High Awareness, Low Trust: Medical Residents’ Readiness to Use ChatGPT in Clinical Practice in an LMIC Context

High Awareness, Low Trust: Medical Residents’ Readiness to Use ChatGPT in Clinical Practice in an LMIC Context | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article High Awareness, Low Trust: Medical Residents’ Readiness to Use ChatGPT in Clinical Practice in an LMIC Context Saeed Zahmatkesh, Elaheh Hooshmand, Marziyhe Meraji This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8394385/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background: Large language models such as ChatGPT are rapidly entering medical education and clinical practice, yet evidence on clinicians’ readiness to adopt these tools in low- and middle-income countries remains limited. This study aimed to assess medical residents’ awareness, trust, perceptions, and predictors of willingness to use ChatGPT in clinical settings in Iran. Methods: A cross-sectional survey was conducted among 651 medical residents at Mashhad University of Medical Sciences between December 2023 and March 2024. A validated Persian questionnaire assessed familiarity with AI, trust in ChatGPT, perceived benefits and risks, and willingness to use the tool. Descriptive statistics, Chi-square tests, and multivariable logistic regression were used for data analysis. Results: Awareness of ChatGPT was universal; however, trust remained limited. Most residents supported the use of ChatGPT for low-risk tasks such as general health information, while none endorsed its independent diagnostic use. Multivariable regression showed that trust in ChatGPT, belief in its potential to improve clinical outcomes, and higher AI familiarity were the strongest independent predictors of willingness to use the technology. Concerns related to accuracy, privacy, and legal responsibility remained the dominant barriers to clinical adoption. Conclusion: Iranian medical residents demonstrate cautious optimism toward ChatGPT, but limited trust and regulatory uncertainty restrict real-world use. Strengthening AI education, establishing institutional policies, and developing national regulatory frameworks are essential to enable the safe and effective integration of large language models into clinical training and practice. ChatGPT Large Language Models Medical residents Trust AI acceptance Digital health Iran Clinical education Introduction Recent advances in artificial intelligence—particularly the emergence of Large Language Models (LLMs) such as GPT-based systems—have created unprecedented opportunities for transforming medical education, clinical decision-making, and health service delivery ( 1 , 2 ). These models can summarize complex clinical information, support diagnostic reasoning, enhance patient communication, and streamline administrative and educational tasks. As their capabilities rapidly expand, LLMs are increasingly viewed as a cornerstone of the next generation of digital health innovation ( 1 – 4 ). However, despite their remarkable potential, their integration into real-world clinical environments remains challenging. One of the most critical concerns surrounding the use of LLMs in healthcare is the phenomenon of hallucination , in which models generate plausible but factually incorrect or clinically unsafe information ( 5 ). Such errors can compromise patient safety, misguide clinical decisions, and undermine trust among healthcare providers. Equally important are the ethical and regulatory implications, including data privacy, security of electronic health records, and uncertainty regarding legal accountability in cases of AI-related harm ( 6 , 7 ). International frameworks such as the GDPR in Europe and HIPAA in the United States have begun outlining boundaries for responsible AI adoption. However, in many low- and middle-income countries (LMICs), including Iran, clear regulatory infrastructures for medical AI are still emerging, creating additional uncertainty for users and policymakers. Importantly, physicians’ acceptance of AI-based tools is deeply influenced by their prior exposure, perceived usability, and confidence in the accuracy and safety of these systems ( 8 , 9 ). Among all healthcare providers, medical residents represent a uniquely critical population for investigation: They form the frontline of patient care in teaching hospitals, They frequently interact with clinical documentation systems, They rely heavily on up-to-date medical knowledge, and Their perceptions often forecast the future trajectory of technology adoption within