoint selection and acupoints combination of FD especially the FD with dysphoria, by using association ruling mining, network analysis, forward inference, reverse inference and Bayes factor (BF) correction to optimize the acupoints combination and improve the therapeutic effect of acupuncture for treating FD.

2. Methods

2.1. Data sources and search strategy

We searched nine electronic databases from inception to December 2022, including five English databases (the Cochrane Library, PubMed, Embase, ScienceDirect, Web of Science) and four Chinese databases (the China National Knowledge Infrastructure [CNKI] database, Chinese Science and Technique Journals [CQVIP], Chinese Biomedical Database [CBM] and Wanfang). Only English or Chinese language literature was selected. In addition, we also read the references of the included papers to search for additional eligible studies. The search strategies were as follows: ("dyspepsia" OR "functional dyspepsia" OR "indigestion" OR "indigestions") AND ("acupuncture" OR "electroacupuncture" OR "transcutaneous electric nerve stimulation" OR "moxibustion" OR "acupoint application") AND ("randomized controlled trial" OR "controlled clinical trial" OR "randomly" OR "trial").

2.2. Inclusion criteria for studies

Patients older than 18 years with FD were included in this study regardless of gender or course of the disease. The diagnostic criteria for FD were based on the Rome II, III, IV or Chinese diagnostic criteria. Patients with any other complications, metabolic disease, or psychiatric sickness were excluded.

In this study, we included four specific acupuncture therapies that were defined as single use: manual acupuncture, electroacupuncture (EA), moxibustion, and transcutaneous electrical acupoint stimulation (TEAS) [15], as shown in Table 1.

Only randomized controlled trials (RCTs) were included.

Eligible included studies reported at least one of the following outcomes based on expert recommendations and the Cochrane Library Guide [16]: (1) an improvement in global or individual dyspepsia symptom scores of FD; (2) the Nepean Dyspepsia Index (NDI); (3) the Functional Digestive Disorder Quality of Life questionnaire (FDDQL); (4) the Satisfaction with Dyspepsia Related Health scale (SODA); (5) Digestive Health Status Instrument (DHSI); (6) the Medical Outcome Study Short Form Health Survey (SF36); (7) the Self-Rating Depression Scale (SDS), Self-Rating Anxiety Scale (SAS).

2.3. Exclusion criteria for studies

The control group was treated with prokinetics except cisapride, which has been removed from the market due to serious adverse effects [17]. Combined therapy with prokinetics and PPIs or H2RAs was also excluded as a substantial side effect [18-20].

2.4. Data selection and extraction

All identified studies were imported into EndNote X8. Duplicate studies and studies that did not meet the inclusion criteria were excluded after reading the titles and abstracts. The final inclusion or exclusion of studies was performed after fulltext reviews of the eligible studies. Data were independently extracted by two authors (Yu-qiao CHEN, Jia-li WU). Any inconsistency in the data extraction was resolved by a third author (Jun-fan FANG). The following information was extracted from each eligible article: author's name, year of publication, country, multicenter clinical study, diagnostic criteria, patient admission time, number of participants, gender and age, course of diseases, style of acupuncture, name of acupuncture points used (basic point, acupoints based on syndrome of traditional Chinese medicine [TCM]), frequency and duration of acupuncture sessions, control group, dosage and duration of control patients received and outcome indicators. The extracted data from eligible articles were imported into SPSS 19.0 and SPSS modeler 14.1 for further analysis.

2.5. Quality assessment

Two reviewers (Meng-ting QIU, Nai-xuan WEI) assessed and summarized the risk of bias of the included literature using the Cochrane collaboration tool for assessing the risk of bias and Review Manager 5.3, respectively. The details were as follows: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other bias. Each study was evaluated as high risk, unclear risk or low risk. A third reviewer (Jun-ying DU) was consulted to resolve any disagreements.

First, we counted the frequency of acupoints and meridians for FD. Acupoints and meridians that were used for FD with irritability symptom were extracted for further analysis.

Second, we summarized the commonly used acupoint groups (two or three acupoints combinations) by association rule mining analysis. The specific methods were used as previously described [21]. The minimum rule confidence level was set as 85%, and the minimum conditional support was set to 15%. Then, we determined the correlation of different acupoints by network analysis [22]. We used Gephi 0.9.2 to perform network image visualization of the acupoint interaction.

