Visual abstract

Introduction

Artificial intelligence (AI) has become increasingly integrated into the scientific process, providing tools to enhance efficiency, accuracy, and objectivity in research activities [1]. Conventional AI systems, such as machine learning models, have aided research tasks including statistical modeling, image analysis, and data classification [2]. More recently, generative AI (GenAI) tools, a subset of AI encompassing large language models like OpenAI’s ChatGPT and Google Gemini, hold particular promise for supporting researchers by streamlining workflows, fostering collaboration, and democratizing access to advanced computational tools [3]. These technologies can potentially enhance productivity and innovation while ensuring rigor and reproducibility in the scientific process [4]. GenAI chatbots can streamline tasks such as literature searches and reviews, analyze and interpret large datasets, and aid in the drafting of manuscripts [5,6]. They can also enhance the readability of scientific articles by summarizing, improving clarity, and bridging the language gap for non-native speakers, potentially decreasing language barriers in research [7]. In contrast, concerns have been raised regarding integrating GenAI tools into scientific research. These include the potential for factual inaccuracies, lack of transparency in decision-making, potential amplification of risks of bias, data privacy issues, and questions surrounding authorship and accountability [3]. Overreliance on such tools may also negatively affect critical thinking and writing skills, which could have greater implications for novice researchers [2]. Recognizing the opportunities and limitations of GenAI is vital for its responsible and equitable integration into research. These considerations have implications in clinical care, as research informs evidence-based outcomes, and patient care.

Traditional, complementary, and integrative medicine (TCIM) encompasses diverse practices, including: AYUSH (Ayurveda, Yoga, Unani, Siddha and Homeopathy; Indian), and Traditional Chinese Medicine; acupuncture, herbal medicine, homeopathy, and mind-body techniques (yoga, meditation, mindfulness, and hypnosis which often fall outside the scope of conventional biomedical frameworks); and integrative approaches that combine conventional with traditional and/or complementary therapies [8].

Traditional medicine refers to the collective knowledge and practices derived from various cultures, encompassing long-established beliefs, skills, and methods for health promotion, disease prevention, diagnosis, and treatment of mental and physical illnesses [9]. Complementary medicine is defined by the World Health Organization as “additional healthcare practices that are not part of a country’s mainstream medicine” [10]. Comparatively, integrative medicine is defined by the World Health Organization as an “interdisciplinary and evidence-based approach to health and well-being by using a combination of biomedical and traditional and/or complementary medical knowledge, skills and practices,” emphasizing holistic well-being [10]. In contrast, the National Center for Complementary and Integrative Health defines complementary approaches as those used together with conventional medicine, and integrative health as combining conventional and complementary approaches in a way to emphasize holistic, patient-centered care [11,12].

TCIM research is deeply intertwined with culturally grounded knowledge systems and diverse epistemologies, therefore, understanding TCIM-specific attitudes toward GenAI is essential to ensure these technologies are integrated in ways that respect TCIM-specific methodologies, and avoid the misappropriation of traditional knowledge [13]. Conducting rigorous research in TCIM can be particularly complex due to the interdisciplinary nature of the field, its reliance on diverse methodologies, and challenges associated with validating treatments [13,14]. GenAI chatbots can provide unique support to TCIM researchers by assisting with tasks such as synthesizing complex information, designing robust methodologies, and improving the clarity and coherence of research outputs [2]. These tools may also contribute to the equitable advancement of TCIM research by bridging gaps in resources and expertise among researchers from different regions and disciplines. Given the rising number of articles being published in TCIM [15] and the increasing prevalence of patients incorporating TCIM into their healthcare [16], GenAI tools may play an increasingly important role in supporting evidence-based practices and helping researchers monitor developments in the TCIM field.

Despite the transformative potential of GenAI in research [17], little is known about how TCIM researchers perceive the role of GenAI chatbots in the scientific process. There is a notable lack of empirical data on their attitudes toward these tools, including their perceived benefits, limitations, and ethical concerns. Understanding these attitudes and perceptions is crucial in the process of informing the responsible development and implementation of GenAI chatbots in ways that align with the unique needs and values of the TCIM research community. In addition, addressing concerns about transparency, accountability, and bias are essential to foster trust and acceptance among researchers [18,19].