the health system. Despite the exponential global growth of LLM-related research, evidence from Middle Eastern health systems remains scarce ( 10 , 11 ). Iran, in particular, is at an early stage of integrating AI technologies into clinical workflows, and very few empirical studies have examined healthcare workers’ readiness, trust, concerns, or ethical apprehensions toward LLMs. This gap is even more pronounced among medical residents, who are both primary end-users and future decision-makers in the digital health ecosystem. Mashhad University of Medical Sciences (MUMS), as one of the largest academic medical centers in eastern Iran, offers a distinctive context for such an investigation. The university hosts a diverse body of residents across a wide range of specialties, manages some of the busiest tertiary hospitals in the region, and has begun exploring digital health innovations despite limited structural and regulatory frameworks. Studying this population therefore provides not only a representative view of residents' perceptions in Iran but also valuable insights for LMICs facing similar challenges in adopting LLMs safely and ethically. Given the rapid expansion of LLMs, the growing global debate on their risks and benefits, and the absence of region-specific evidence, this study aims to assess medical residents’ awareness, attitudes, operational barriers, and privacy-related concerns regarding the use of LLMs in healthcare. Understanding these perspectives is essential for shaping evidence-based policies, designing targeted training programs, and developing ethical frameworks to guide responsible AI adoption in clinical settings. Method The study population included all medical residents enrolled in specialty training programs at MUMS during the 2023–2024 academic year. Eligibility criteria were: Active enrollment in a residency program, Willingness to participate voluntarily, and Ability to complete a Persian-language online questionnaire. Residents who declined participation, submitted incomplete questionnaires (defined as more than 20% missing responses), or withdrew before completion were excluded from the final analysis. A stratified non-random sampling strategy was used to ensure proportional representation across major clinical specialties. The strata were defined based on clinical specialty, and proportional allocation was applied according to the distribution of residents within each department. The survey link was distributed through official residency communication platforms, hospital-based academic groups, and institutional mailing lists. A total of 651 residents completed the questionnaire and were included in the final analysis. Data were collected using a structured questionnaire adapted from the instrument originally developed by Nri-Ezedi et al. ( 12 ), designed to assess physicians’ awareness, attitudes, perceived benefits, operational barriers, and privacy concerns regarding the clinical use of Large Language Models (LLMs). Because the original questionnaire was developed in English, a standardized forward–backward translation procedure was implemented to ensure linguistic and conceptual equivalence. Two independent bilingual translators performed the forward translation into Persian. A reconciled version was subsequently back-translated into English by two independent translators who were blinded to the original instrument. An expert panel consisting of two specialists in Medical Informatics and one specialist in Medical Ethics evaluated the translated items for clarity, relevance, and cultural appropriateness. Quantitative content validity indices demonstrated excellent validity: All items achieved a Content Validity Ratio (CVR) of 1.00, The Item-level Content Validity Index (I-CVI) ranged from 0.83 to 1.00, and The Scale-level CVI/Average (S-CVI/Ave) was 0.92. A pilot study was conducted with 30 medical residents to evaluate the internal consistency and comprehensibility of the Persian version of the questionnaire. Minor linguistic refinements were applied based on participant feedback. The final instrument demonstrated acceptable internal reliability with a Cronbach’s alpha coefficient of 0.76, indicating satisfactory internal consistency. Data were collected electronically using Google Forms. The online questionnaire included an embedded informed consent form at the beginning of the survey. Participation was entirely voluntary, anonymous, and without financial or academic incentives. Participants were informed about the purpose of the study, confidentiality of the