Third, we summarized the relationship between irritability symptom and acupoints by forward inference, reverse inference, and B F correction [23]. In forward inference and reverse inference, probability values were Z-transformed to assess statistical significance (Z>1.96). X =P(AT)/P(A), P(AT) is the probability that an acupoint would be selected for a given vexation symptom. P(A) is the probability that a given vexation symptom would be selected for four symptoms. The /z is the mean of X. 8 is standard deviation of X. n is the number of samples. In BF correction, BF>3 indicated substantial evidence in favor of the inference.

3. Results

3.1. Eligible studies

The research process is shown in Fig. 1. First, we identified 6 354 related citations through an initial search. After removing duplicated articles, we identified 2 269 potential studies. Then, 2 026 studies were deleted after screening the titles and abstracts, leaving only 243 potentially eligible studies. Finally, 45 studies were included after screening the full texts.

3.2. Risk of bias assessment

All 45 RCTs had a low risk of bias in random sequence generation. In terms of allocation concealment, 11 trials used central randomization, and 1 trial adopted an opaque envelope, both of which had low risk. In terms of blinding of participants and personnel, 8 trials had low risk with adopting blinding of investigators and patients (participants and personnel are separated), which was not easy to destroy. In terms of blinding of the outcome assessment, 9 trials blinded the outcome evaluators, and these trials had low risk. A total of 3 trials did not blind the outcome assessment, and these trials were scored as high risk. Most trials had a low risk of bias in incomplete outcome data and selective reporting. A total of 44 RCTs had unclear risk of bias in other biases. The risk of bias assessment is summarized in Figs. 2 and 3.

3.3. Characteristics of included RCTs

3.3.1. Participants

A total of 45 RCTs included a total of 4 040 FD patients. The age of the participants ranged between 18 and 70 years. Average sample size of the RCTs was 90. Duration of disease ranged from 1 to 10 years.

3.3.2. Diagnostic criteria

Among all these studies, 7 trials applied the Rome II criteria, 28 applied Rome III and 10 applied Rome IV.

3.3.3. Intervention

Five different forms of acupuncture and related therapies were evaluated: manual acupuncture plus moxibustion (6 comparisons), manual acupuncture (22 comparisons), EA (12 comparisons), moxibustion (3 comparisons), and TEAS (2 comparison). A total of 33 out of the 45 RCTs applied basic acupoints (based on unified acupoints formula), and 8 trials only applied acupoints based on one TCM syndromes. The frequency and duration of acupuncture, control, dosage and duration of control patients received and outcomes are shown in Table 2.

3.4. Frequency statistics of acupoints and meridians for FD

A total of 30 acupoints were included, with a total frequency of 172 times (Table 3). Among these acupoints, 29 were acupoints belonged to regular meridians and vessies, and one was an extraordinary acupoint. The top 6 acupoints were Zusanli (ST36), Neiguan (PC6), Zhongwan (CV12), Tianshu (ST25), Taichong (LR3), and Gongsun (SP4). A total of 10 meridians were involved in the treatment of FD (Table 4). Among these meridians, the most frequent meridian is the foot yangming stomach meridian, which is used for 27 times and covers 5 acupoints of this meridian.

3.5. Association rules and network of acupoints for FD

The association rules of acupoints for FD was shown in Table 5. There were 20 acupoints combinations with confidence levels above 85%, 10 of which were 100%, such as ST25-ST36, ST25-CV12-ST36, LR3-ST36, LR3-CV12-PC6, LR3-PC6-ST36, ST25PC6-ST36, SP4-CV12-PC6, and SP4-ST36-PC6. The acupoints combinations of PC6 and ST36 had the highest frequency and highest support, with 22 times and 75.86%, respectively. The highest support of the three acupoints combinations was CV12-PC6-ST36, with 48.276%. The support of acupoints combinations for FD was shown in Fig. 4. These results were consistent with the results in the cooccurrence matrix of acupoints (Fig. 5).

After association rule mining, we finally exported 30 node acupoints and 148 edge weights. Therefore, we constructed a com- phoria by forward inference. The Z-scores illustrated the frequency of each acupoints compared to the overall acupoints selection pattern for that syndrome. As shown in Fig. 8, the following acupoints had a high Z-score PC6 (Z=5.15), LR3 (Z=7.74), LR14 (Z=3.87), CV12 (Z=11.61 ), CV17 (Z=2.58), ST25 (Z=11.61) and ST36 (Z=9.03).