Investigating the attitudes and perceptions of TCIM researchers toward GenAI chatbots in the scientific process is both timely and significant. By automating routine tasks, providing intelligent recommendations, and enhancing data interpretation, GenAI chatbots can potentially reduce researchers’ workload and improve the overall efficiency and quality of scientific inquiry [17]. Furthermore, these tools may help researchers address complex challenges in TCIM research, which are related to fostering interdisciplinary collaboration, by synthesizing information across multiple disciplines, translating terminology, facilitating communication between researchers from diverse backgrounds and specializations, and enhancing methodological rigor [20]. For example, AI-powered clinical decision support systems integrate data from multiple sources to provide evidence-based recommendations, facilitate communication among healthcare professionals, and thus, improve patient outcomes [17,20]. Exploring ethical considerations, such as data security and algorithmic bias, will be critical to ensuring the responsible adoption of GenAI chatbots in this field [17,20]. Intellectual property remains a foundational element that enables access to traditional knowledge for research and represents a core ethical concern requiring careful navigation to prevent misuse of GenAI chatbots [21]. Insights from the proposed study may help guide the development and integration of GenAI tools that align with the goals of TCIM researchers and support the advancement of evidence-based TCIM practices.

In summary, this study aims to fill an existing knowledge gap by exploring the attitudes and perceptions of TCIM researchers regarding the use of GenAI chatbots in the scientific process. The findings of the study may offer useful insights into how these tools could be responsibly implemented to support TCIM research, with potential implications for the broader scientific community.

Materials and Methods

1. Approach and open science statement

This study will employ a cross-sectional, web-based survey design targeting TCIM researchers identified as corresponding authors of subject-specific MEDLINE-indexed articles. This study’s protocol has been registered on the Open Science Framework [22]. Prior to data collection, ethics approval has been obtained from the University Hospital Tübingen, Germany, Research Ethics Board (REB #: 079/2025BO2). Study materials and data will also be shared on the Open Science Framework as they become available. After all data collection and analysis, the findings will be incorporated into a manuscript and posted as a preprint before submission to a peer-reviewed journal.

2. Sampling framework

We conducted a search on Ovid MEDLINE to identify relevant articles published between December 15, 2014, and December 15, 2024. The search targeted publications related to a broad range of TCIM disciplines. The search strategy combined MeSH terms and keywords for major TCIM domains, including acupressure, acupuncture, chiropractic, complementary therapies, Chinese herbal medicine, electroacupuncture, herbal medicine, integrative medicine, applied kinesiology, osteopathic manipulation, Ayurvedic medicine, traditional Chinese medicine, mind-body therapies, naturopathy, phytotherapy, medicinal plants, and yoga. Based on this search strategy, it can be assumed that a significant proportion of the corresponding authors of these articles are TCIM researchers. Non-research articles (i.e., records not reporting original research and reviews), such as editorials, opinions, and commentaries, were excluded using a dedicated search strategy line. PubMed Identifier numbers from the resulting articles will be exported from Ovid as .csv files and processed using an R script (based on the 2019 easyPubMed and RedEye packages) to extract author names, affiliated institutions, and email addresses [23,24]. Once the PubMed Identifier numbers are imported to R, corresponding author email extraction will be finalized, with the data subsequently cleaned for errors or duplicates before survey distribution. In addition, extracted email addresses will be manually checked for truncations and verified using a hex recorder developed in PyCharm before survey distribution. Authors who meet the inclusion criteria as outlined in the sampling framework will be invited to participate in an anonymous online survey. Details about the process for retrieving authors’ names and email addresses can be found [25].

3. Participant recruitment

Only researchers identified through the sampling framework will be invited to participate in the study. The prospective participants will be contacted via email and invited to complete the survey voluntarily and anonymously. Emails will be sent via SurveyMonkey, and will include details about the study objectives, and a link to an informed consent form. Participants will be required to provide consent before accessing the online survey questions. Duplicate entries of name and email address pairs will be removed from the sample. A response rate of 3%–7% is anticipated based on prior research previously conducted on related topics [5,26,27]. Participants will have 5 weeks to complete the survey, with reminder emails sent 1, 2, and 3 weeks after the initial invitation. A 2-week final-response window will follow the final reminder email. Participation will be entirely voluntary, without financial compensation, and respondents may skip any questions they do not wish to answer.

