|Year : 2022 | Volume
| Issue : 4 | Page : 389-393
Bibliometric analyses of applications of artificial intelligence on tuberculosis
Miguel Cabanillas-Lazo1, Carlos Quispe-Vicuña1, Milagros Pascual-Guevara2, John Barja-Ore3, Maria Eugenia Guerrero4, Arnaldo Munive-Degregori4, Frank Mayta-Tovalino5
1 Department of Academic, Grupo Peruano De Investigación Epidemiológica, Unidad Para La Generación Y Síntesis De Evidencias En Salud, Universidad San Ignacio De Loyola; Department of Academic, Sociedad Científica De San Fernando, Lima, Perú
2 Department of Academic, Sociedad Científica De San Fernando; Department of Academic, Faculty of Medicine, Universidad Nacional Mayor De San Marcos, Lima, Perú
3 Department of Academic, Research Direction, Universidad Privada Del Norte, Lima, Perú
4 Department of Academic, Universidad Nacional Mayor De San Marcos, Lima, Perú
5 Department of Posgraduate, Vicerrectorado De Investigación, Universidad San Ignacio De Loyola, Lima, Perú
|Date of Submission||21-Aug-2022|
|Date of Decision||01-Sep-2022|
|Date of Acceptance||20-Oct-2022|
|Date of Web Publication||10-Dec-2022|
Universidad San Ignacio De Loyola, Av. La Fontana 550, La Molina 15024, La Molina
Source of Support: None, Conflict of Interest: None
Background: Tuberculosis is one of the leading causes of death worldwide affecting mainly low- and middle-income countries. Therefore, the objective is to analyze the bibliometric characteristics of the application of artificial intelligence (AI) in tuberculosis in Scopus. Methods: A bibliometric study, the Scopus database was used using a search strategy composed of controlled and free terms regarding tuberculosis and AI. The search fields “TITLE,” “ABSTRACT,” and “AUTHKEY” were used to find the terms. The collected data were analyzed with Scival software. Bibliometric data were described through the figures and tables summarized by absolute values and percentages. Results: Thousand and forty-one documents were collected and analyzed. Yudong Zhang was the author with the highest scientific production; however, K. C. Santosh had the greatest impact. Anna University (India) was the institution with the highest number of published papers. Most papers were published in the first quartile. The United States led the scientific production. Articles with international collaboration had the highest impact. Conclusion: Articles related to tuberculosis and AI are mostly published in first quartile journals, which would reflect the need and interest worldwide. Although countries with a high incidence of new cases of tuberculosis are among the most productive, those with the highest reported drug resistance need greater support and collaboration.
Keywords: Artificial intelligence, bibliometrics, machine learning, tuberculosis
|How to cite this article:|
Cabanillas-Lazo M, Quispe-Vicuña C, Pascual-Guevara M, Barja-Ore J, Guerrero ME, Munive-Degregori A, Mayta-Tovalino F. Bibliometric analyses of applications of artificial intelligence on tuberculosis. Int J Mycobacteriol 2022;11:389-93
|How to cite this URL:|
Cabanillas-Lazo M, Quispe-Vicuña C, Pascual-Guevara M, Barja-Ore J, Guerrero ME, Munive-Degregori A, Mayta-Tovalino F. Bibliometric analyses of applications of artificial intelligence on tuberculosis. Int J Mycobacteriol [serial online] 2022 [cited 2023 Feb 4];11:389-93. Available from: https://www.ijmyco.org/text.asp?2022/11/4/389/363152
| Introduction|| |
Considered by some as the worst epidemic of the 21st century, tuberculosis is the thirteenth leading cause of death and the most lethal infection after AIDS. It is caused by the airborne transmission of Mycobacterium tuberculosis (Mtb) which, despite being curable and preventable, 9.9 million cases were reported in 2020 with 98% of cases and deaths attributed to low- and middle-income countries (LMICs), reflecting the enormous economic and social differences in public health., Thus, its poor infrastructure, self-medication, drug resistance, poor adherence to treatment and professionals with limited training, project catastrophic expenses of 13 billion dollars annually to try to contain this epidemic, so alternative methods that achieve early and timely diagnosis and treatment such as artificial intelligence (AI) are necessary.
AI is a technological tool formed by a series of logical algorithms that simulate human intelligence processes (learning, reasoning, and self-correction) from which machines are able to assist in clinical decision-making and thereby lessen the burden on health care workers. Therefore, most AI studies have focused on contributing to diagnosis in health problems common to LMICs led by communicable diseases such as tuberculosis, malaria, and dengue based on machine learning and signal processing. In addition, some AI algorithms have also succeeded in identifying the concentrations of levofloxacin required for microbiological eradication of Mtb and suppression of multidrug-resistant tuberculosis.