It was listed in Fig. 9. that the FD with dysphoria, their overall trial frequencies of their Z-scores greater than 1.96 for the 25 acupoints by reverse inference. The Z-scores illustrate the frequency of acupoints treatment compared to the indications for that acupoints. As shown in Fig. 9, the Z-scores of dysphoria for PC6, LR2, LR3, LR14, CV4, CV12, CV17, SP6, ST25, ST36, and TE6 were 3.00, 3.46, 3.46, 3.46, 3.46, 3.46, 2.73, 3.39, 2.00, 2.67, 2.90, and 3.46, respectively.

Furthermore, we used BF correction, which is more specific to each TCM syndrome, to analyze the specificity of acupoint indications. BF>3 was depicted on a heat map. As shown in Fig. 10, the BF of FD with dysphoria for PC6, CV12, CV17, SP6, ST25, and ST36 were 3.50, 3.43, 10.01, 4.01, 3.64, and 3.70, respectively.

4. Discussion

4.1. Summarize key finds

In this study, we presented a novel application of data mining that uses association rule mining and network analysis to indicate the acupoints combination of FD and forward and reverse inference and BF correction to directly examine the specificity relationship between acupoints selection and FD symptoms.

It has been reported that acupuncture therapy is an effective alternative method to treat FD, but acupoints prescriptions arc still lacking. We used association rule mining and network analysis to analyze the acupoints rules. The top 6 acupoints were ST36, PC6, CV12, ST25, LR3 and SP4, which correspond the foot yangming stomach meridian, the hand jueyin pericardium meridian, conception vessel, the foot yangming stomach meridian, the foot jueyin liver meridian, and the foot taiyin spleen meridian, respectively. The disease is located in the stomach, ST36 is the sea point of the foot yangming stomach meridian and the lower sea point of the stomach, which can "combined to treat the internal organs" and regulate the stomach qi. PC6 is the /uo-connecting point of the hand jueyin pericardium meridian, and it is also the confluence points of the eight vessels, which runs through the yinwei meridian. Stimulation of PC6 may smooth the qi movement of triple energizer, harmonizing stomach and lower adverse qi. The indication lies in the place where the acupoints are located. CV12 is the frontmu point of stomach, which can strengthen the spleen and regulate the stomach. When matching with ST36, it can regulate the fu qi. ST25 is the front-mu point of large intestine, and located around the umbilical. It can regulate the stomach, regulate qi and invigorate the spleen, have function of clear the intestine, and separate clear and excrete turbid. LR3 is the shu-acupoint and yuanprimary acupoint of the foot jueyin liver meridian. The foot jueyin liver meridian reaches the lower abdomen and the perigastric region. The curative effect of treatment lies in the place where the meridians pass. SP4 is the foot taiyin spleen meridian point, and it is also eight confluence point, which runs through the chong vessel. Diseases of chong vessel are characterized by circulation of vital energy in the wrong direction and abdominal pain. It can regulate the spleen and stomach, descend qi and relieve pain, and match with PC6, specifically for the treatment of stomach, heart, chest diseases.

4.2. Differences from other studies

Different acupoints combinations were applied to treat the same disease when the patients had different clinical symptoms in the clinic, but few studies have discussed them. However, several studies have identified that acupoints selection should be based on symptoms [24-27]. These findings did not directly reveal the specificity of acupoints indications, especially acupuncture for FD treatment. Previous studies have determined that forward and reverse inference can link associations between acupoints selection and symptoms directly, and the causal relationship without possible logical errors [28,29]. In this study, we used forward inference, reverse inference, and BF correction to analyze the direct relationship between acupoints and FD syndrome. Our results suggested that PC6, CV12, and ST36 have high specificity for any clinical symptoms of FD by forward inference. Because the total number of trials could influence the specificity of acupoints selection, we used BF correction to avoid possible fallacy. It is well known that using the method of BF correction could avoid fallacies and logical errors in reverse and forward inference analyses [28-31], which help define the specificities of acupoints indications by correcting the odds of the prior probability or the odds of the diseases presented in the data. The results suggested that irritability symptom was a specific indication for acupoints PC6, CV12, CV17, SP6, ST25, and ST36.

4.3. Limitations

It is easy to ignore the application changes of low-frequency acupoints while analyzing the law of selecting high-frequency acupoints, which still needs to be supplemented and improved continuously [32]. FD patients have various clinical manifestations, such as epigastric pain, epigastric distension, early satiety, warm breath, loss of appetite, nausea and vomiting, etc. Patients with the different symptoms need different treatments [33]. The number of multicenter clinical studies included in this study is low, and the description of blind method and follow-up is not specific, which makes the article have certain limitations.