4. Survey design

The survey will be developed using SurveyMonkey software [28]. Researchers will receive an email invitation containing a link to the survey on SurveyMonkey, which begins with an implied consent agreement that participants must agree to, before viewing the survey questions. The survey will consist of multiple-choice questions, beginning with demographic questions. The demographic section will include questions adapted from previous studies on TCIM [26,27]. Following the demographics section, participants will answer questions regarding their awareness, attitudes, and use of GenAI chatbots at different stages of the scientific process, including study design, data analysis, manuscript preparation, and peer communication. These AI-related questions were adapted from a recent study on AI use in research workflows [5]. The survey is expected to take approximately 15 minutes to complete. A complete version of the survey is available [25].

5. Data analysis

This proposed study will not include a formal hypothesis. Descriptive statistics, such as frequencies and percentages, will be used to analyze the quantitative data. If a significant number of valuable qualitative responses are collected, a thematic content analysis will be performed. Subgroup differences will be examined by stratifying responses across key demographic variables (e.g., age, sex, career stage). In addition, analyses will stratify participants by whether they identify TCIM as their primary or secondary area of research, which will allow for comparisons based on their depth of engagement in TCIM research. Two members of the research team will independently code the qualitative data and engage in multiple rounds of discussion to reach consensus on the coding framework. Using an inductive approach, the coding categories will be developed directly from the data, with careful attention to ensure that each category is clear, distinct, and accurately reflects participants’ responses [29,30]. The survey study will be reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology guidelines for cross-sectional studies and the Checklist for Reporting Results of Internet E-Surveys reporting guidelines [31,32].

Discussion

The primary objective of the proposed survey study is to understand the attitudes and perceptions of TCIM researchers regarding the use of GenAI chatbots in the scientific process. TCIM researchers may hold distinct perspectives compared with those in biomedical fields due to the diversity of TCIM therapies, their frequent engagement with traditional knowledge systems, and the cultural nature of many TCIM approaches to patient care [14,33,34]. Consequently, TCIM researchers may have differing views on the use of GenAI for study ideation, methodological design, and data sovereignty. By exploring these attitudes and perceptions, the proposed study seeks to identify the perceived benefits, challenges, and ethical considerations associated with integrating GenAI tools into TCIM research workflows. The findings may have important implications for the responsible adoption of GenAI chatbots in TCIM research, provide valuable insights into researchers’ acceptance of these tools, and highlight the potential barriers to their effective use. Understanding TCIM researchers’ perspectives will inform the development of GenAI technologies that align more closely with the specific needs and values of the TCIM research community, ultimately supporting the advancement of high-quality, evidence-based research in this field. Furthermore, this study will contribute to broader discussions about the ethical use of GenAI in scientific research, emphasizing the importance of transparency, accountability, and equity in the design and implementation of these technologies [4]. Outside of the direct research context, the study findings may also inform how TCIM institutions, journals, and professional networks develop policies and guidance to support the responsible and ethical use of GenAI chatbots in research workflows.

GenAI tools have the potential to meaningfully strengthen TCIM research practice by supporting interdisciplinary collaboration and improving methodological coherence across diverse evidence traditions. Within integrative medicine teams, where clinicians and researchers often navigate differing terminologies, epistemologies, and documentation norms, GenAI can facilitate communication by synthesizing information across systems, translating concepts between TCIM and conventional medicine frameworks, and help harmonize documentation and reporting standards used in clinical trials. These capabilities may also advance transparency and rigor in TCIM research by assisting with standardized data extraction, protocol adherence checks, and consistent reporting structures that align with international guidelines. In addition, GenAI can expand the capacity for meta-research in TCIM by enabling efficient mapping, appraisal, and synthesis of heterogeneous evidence bases, thereby supporting more robust evaluations of research quality and methodological trends. Together, these applications underscore the role of GenAI as a potential catalyst for improving the environment for integrative research and strengthen the evidence, the foundation to underpin TCIM practice.