data, and their right to withdraw at any time without any consequences. The questionnaire assessed the following domains: Demographic characteristics (age, gender, specialty, years of experience, patient load) Familiarity with artificial intelligence and ChatGPT Trust in ChatGPT for clinical use Willingness to use ChatGPT in healthcare practice Perceived benefits (e.g., access to information, efficiency) Perceived risks (e.g., hallucination, privacy, legal responsibility) Appropriate clinical use-cases Prior experience and recommendation behavior toward AI tools Data analysis was performed using SPSS version 22 . Descriptive statistics, including frequencies and percentages, were used to summarize demographic characteristics and key study variables. Inferential analysis was conducted using: Chi-square tests to examine associations between categorical variables, and Fisher’s Exact Test when expected cell counts were less than five. The following associations were specifically tested: Gender and trust in ChatGPT Specialty and familiarity with AI Clinical experience and concern about information accuracy Willingness to use ChatGPT and belief in its potential to improve clinical outcomes All statistical tests were two-tailed, and statistical significance was defined as p < 0.05. All ethical requirements for human research were strictly observed. Participation was voluntary, anonymity was maintained, and no identifiable personal data were collected. Participants were informed about the objectives of the study, data confidentiality, and their right to withdraw at any time. Results A total of 651 medical residents participated in the study. The demographic characteristics are presented in Table 1 . Table 1 Demographic Characteristics of Participating Medical Residents (N = 651) Variable Category Frequency (%) Age 20–29 years 267 (41.0%) 30–39 years 225 (34.6%) 40–49 years 134 (20.5%) ≥ 50 years 25 (3.8%) Gender Female 376 (57.7%) Male 275 (42.3%) Specialty Surgery 234 (35.9%) Internal Medicine 167 (25.6%) Pediatrics 116 (17.9%) Emergency Medicine 92 (14.1%) Oncology 42 (6.4%) Years of Experience 15 years 75 (11.5%) Daily Patient Visits 30 patients 83 (12.8%) The sample consisted predominantly of young residents, with more than 75% under the age of 40. This distribution indicates a workforce that is digitally exposed in personal life but may lack structured training in clinical AI. The strong presence of surgical and internal medicine residents provides insights across both procedural and cognitive specialties. Additionally, the substantial daily patient load observed among most participants highlights the potential relevance of time-saving digital tools in this context. Table 2 summarizes residents’ awareness of and familiarity with artificial intelligence and ChatGPT. Table 2 Awareness and Familiarity with AI and ChatGPT (N = 651) Variable Category Frequency (%) AI Familiarity Highly familiar 100 (15.4%) Somewhat familiar 367 (56.4%) Not very familiar 167 (25.6%) Not familiar at all 17 (2.6%) Heard of ChatGPT Yes 651 (100%) Description of ChatGPT AI system 501 (76.9%) Virtual assistant 150 (23.1%) Medical chatbot 0 (0%) While familiarity with AI varied, awareness of ChatGPT was universal. Residents mostly perceived ChatGPT as a general AI system rather than a clinical tool, revealing a conceptual gap between exposure and specialized understanding. This pattern is typical in settings where formal AI education is limited.Levels of trust in ChatGPT and residents’ willingness to use the tool in clinical care are presented in Table 3 . Table 3 Trust and Willingness to Use ChatGPT in Clinical Care (N = 651) Variable Category Frequency (%) Trust in ChatGPT Very confident 0 (0%) Moderately confident 250 (38.5%) Not very confident 334 (51.3%) Not confident at all 67 (10.3%) Willingness to Use Strongly willing 84 (12.8%) Somewhat willing 267 (41.0%) Not very willing 217 (33.3%) Not willing at all 83 (12.8%) No resident expressed complete trust in ChatGPT, and more than half were not very confident in its clinical reliability. Despite this, moderate willingness to use the tool suggests a cautious openness—residents see potential utility but remain concerned about risks, accuracy, and patient safety. Table 4 summarizes residents’ perceptions of the potential advantages of ChatGPT use in healthcare Table 4 Perceived Advantages of Using ChatGPT in Healthcare (N = 651) Advantage Frequency (%) Improving access to information 284 (43.6%) Enhancing communication 8 (1.3%) Increasing clinical efficiency 234 (35.9%) Combined advantages 