4.4. Inspiration and hints for future research

In future research, the sample size should be broadening, more articles with clear and reliable data sources should be screened, and the number of included articles should be increased to reduce the limitations of the articles. What is more, the research on acupuncture and moxibustion treatment schemes with different clinical manifestations of FD can be added to strengthen the pertinence of clinical treatment of FD.

5. Conclusion

ST36, PC6, ST25, CV12 could be used as the main prescriptions in treating FD. For irritability symptom of FD, the acupoints prescription is PC6, CV12, CV17, ST25 and ST36. Our paper provides that acupoints prescriptions and specificity of acupoints selection with irritability symptom can be inferred from clinical data for treating the FD, which will promote to establish a better acupuncture treatment prescription.

Acknowledgments

We appreciate the great help from the Public Platform of Medical Research Center, Academy of Chinese Medical Science, Zhejiang Chinese Medical University.

CRediT authorship contribution statement

Yuqiao CHEN, Jiali WU and Junfan FANG: Writing-original draft of the manuscript, Reviewing the eligible studies. Mengting QIU, NaiXuan WEI, Zi GUO and Junying DU: Formal analysis, Data curation, Visualization. Xiaomei SHAO: Supervision. Jianqiao FANG: Methodology, Project administration, Resources.

Role of the funder/sponsor

The study sponsor Zhejiang Provincial Natural Science Fund of China, Zhejiang Medical and Health Science and Technology Program, National Natural Science Fund of China had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Declaration of Competing Interest

The authors declare that there is no conflicts of interest regarding the publication of this artele.

Data sharing statement

You can contact the corresponding author for the data.

State of human and animal rights

No need to declare in this paper.

Footnote

Supported by Zhejiang Provincial Natural Science Fund of China: LY20H270006, LY19H270003; Zhejiang Medical and Health Science and Technology Program: 2021RC098; National Natural Science Fund of China: 81873360.

* Corresponding authors.

E-mail addresses: [email protected] (J.-f. FANG), [email protected] (J.-y. DU).

References

References

[1] El-Serag HB, Talley NJ. Systemic review: the prevalence and clinical course of functional dyspepsia. Aliment Pharmacol Ther 2004;19:643-54.

[2] Ford AC, Mahadeva S, Carbone MF, Lacy BE, Talley NJ. Functional dyspepsia. Lancet 2020;396:1689-702.

[3] Enck P, Azpiroz F, Boeckxstaens G, Elsenbruch S, Feinle-Bisset C, Holtmann G, et al. Functional dyspepsia. Nat Rev Dis Primers 2017;3:17081.

[4] Madisch A, Andresen V, Enck P, Labenz J, Frieling T, Schemann M. The diagnosis and treatment of functional dyspepsia. Dtsch Arztebl Int 2018;115:222-32.

[5] Ye Y, Wang XR, Zheng Y, Yang JW, Yang NN, Shi GX, et al. Choosing an animal model for the study of functional dyspepsia. Can J Gastroenterol Hepatol 2018.

[6] Wauters L, Talley NJ, Walker MM, Tack J, Vanuytsel T. Novel concepts in the pathophysiology and treatment of functional dyspepsia. Gut 2020;69:591-600.

[7] Doggrell SA, Hancox JC. Cardiac safety concerns for domperidone, an antiemetic and prokinetic, and galactogogue medicine. Expert Opin Drug Saf 2014;13:131-8.

[8] Lan L, Zeng F, Liu GJ, Ying L, Wu X, Liu ML, et al. Acupuncture for functional dyspepsia. Cochrane Database Syst Rev 2014;13(10):CD008487.

[9] Pang B, Jiang T, Du YH, Li J, Bo Li, Hu YC, et al. Acupuncture for functional dyspepsia: what strength does it have? a systematic review and meta-analysis of randomized controlled Trials. Evid Based Complement Alternat Med 2016:2016:3862916.

[10] Masuy I, Van Oudenhove L, Tack J. Review article: treatment options for functional dyspepsia. Aliment Pharmacol Ther 2019;49:1134-72.

[11] Guo Y, Wei W, Chen JD. Effects and mechanisms of acupuncture and electroacupuncture for functional dyspepsia: a systematic review. World J Gastroenterol 2020;26:2440-57.

[12] Ho RST, Chung VCH, Wong CHL, Wu JCY, Wong SYS, Wu IXY. Acupuncture and related therapies used as add-on or alternative to prokinetics for functional dyspepsia: overview of systematic reviews and network meta-analysis. Sci Rep 2017;7:10320.