This protocol to conduct a cross-sectional survey will enable the collection of data at a single time point. This makes a study both efficient and cost-effective [35]. By employing a sampling strategy that seeks a large and diverse international participant pool, the aim is to enhance the generalizability of the survey study findings regarding TCIM researchers’ attitudes and perceptions of AI chatbots in the scientific process. In addition, all data extraction and cleaning procedures will be conducted first by software, then by manual verification to ensure accuracy and minimize errors. Ethical considerations were applied to all email and contact information extraction procedures, with data handling compliant with applicable data protection standards, such as the General Data Protection Regulation or equivalent local frameworks, to ensure that personal contact information is stored securely and used solely for research purposes.

However, the proposed study also has limitations [36]. Despite planning to survey many researchers, we anticipate a relatively low response rate due to several factors. These include participants being unfamiliar with the research team and not expecting a survey invitation. Other logistical challenges, such as outdated contact information, changes in affiliations, email inaccessibility, bounce-back emails, vacations, and retirements, also contribute to non-response bias. In addition, some participants may have passed away, which can further impact response rates. Furthermore, non-English speaking researchers were largely excluded from the sample (via the MEDLINE search strategy excluding articles not published in English) due to language and resource constraints. This may also have implications on TCIM knowledge representation, as non-English speaking researchers working with nonbiomedical, spiritual, symbolic, or narrative knowledge systems prevalent in TCIM, may be underrepresented, and this may result in limited diversity in epistemic perspectives captured through the survey. Limited English proficiency among respondents may affect the completion of the survey, potentially restricting their ability to express their attitudes and opinions comprehensively. Consequently, this study may not be applicable towards researchers who publish in languages outside of English. In the context of recent developments in large language models capable of automated multilingual translation, future research should employ multilingual survey tools combined with AI-based translation scripts to ensure global participation can be accommodated. The findings may be skewed toward respondents with stronger views about AI chatbots (whether positive or negative) and those with greater experience or knowledge of these tools, as individuals with more neutral perspectives or less familiarity may have been less motivated to participate. Another potential limitation is recall bias; participants may have difficulty accurately remembering their use of GenAI chatbots or omit details about their experiences, especially if significant time has elapsed since the research activities being surveyed took place. Finally, the survey primarily focuses on GenAI chatbots rather than rule-based or retrieval-based systems, which may limit applicability to the full spectrum of AI tools used in TCIM.

Conclusion

This proposed survey study will offer insights into how TCIM researchers perceive the role of GenAI chatbots in the scientific process. By exploring their attitudes, experiences, and ethical concerns, the findings could help guide the responsible development and implementation of these tools in TCIM research. The results may also inform and enhance research quality, support interdisciplinary collaboration, and promote equitable access to GenAI technologies. Ultimately, this work may contribute to ongoing discussions about evidence-based practices in TCIM and the thoughtful integration of GenAI chatbots across diverse research contexts.

Article information

Author Contributions

Conceptualization: JYN. Methodology: JYN, KA, PKG, SL, DM, MR, MS, JPS, TY, MSL, and HC. Resources: JYN and HC. Materials: JYN and HC. Writing - Original Draft: JYN and JT. Writing – Review & Editing: JYN, JT, KA, PKG, SL, DM, MR, MS, JPS, TY, MSL, and HC. Supervision: JYN. Project administration: JYN.

Conflicts of Interest

The authors have no conflicts of interest to declare.

Author Use of AI Tools Statement

No AI tools were used in the writing of this article.

Funding

This study was unfunded.

Ethical Statement

Ethics approval has been obtained from the University Hospital Tübingen, Germany, Research Ethics Board (REB no.: 079/2025BO2).