However, the application of AI in LMIC public health still faces inadequate designs and developments, so it is necessary to make the use of bibliometric studies to obtain a quantitative synthesis and follow-up of scientific advances, as well as to compare results in quality and quantity with developed countries to plan the programs and formulate better health policies, In the last 5 years, the growth of bibliometric studies on AI has been exponential and has even increased tenfold, possibly due to the appearance of different techniques and computer advances, although publications are concentrated in only three developed countries such as the USA, China, and the United Kingdom.
Our results using bibliometric indicators would be useful to identify whether LMICs are beginning to integrate AI into decision-making for their most prevalent diseases such as tuberculosis and thereby improve the quality of their research. Therefore, the objective of the research was to analyze the bibliometric characteristics of the scientific production on the application of AI in tuberculosis in Scopus.
| Methods|| |
A descriptive, cross-sectional, bibliometric study, which consisted of evaluating the documents published on the application of AI in tuberculosis in the Scopus database up to June 18, 2022.
Data were acquired on June 18, 2022 from Elsevier's Scopus database (https://www.scopus.com/). There were no ethical issues as all information is public. Controlled MeSH and Emtree terms were used, as well as free terms related to AI and tuberculosis. The Boolean operators “OR” and “AND” were used to combine these terms to form a strategy. The search fields “TITLE,” “ABSTRACT,” and “AUTHKEY” were used to find the terms. To avoid the changes in citation rates, all searches were performed on the same day. Two authors independently developed strategies, then, by consensus, a final strategy was obtained [Supplementary Material 1].
We included all papers published during the period 2016–2022 in the scientific journals. The sample size of our study was the data generated during that period.
The bibliometric indicators analyzed were the following: institutions, countries, journal quartile, type of collaboration, field-weighted citation impact, source-Normalized Impact per Paper, SCImago Journal Rank, and CiteScore. In addition, authors with the highest scientific production, their h-index, the number of publications and citations, as well as the number of citations per published paper, to assess their impact.
After the identification of the documents, the metadata were analyzed with the Scival program of Elsevier (https://www.scival.com/). In addition, Microsoft Excel 2019 was used for the estimation of absolute values and percentages, as well as for the design of tables and graphs.
| Results|| |
A total of 1041 documents (47.6% were published in open access journals) and 4941 authors related to tuberculosis and AI were collected. The countries with the highest production were the United States (223 papers), India (216 papers), and China (174 papers). These countries had the highest citation (United States: 4293 citations; India: 1599 citations; China: 1739 citations) [Figure 1].
|Figure 1: Top ten productive countries on tuberculosis and artificial intelligence|
Click here to view
Yudong Zhang was the author with the highest number of publications (15 published papers), while K. C. Santosh had the highest impact with 36.3 citations per paper in his 8 publications. Sean Ekins and Xin Zhang were the second and third highest producing authors, respectively. Most of the authors were from the United States [Table 1].
|Table 1: Top ten authors publishing on tuberculosis and artificial intelligence|
Click here to view
The 10 institutions with the highest number of papers are shown in [Table 2]. Anna University (India) was the institution with the highest scientific output (30) while Harvard University (United States) had the highest impact (32.9 citations per paper). The University of Cape Town (South Africa) and the National Institutes of Health (United States) were the second and third institutions with the highest scientific production with 26 and 24 published papers, respectively.
|Table 2: Top ten productive institutions on tuberculosis and artificial intelligence|
Click here to view
With respect to the journals, CEUR Workshop Proceedings, Lecture Notes in Computer Science and Scientific Reports are the ones with the highest number of publications on the subject under the study with 49, 28, and 23 documents, respectively. However, the journal with the highest impact was Clinical Infectious Diseases with 22.9 citations per paper (13 publications) [Table 3].
|Table 3: Bibliometric indicators of production and impact on journals on tuberculosis and artificial intelligence|
Click here to view
In addition, [Figure 2] shows the number of documents according to the quartile of the journal. During the study period, a higher percentage is observed in the first quartile (374 documents). This proportion remained constant in all the years of the study.
|Figure 2: Documents published according CiteScore Quartil on tuberculosis and artificial intelligence|
Click here to view
[Figure 3] shows the type of collaboration and its bibliometric indicators. Most of the retrieved papers had only institutional (334 papers; 33.5%), followed by only national collaboration (326 papers; 32.7%), and international collaboration (427 papers; 22.3%). However, in terms of impact, international collaboration (3958 citations; 12.8 citations per paper) outperforms both national (3046 citations; 9.3 citations per paper) and institutional (3052 citations; 9.1 citations per paper). The remaining documents belong to the “single authorship” or “no collaboration” category (27 documents; 2.7%).
|Figure 3: Production and impact according to type of collaboration on tuberculosis and artificial intelligence|
Click here to view
| Discussion|| |
Tuberculosis is one of the leading causes of death after HIV/AIDS infection; moreover, this disease is highly prevalent in low and marginal socioeconomic strata. For this reason, in recent years, AI algorithms have been developed for prevention, diagnosis, and treatment of this disease. In this context, the aim of our study was to evaluate the characteristics of publications on this disease in the Scopus database. Our analysis found 1041 publications during the period 2016–2022.