[13] Zhang J, Liu Y, Huang X, Chen Y, Hu L, Lan K, et al. Efficacy comparison of different acupuncture treatments for functional dyspepsia: a systematic review with network meta-analysis. Evid Based Complement Alternat Med 2020:2020:3872919.

[14] Hwang YC, Lee YS, Ryu Y, Lee IS, Chae Y. Statistical inference of acupoint specificity: forward and reverse inference. Integr Med Res 2020;9:1720.

[15] Lao L. Acupuncture techniques and devices. J Altern Complement Med 1996;2:23-5.

[16] Wheeler DC, Archer KJ, Burstyn I, Yu K, Stewart PA, Colt JS, et al. Comparison of ordinal and nominal classification trees to predict ordinal expert-based occupational exposure estimates in a case-control study. Ann Occup Hyg 2015;59:324-35.

[17] Ang D, Talley NJ, Simren M, Janssen P, Boeckxstaens G, Tack J. Review article: endpoints used in functional dyspepsia drug therapy trials. Aliment Pharmacol Ther 2011;33:634-49.

[18] Ren LH, Chen WX, Qian LJ, Li S, Gu M, Shi RH. Addition of prokinetics to PPI therapy in gastroesophageal reflux disease: a meta-analysis. World J Gastroenterol 2014;20:2412-19.

[19] Lam JR, Schneider JL, Zhao W, Corley DA. Proton pump inhibitor and histamine 2 receptor antagonist use and vitamin B12 deficiency. JAMA 2013;310:2435-42.

[20] Hanlon JT, Landerman LR, Artz MB, Gray SL, Fillenbaum GG, Schmader KE. Histamine2 receptor antagonist use and decline in cognitive function among community dwelling elderly. Pharmacoepidemiol Drug Saf 2004;13:781-7.

[21] Wang JY, Liu CS, Lung CH, Yang YT, Lin MH. Investigating spousal concordance of diabetes through statistical analysis and data mining. PLoS One 2017;12(8):e0183413.

[22] Wu HH, Wen J, Han P, Xing HJ, Zhang XQ Jia CS. The exploration on the characteristics of acupoints selection and the rule of acupoint combination in treatment of insomnia with acupuncture and moxibustion: complex network analysis. World J Acupunct Moxibust 2022;32(01):61-9.

[23] Hwang YC, Lee IS, Ryu Y, Lee YS, Chae Y. Identification of acupoint indication from reverse inference: data mining of randomized controlled clinical trials. J Clin Med 2020;9:3027.

[24] Li DD, Yue ZH, Xu LC, Xie T, Hu GZ, Yang J. Clinical evalutaion study on long-term effect of acupuncture with pattern/syndrome differentiation on functional dyspepsia. Chin Acupunct Moxibust 2014;34(5):431-4.

[25] Kim EJ, Lim CY, Lee EY, Lee SD, Kim KS. Comparing the effects of individualized, standard, sham and no acupuncture in the treatment of knee osteoarthritis: a multicenter randomized controlled trial. Trials 2013; 14:129.

[26] Hou LW, Fang JL, Zhang JL, Wu D, Wang JY, Rong PJ. Effects of transcutaneous auricular vagus nerve stimulation on intestinal ligandins in a rat model of functional dyspepsia. World J Acupunct Moxibust 2022;32(01):33-9.

[27] Zhou W, Li YX, Zhang YX, Xue RD, Liu Y, Cao LJ, et al. Acupuncture based on meridian diagnosis for chronic atrophic gastritis. Chin Acupunct Moxibust 2020;40(9):928-32.

[28] Henson R. Forward inference using functional neuroimaging: dissociations versus associations. Trends Cogn Sei 2006;10:64-9.

[29] Hutzler F. Reverse inference is not a fallacy per se: cognitive processes can be inferred from functional imaging data. Neuroimage 2014;84:1061-9.

[30] Heit E. Brain imaging, forward inference, and theories of reasoning. Front Hum Neurosci 2015;8:1056.

[31] Poldrack RA. Can cognitive processes be inferred from neuroimaging data? Trends Cogn Sci 2006;10(2):59-63.

[32] XI MD, Gao XF. Based on data mining to explore the clinical medication rules of traditional Chinese medicine in the treatment of hypertensive renal injury. J Zhejiang Chin Med Univ 2023;47(l):84-90.

[33] Zhou TY, Qian FH, Huang XY, Gu ME, An XF, Zheng X, et al. Study on the rules of traditional Chinese medicine and acupuncture diagnosis and treatment of acute gastrointestinal injury in critically patients based on data mining. Modern Tradit Chin Med Materia Mater 2019;21(8):1606-14.