References

• [1] Xu Y, Wang Q, An Z, Wang F, Zhang L, Wu Y, et al. Artificial intelligence: a powerful paradigm for scientific research. Innovation 2021;2(4):100179. ArticlePubMedPMC
• [2] Salvagno M, Taccone FS, Gerli AG. Can artificial intelligence help for scientific writing? Crit Care. 2023, 27:p 75. ArticlePubMedPMCPDF
• [3] Chen A, Liu L, Zhu T. Advancing the democratization of generative artificial intelligence in healthcare: a narrative review. J Hosp Manag Health Policy 2024;8:12. Article
• [4] Singhal A, Neveditsin N, Tanveer H, Mago V. Toward fairness, accountability, transparency, and ethics in ai for social media and health care: scoping review. JMIR Med Inform 2024;12(1):e50048. ArticlePubMedPMC
• [5] Ng JY, Maduranayagam SG, Suthakar N, Li A, Lokker C, Iorio A, et al. Attitudes and perceptions of medical researchers towards the use of artificial intelligence chatbots in the scientific process: an international cross-sectional survey. Lancet Digit Health 2025;7(1):e94−102.ArticlePubMed
• [6] Fabiano N, Gupta A, Bhambra N, Luu B, Wong S, Maaz M, et al. How to optimize the systematic review process using AI tools. JCPP Adv 2024;4(2):e12234. ArticlePubMedPMC
• [7] Kirchner GJ, Kim RY, Weddle J, Bible JE. Can artificial intelligence improve the readability of patient education materials? Clin Orthop Relat Res 2023;481(11):2260−7.ArticlePubMedPMC
• [8] Ng JY, Dhawan T, Dogadova E, Taghi-Zada Z, Vacca A, Wieland LS, et al. Operational definition of complementary, alternative, and integrative medicine derived from a systematic search. BMC Complement Med Ther 2022;22(1):104. ArticlePubMedPMCPDF
• [9] World Health Organization [Internet]. Traditional medicine: Geneva (Switzerland); World Health Organization: 2023 Available from: https://www.who.int/news-room/questions-and-answers/item/traditional-medicine
• [10] World Health Organization [Internet]. Traditional, complementary and integrative medicine: Geneva (Switzerland); World Health Organization: 2019 Available from: https://www.who.int/publications/i/item/978924151536
• [11] National Center for Complementary and Integrative Health [Internet]. Complementary, alternative, or integrative health: what’s in a name?: NCCIH: 2021 Available from: https://www.nccih.nih.gov/health/complementary-alternative-or-integrative-health-whats-in-a-name
• [12] Ng JY, Boon HS, Thompson AK, Whitehead CR. Making sense of “alternative”, “complementary”, “unconventional” and “integrative” medicine: exploring the terms and meanings through a textual analysis. BMC Complement Altern Med 2016;16(1):134. ArticlePubMedPMC
• [13] Raja M, Cramer H, Lee MS, Wieland LS, Ng JY. Addressing the challenges of traditional, complementary, and integrative medicine research: an international perspective and proposed strategies moving forward. Perspect Integr Med 2024;3(2):86−97.ArticlePDF
• [14] Veziari Y, Leach MJ, Kumar S. Barriers to the conduct and application of research in complementary and alternative medicine: a systematic review. BMC Complement Altern Med 2017;17(1):166. ArticlePubMedPMCPDF
• [15] Ng JY. Insight into the characteristics of research published in traditional, complementary, alternative, and integrative medicine journals: a bibliometric analysis. BMC Complement Med Ther 2021;21(1):185. ArticlePubMedPMCPDF
• [16] World Health Organization [Internet]. WHO traditional medicine strategy: 2014-2023: Geneva (Switzerland); World Health Organization: 2013 [cited 2025 Aug 14]. Available from: https://www.who.int/publications/i/item/9789241506096
• [17] Ng JY, Cramer H, Lee MS, Moher D. Traditional, complementary, and integrative medicine and artificial intelligence: Novel opportunities in healthcare. Integr Med Res 2024;13(1):101024. ArticlePubMedPMC