125 (19.2%) Residents viewed ChatGPT primarily as an information-support tool. The extremely low emphasis on communication enhancement implies that residents do not yet associate LLMs with patient–provider relational improvements, consistent with early-phase AI adoption literature. Residents’ perceived disadvantages and risks associated with the use of ChatGPT are presented in Table 5 . Table 5 Perceived Disadvantages and Risks of ChatGPT (N = 651) Disadvantage Frequency (%) Misinterpretation of information 267 (41.0%) Dependency on technology 117 (17.9%) Reduced human interaction 75 (11.5%) Combined disadvantages 192 (29.5%) Concerns centered heavily on information accuracy, reflecting global apprehensions regarding AI hallucinations. The prominence of multi-dimensional concerns indicates that skepticism is driven by complex and interconnected risks rather than a single barrier. Table 6 summarizes residents’ views on appropriate clinical applications of ChatGPT. Table 6 Appropriate Clinical Use-Cases for ChatGPT (N = 651) Use-Case Frequency (%) Providing general health information 392 (60.3%) Diagnostic use alone 0 (0%) Medication management 67 (10.3%) Combined use-cases 192 (29.5%) Participants strongly supported low-risk applications such as patient education but unanimously rejected ChatGPT as an independent diagnostic tool. This reflects professional caution and aligns with international recommendations advising against AI-driven diagnosis without supervision. Residents’ concerns regarding accuracy, privacy, and legal responsibility related to the use of ChatGPT are presented in Table 7 . Table 7 Privacy, Accuracy, and Legal Concerns (N = 651) Concern Frequency (%) Accuracy 184 (28.2%) Privacy and security 159 (24.4%) Legal responsibility 50 (7.7%) Combined concerns 258 (39.6%) Accuracy and privacy collectively accounted for most concerns, emphasizing the need for robust regulatory frameworks in early-stage AI adoption environments. The high rate of combined concerns underscores that residents view AI risk as multi-factorial. Table 8 summarizes residents’ prior experience with ChatGPT, recommendation to patients, and perceived importance of ChatGPT-related knowledge. Table 8 Prior Use of ChatGPT and Perceived Importance (N = 651) Variable Category Frequency (%) Recommended AI to patients Yes 67 (10.3%) No 584 (89.7%) Importance of ChatGPT knowledge Very important 100 (15.4%) Moderately important 343 (52.6%) Not important 208 (32.0%) Belief in improved outcomes Yes 526 (80.8%) No 125 (19.2%) Although most residents believed ChatGPT could enhance healthcare outcomes, very few had ever recommended AI tools to patients. This gap indicates that theoretical acceptance has not yet translated into real-world behavior. Table 9 presents the results of chi-square analyses assessing associations between demographic and attitudinal variables related to ChatGPT use. Table 9 Inferential Analysis of Key Associations (Chi-Square Tests, N = 651) Comparison χ² df p-value Interpretation Gender × Trust 8.71 3 0.033 Significant Specialty × AI familiarity 21.54 12 0.043 Significant Experience × Accuracy concern 14.62 9 0.101 Not significant Willingness × Belief in improved outcomes 19.83 3 < 0.001 Highly significant Trust varied significantly by gender, and familiarity differed by specialty. Experience did not influence accuracy concerns, implying widespread skepticism across seniority levels. The strongest relationship was between willingness to use AI and belief in its benefits, emphasizing the central role of perceived value in adoption decisions. To identify independent predictors of residents’ willingness to use ChatGPT in clinical practice, a binary logistic regression analysis was performed. The dependent variable was defined as willingness to use ChatGPT (coded as 1 = willing [strongly or somewhat willing], and 0 = not willing [not very willing or not willing at all]). Independent variables entered into the model included: Gender Specialty Level of familiarity with AI Trust in ChatGPT Belief in ChatGPT’s potential to improve clinical outcomes The results of the multivariable logistic regression analysis predicting residents’ willingness to use ChatGPT in clinical practice are presented in Table 10 . Table 10 Multivariable Logistic Regression Predicting Willingness to Use ChatGPT (N = 651) Predictor Adjusted OR 95% CI p-value Male gender 1.28 0.91–1.80 0.148 Surgical specialty 1.94 1.31–2.89 0.001 High AI familiarity 2.63 1.74–3.99 < 0.001 Moderate/High trust in ChatGPT 4.11 2.88–5.86 < 0.001 Belief in improved