Bibliometric studies explore and analyze the scientific data in a rigorous manner. This methodology facilitates the acquisition and evaluation of characteristics of publications. This allows us to quantify and qualify the production of countries, institutions, and authors.
Yudong Zhang was the author with the highest production. One of his most recent publications aimed to develop a computer-aided diagnosis system based on deep learning to automatically classify subjects with COVID-19, tuberculosis, and healthy control based on chest CT scan. The model proposed by the authors had a classification accuracy >90% On the other hand, K. C. Santosh was the author with the greatest impact. One of his most cited articles aimed to develop a system for automatic detection of tuberculosis in chest radiographs of Kenyan patients through a combination of features that minimized the classification error rate reaching 0.99 area under the curve.
Regarding the most productive institutions, although Anna University of India was the institution with the highest number of publications, the institution with the highest impact worldwide was Harvard University. This may be because, among its most cited publications is a deep-learning approach to predict the molecules with antibacterial activity against a broad phylogenetic spectrum including Mtb. This allowed the identification of numerous compounds that could serve as antibiotics, as well as setting a precedent in the development of new drugs against multidrug-resistant pathogens. In addition, Harvard University leads in the scientific production of other topics such as acute respiratory distress syndrome and medicine-related AI.,
The CEUR Workshop Proceedings journal published the largest number of documents related to our study topic; however, the Clinical Infectious Diseases journal had the greatest impact among the 10 with the highest production. According to other bibliometric analyses, this journal is also among the most cited in the topics related to Acinetobacter bacteremia and antimicrobial stewardship., On the other hand, regarding the quartile of the journals, most of the publications were published in Q1 journals. This distribution is maintained throughout the study years and with an overall sustained annual increase. This could be since developments in AI and tuberculosis have been of great interest and need worldwide.
Regarding collaboration between authors, national collaboration was the most productive. However, the international one was the one that had the greatest impact with a wide difference to the others. This should be considered by the authors since collaboration networks play an important role in the dissemination and impact of publications.
With respect to the most productive countries, the United States tops the list. This agrees with other bibliometric studies on the use of AI for prostate cancer where this country also leads in the production. On the other hand, China and India also stood out in number and impact of their publications. This reflects that both countries had a marked development on AI, as reported by other bibliometric analyses., In addition, five (India, China, South Africa, Indonesia, and Pakistan) of the 10 most productive countries are the countries with the highest incidence of tuberculosis worldwide, reflecting a concern for the urgent need for better solutions. However, other countries such as Angola, Nigeria, and Peru, which have a high incidence of multidrug-resistant tuberculosis are not on the list despite having a greater need for the treatment and follow-up strategies.,,,
| Conclusion|| |
In conclusion, articles related to tuberculosis and AI are mostly published in first quartile journals, which would reflect the need and interest worldwide. Publications with international collaboration reported a higher impact. Although countries with a high incidence of new TB cases are among the most productive, those with the highest reported drug resistance need greater support and collaboration.
Limitation of the study
Our study had some limitations that should be considered when interpreting the results. First, we only obtained the information from the Scopus database so those articles published in nonindexed journals were omitted; however, this database covers a wide spectrum of journals worldwide. Second, an analysis was made from 2016 onward so past publications that may have had greater relevance or impact could be omitted. Finally, although the search strategy was paired, false positives and false negatives are possible.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Benach J, Vergara M, Muntaner C. Desigualdad en salud: La mayor epidemia del siglo XXI. Madrid: Sesión epidemiología social; 2008.
Lozano J. Tuberculosis. Patogenia, diagnóstico y tratamiento. Offarm 2002;21:102-10.
Curioso WH, Brunette MJ. Artificial intelligence and innovation to optimize the tuberculosis diagnostic process. Rev Peru Med Exp Salud Publica 2020;37:554-8.
Avila-Tomás JF, Mayer-Pujadas MA, Quesada-Varela VJ. Artificial intelligence and its applications in medicine I: Introductory background to AI and robotics. Aten Primaria 2020;52:778-84.