• [18] Amabie T, Izah SC, Ogwu MC, Hait M. Harmonizing tradition and technology: the synergy of artificial intelligence in traditional medicine. Herbal Medicine Phytochemistry: Applications and Trends. Springer, 2024, pp 2103−25.ArticlePubMed
• [19] Ofosu-Asare Y. Cognitive imperialism in artificial intelligence: counteracting bias with indigenous epistemologies. AI Soc 2024;40:3045−61.ArticlePDF
• [20] World Health Organization [Internet]. Mapping the application of artificial intelligence in traditional medicine: technical brief: [cited 2025 Jun 25]. Available from: https://www.who.int/publications/i/item/9789240107663
• [21] Elendu C, Amaechi DC, Elendu TC, Jingwa KA, Okoye OK, Okah MJ, et al. Ethical implications of AI and robotics in healthcare: a review. Medicine 2023;102(50):e36671. ArticlePubMedPMC
• [22] Ng JY, Tan J, Adapa K, Gupta PK, Li S, Mehta DH, et al. [Internet]. Traditional, complementary, and integrative medicine researcher attitudes and perceptions of generative artificial intelligence chatbots in the scientific process: a large-scale, international cross-sectional survey: OSF: 2025 Available from: https://doi.org/10.17605/OSF.IO/Q9UNK
• [23] Fantini D. [Internet]. easyPubMed: Search and Retrieve Scientific Publication Records from PubMed. R-Packages: 2019 Mar 29 Available from: https://cran.r-project.org/web/packages/easyPubMed/index.htmlArticle
• [24] Fry D, Al-Khafaji W. [Internet]. RedEye Pipeline. (V1.0): Toronto (Canada); Daniel Fry: 2025 [cited 2025 Aug 29]. Available from: https://doi.org/10.5281/zenodo.16996504
• [25] OSF [Internet]. STRATEGY_TCIM Researchers GenAI Chatbots Scientific Process_Aug1425: Available from: https://osf.io/68rjc
• [26] Ng JY, Santoro LJ, Cobey KD, Steel A, Cramer H, Moher D. Complementary, alternative, and integrative medicine researchers’ practices and perceived barriers related to open science: An international, cross-sectional survey. PLoS One 2024;19(5):e0301251. ArticlePubMedPMC
• [27] Ng JY, Lin BX, Kreuder L, Cramer H, Moher D. Open science practices among authors published in complementary, alternative, and integrative medicine journals: an international, cross-sectional survey. Medicine 2024;103(44):e40259. ArticlePubMedPMC
• [28] SurveyMonkey Inc [Internet]. SurveyMonkey: San Mateo (CA); SurveyMonkey Inc: [cited 2025 Aug 14]. Available from: https://www.surveymonkey.com/
• [29] Raskind IG, Shelton RC, Comeau DL, Cooper HLF, Griffith DM, Kegler MC. A review of qualitative data analysis practices in health education and health behavior research. Health Educ Behav 2019;46(1):32−9.ArticlePubMedPMCPDF
• [30] Thomas DR. A general inductive approach for analyzing qualitative evaluation data. Am J Eval 2006;27(2):237−46.ArticlePDF
• [31] von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP. Strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. BMJ 2007;335(7624):806−8.ArticlePubMedPMC
• [32] Eysenbach G. Improving the quality of web surveys: the checklist for reporting results of internet e-surveys (CHERRIES). J Med Internet Res 2004;6(3):e34. ArticlePubMedPMC
• [33] Ng JY, Dhawan T, Fajardo RG, Masood HA, Sunderji S, Wieland LS, et al. The brief history of complementary, alternative, and integrative medicine terminology and the development and creation of an operational definition. Integr Med Res 2023;12(4):100978. ArticlePubMedPMC
• [34] Ng JY, Lee MS, Liu JP, Steel A, Wieland LS, Witt CM, et al. How can meta-research be used to evaluate and improve the quality of research in the field of traditional, complementary, and integrative medicine? Integr Med Res 2024;13(3):101068. ArticlePubMedPMC
• [35] Wang X, Cheng Z. Cross-sectional studies: strengths, weaknesses, and recommendations. Chest 2020;158(1S):S65−71.ArticlePubMedPMC
• [36] Choi BC, Pak AW. A catalog of biases in questionnaires. Prev Chronic Dis 2004;2(1):A13. PubMedPMC

Figure & Data

References

Citations

Citations to this article as recorded by