outcomes 5.27 3.46–8.02 < 0.001 Model fit: Hosmer–Lemeshow test: p = 0.41 Nagelkerke R² = 0.38 Multivariable logistic regression analysis demonstrated that trust in ChatGPT and belief in its potential clinical benefits were the strongest independent predictors of residents’ willingness to use the tool in practice. Residents who expressed moderate to high trust in ChatGPT were 4.11 times more likely to report willingness to use it clinically (OR = 4.11, 95% CI: 2.88–5.86, p < 0.001). Similarly, those who believed that ChatGPT could improve clinical outcomes were more than five times as likely to be willing users (OR = 5.27, 95% CI: 3.46–8.02, p < 0.001). Higher familiarity with AI also significantly increased willingness to use ChatGPT (OR = 2.63, p < 0.001). Surgical residents were nearly twice as likely to report willingness compared with non-surgical specialties (OR = 1.94, p = 0.001). Gender was not a significant independent predictor in the adjusted model. The Hosmer–Lemeshow test indicated acceptable model fit (p = 0.41), and the model explained approximately 38% of the variance in willingness to use ChatGPT. All predictors were entered simultaneously using the Enter method. The goodness-of-fit of the model was assessed using the Hosmer–Lemeshow test, and adjusted odds ratios (ORs) with 95% confidence intervals (CIs) were calculated. A p-value of < 0.05 was considered statistically significant. Discussion Medical residents’ perceptions of large language models, particularly ChatGPT, reflect a complex and transitional stage between widespread awareness and cautious clinical engagement. Although awareness of ChatGPT was universal in this study, both descriptive and regression findings demonstrate that awareness alone is insufficient to drive adoption. Trust and perceived clinical usefulness emerged as the most influential determinants of willingness to use ChatGPT, underscoring that psychological confidence and perceived performance enhancement remain the central pillars of technology acceptance in clinical environments ( 13 ). The multivariable regression model provides important insights beyond bivariate associations by demonstrating that belief in ChatGPT’s potential to improve clinical outcomes was the strongest independent predictor of willingness to use the technology. Residents holding this belief were more than five times as likely to express willingness for clinical use. This strong predictive effect aligns closely with the core assumptions of the Technology Acceptance Model and related behavioral frameworks, which posit perceived usefulness as the primary driver of adoption behavior. These findings mirror international evidence showing that clinicians’ acceptance of generative AI is shaped far more by demonstrated clinical value than by novelty or general technological enthusiasm ( 14 , 15 ). Trust in ChatGPT also remained a dominant predictor in the adjusted model, with residents who reported moderate to high trust being more than four times as likely to report willingness to use the tool. This finding reinforces concerns surrounding hallucination, misinformation, and clinical safety that have been widely reported in previous studies ( 5 , 16 ). In environments where accountability structures for AI-supported decisions remain poorly defined, trust becomes not only a technical issue but also a legal and ethical one. Persistent uncertainty regarding medical liability and professional responsibility in cases of AI-related harm continues to undermine clinicians’ confidence, particularly in low- and middle-income health systems where regulatory guidance is still evolving ( 6 , 17 , 18 ). The independent effect of AI familiarity on willingness to use ChatGPT further highlights the critical role of education and structured exposure. Residents with higher familiarity were significantly more likely to express behavioral intention to use AI in practice. This finding aligns with evidence from medical education research showing that formal AI training markedly improves confidence, interpretability skills, and appropriate clinical use of intelligent systems ( 19 – 21 ). In contrast, the predominance of informal exposure in the Iranian context appears insufficient to establish durable clinical trust or responsible adoption practices. Specialty-based differences also persisted in the adjusted model, with surgical residents demonstrating significantly higher odds of willingness to use ChatGPT. This likely reflects greater exposure to technology-driven clinical environments, algorithm-supported