Schwalbe N, Wahl B. Artificial intelligence and the future of global health. Lancet 2020;395:1579-86.
Deshpande D, Pasipanodya JG, Mpagama SG, Bendet P, Srivastava S, Koeuth T, et al
. Levofloxacin pharmacokinetics/pharmacodynamics, dosing, susceptibility breakpoints, and artificial intelligence in the treatment of multidrug-resistant tuberculosis. Clin Infect Dis 2018;67 Suppl 3:S293-302.
Gauthier É. Bibliometric Analysis of Scientific and Technological Research: A User´s Guide to the Methodology: Statistics Canada; 1998.
González De Dios J, Moya M, Mateos Hernández MA. Indicadores bibliométricos: Características y limitaciones en el análisis de la actividad científica. An Esp Pediatr 1997;47:235-44.
Mayta-Tovalino F, Quispe-Vicuña C, Cabanillas-Lazo M, Munive-Degregori A, Guerrero ME, Mendoza R. A Scientometric Analysis of Scholarly Output on COVID-19 and Dentistry. Int Dent J 2022;72:725-30.
Fosso Wamba S, Queiroz MM. Responsible Artificial Intelligence as a Secret Ingredient for Digital Health: Bibliometric Analysis, Insights, and Research Directions. Inf Syst Front 2021;1:1-16.
Natarajan A, Beena PM, Devnikar AV, Mali S. A systemic review on tuberculosis. Indian J Tuberc 2020;67:295-311.
Donthu N, Kumar S, Mukherjee D, Pandey N, Lim WM. How to conduct a bibliometric analysis: An overview and guidelines. J Bus Res 2021;133:285-96.
Yao X, Zhu Z, Kang C, Wang SH, Gorriz JM, Zhang YD. AdaD-FNN for Chest CT-Based COVID-19 Diagnosis. IEEE Trans Emerg Top in Comput Intell; 2022. p. 1-10.
Vajda S, Karargyris A, Jaeger S, Santosh KC, Candemir S, Xue Z, et al
. Feature selection for automatic tuberculosis screening in frontal chest radiographs. J Med Syst 2018;42:146.
Stokes JM, Yang K, Swanson K, Jin W, Cubillos-Ruiz A, Donghia NM, et al.
A deep learning approach to antibiotic discovery. Cell 2020;180:688-702.e13.
Yildirim F, Gulhan PY, Karaman I, Kurutkan MN. Bibliometric analysis of acute respiratory distress syndrome (ARDS) studies published between 1980 and 2020. Adv Clin Exp Med 2022;31:807-13.
Yoon HY, Lee H, Yee J, Gwak HS. Global research trends of gender-related artificial intelligence in medicine between 2001-2020: A bibliometric study. Front Med (Lausanne) 2022;9:868040.
Alkan S, Dindar Demiray EK, Yıldız E, Özlü C. Analysis of scientific publications on acinetobacter bacteremia in web of science. Infect Dis Clin Microbiol 2021;3:39-44.
Sweileh WM. Bibliometric analysis of peer-reviewed literature on antimicrobial stewardship from 1990 to 2019. Global Health 2021;17:1.
Shen Z, Wu H, Chen Z, Hu J, Pan J, Kong J, et al
. The global research of artificial intelligence on prostate cancer: A 22-year bibliometric analysis. Front Oncol 2022;12:843735.
Islam MM, Poly TN, Alsinglawi B, Lin LF, Chien SC, Liu JC, et al.
application of artificial intelligence in COVID-19 pandemic: Bibliometric analysis. Healthcare (Basel) 2021;9:441.
Huang J, Liu Y, Huang S, Ke G, Chen X, Gong B, et al
. Research output of artificial intelligence in arrhythmia from 2004 to 2021: A bibliometric analysis. J Thorac Dis 2022;14:1411-27.
WHO Global Lists of High Burden Countries for Tuberculosis (TB), TB/HIV and Multidrug/Rifampicin-Resistant TB (MDR/RR-TB), 2021-2025 Background Document. 2021. Available from: http://apps.who.int/bookorders
. [Last accessed on 2022 Jun 02].
Falagas ME, Pitsouni EI, Malietzis GA, Pappas G. Comparison of PubMed, scopus, web of science, and google scholar: Strengths and weaknesses. FASEB J 2008;22:338-42.
Robinson ED. Artificial intelligence in healthcare; its knowledge, practice, and perception among medical personnel in the developing economy. J Radiat Med Trop 2020;1:13-9. [Full text]
Tai MC. The impact of artificial intelligence on human society and bioethics. Tzu Chi Med J 2020;32:339-43. [Full text]
[Figure 1], [Figure 2], [Figure 3]
[Table 1], [Table 2], [Table 3]