workflows, and decision-support systems in procedural disciplines ( 22 – 25 ). The absence of a significant independent effect of gender in the regression model suggests that previously observed gender-based differences in trust may largely be mediated by familiarity and perceived usefulness rather than inherent attitudinal differences. Despite widespread belief in the potential benefits of ChatGPT, actual clinical recommendation behavior remained strikingly low. Only a small minority of residents reported ever recommending AI tools to patients. This behavioral gap reflects a broader pattern observed in other LMIC settings, where high conceptual acceptance coexists with limited real-world implementation due to regulatory ambiguity, institutional caution, and fear of ethical or legal repercussions ( 26 – 29 ). This discrepancy underscores that favorable attitudes do not automatically translate into clinical behavior in the absence of formal governance and institutional endorsement. Residents’ strong preference for low-risk use cases such as general health information and educational support, coupled with unanimous rejection of independent diagnostic use, further reflects professional caution and adherence to prevailing ethical standards. These attitudes closely align with international recommendations cautioning against unsupervised use of generative AI in high-stakes medical decision-making ( 30 ). The dominance of concerns related to accuracy, privacy, and legal responsibility in this study further reinforces that residents perceive AI-related risks as multidimensional and interconnected rather than isolated technical issues ( 18 , 31 ). Compared with high-income settings where formal AI curricula and regulatory pathways are increasingly embedded within medical education systems, the Iranian context remains at an early stage of institutional integration. Studies from technologically advanced countries have demonstrated substantially higher trust and acceptance among trainees who are exposed to structured educational interventions and supervised clinical AI applications ( 32 – 34 ). The absence of such frameworks in Iran likely explains the coexistence of optimism and deep skepticism observed among residents in this study. From a policy and educational perspective, these findings indicate that successful integration of LLMs into clinical practice cannot rely solely on technological availability. Structured AI literacy programs, clearly defined institutional policies, and national-level regulatory frameworks are essential to transform willingness into safe, ethical, and sustainable clinical use ( 20 , 35 , 36 ). Supervised pilot implementations in low-risk domains such as discharge summaries, guideline synthesis, and patient education may serve as controlled entry points for building experiential trust while minimizing patient safety risks ( 37 , 38 ). Several limitations should be considered when interpreting these findings. The cross-sectional design precludes causal inference, and attitudes were assessed using self-reported measures that may be influenced by social desirability or recall bias. The single-center nature of the study may limit generalizability to other training institutions with different technological infrastructures. Nonetheless, the large and diverse sample, rigorous instrument validation, and integration of both descriptive and predictive analyses substantially strengthen the credibility of the findings. In summary, medical residents in Iran appear to occupy an intermediate stage of AI adoption characterized by high awareness, cautious optimism, and persistent structural and ethical uncertainty. The regression findings clearly demonstrate that trust, perceived clinical benefit, and structured familiarity are the primary forces shaping adoption behavior. Without parallel investments in education, governance, and institutional safeguards, the clinical potential of large language models is unlikely to be realized safely or equitably within the Iranian health system or similar LMIC contexts. Conclusion This study provides timely evidence on Iranian medical residents’ early engagement with large language models such as ChatGPT in clinical settings. Although awareness was universal and most residents acknowledged potential benefits, trust remained limited and actual clinical use cautious. The findings confirm that awareness alone is insufficient to drive meaningful adoption. Multivariable regression analysis showed that trust in ChatGPT, belief in its potential to improve clinical outcomes, and familiarity with AI were the strongest independent predictors of willingness to use the technology. These results indicate that adoption is primarily driven by perceived value and professional confidence rather than by mere exposure. The persistence of concerns related to accuracy, privacy, and legal responsibility explains the observed gap between positive attitudes and real-world clinical recommendation. Overall, the safe and sustainable integration of large language models into clinical training in Iran requires structured AI education, formal institutional policies, and clear national regulatory frameworks to translate cautious optimism into responsible clinical practice. Declarations Ethics approval and consent to participate : This study was conducted in accordance with the principles of the Declaration of Helsinki . This research is approved by Ethics Committee of Mashhad University of Medical Sciences (IR.MUMS.FHMPM.REC.1403.253) Consent for publication : The authors declare their Consent for publication Availability of data and material : The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Competing interests : The authors declare that they have no competing interests. Funding : this research is part of master thesis in field of health care management supported by Mashhad University of medical sciences. (Gant NO. 4032023) Acknowledgements The research team would like to thank all Resident who participated in this research despite all the limitations Authors’ Contributions SZ and EH contributed equally to the conception and design of the study. SZ was involved in data collection and initial drafting of the manuscript. EH led the methodological design, statistical analysis, interpretation of results, and critical revision of the manuscript. MM supervised the study, contributed to questionnaire adaptation and validation, and provided substantial intellectual input throughout the research process. All authors reviewed and approved the final version of the manuscript and agree to be accountable for all aspects of the work. References Wang D, Zhang S. Large language models in medical and healthcare fields: applications, advances, and challenges. Artif Intell Rev. 2024;57(11):299. Jung KH. Large Language Models in Medicine: Clinical Applications, Technical Challenges, and Ethical Considerations. Healthc Inf Res. 2025;31(2):114–24. Nazi ZA, Peng W. Large Language Models in Healthcare and Medical Domain: A Review. Informatics. 2024;11(3):57. Chow JCL, Li K. Large Language Models in Medical Chatbots: Opportunities, Challenges, and the Need to Address AI Risks. Information. 2025;16(7):549. Jonnagaddala J, Wong ZS-Y. Privacy preserving strategies for electronic health records in the era of large language models. npj Digit Med. 2025;8(1):34. Zhang J, Zhang Z-m. Ethics and governance of trustworthy medical artificial intelligence. BMC Med Inf Decis Mak. 2023;23(1):7. Chang Y, Liu H, Lu C, Zhang N, SoK. Security and Privacy Risks of Healthcare AI. arXiv preprint arXiv:240907415. 2024. Topol EJ. High-performance medicine: the convergence of human and artificial intelligence. Nat Med. 2019;25(1):44–56. Jiang F, Jiang Y, Zhi H, Dong Y, Li H, Ma S et al. Artificial intelligence in healthcare: past, present and future. Stroke Vascular Neurol. 2017;2(4). Behzadifar M, Azari S, Sajedimehr N, Aalipour A, Nematkhah M, Teli BD, et al. Challenges of using artificial intelligence in Iran's health system: a qualitative study. J Prev Med Hyg. 2025;66(3):E331–40. Dehnavieh R, Inayatullah S, Yousefi F, Nadali M. Artificial Intelligence (AI) and the future of Iran’s Primary Health Care (PHC) system. BMC Prim Care. 2025;26(1):75. Nri-Ezedi CA, Ezidiegwu US, Ohuche IO, Musa S, Chikani UN, Okonkwo IO, et al. Integration of large language models in Nigeria's healthcare: doctors early perspectives, operational barriers, and privacy concerns. Discover Health Syst. 2024;3(1):104. Zhou J, Li H, Chen S, Chen Z, Han Z, Gao X. Large language models in biomedicine and healthcare. npj Artif Intell. 2025;1(1):44. DSouza H, Mohammed A, Almuntashiri AA, Kumar VVH, Mulla M, Khader AA, et al. Knowledge, awareness and attitude of healthcare professionals toward the use of ChatGpt: A cross-sectional study. J Educ health promotion. 2025;14(1):304. Dai Q, Li M, Yang M, Shi S, Wang Z, Liao J, et al. Attitudes, Perceptions, and Factors Influencing the Adoption of AI in Health Care Among Medical Staff: Nationwide Cross-Sectional Survey Study. J Med Internet Res. 2025;27:e75343. Alhur A, Sendi D, AlZahrani M, Abusharha L, Abudaak R, Alsinan R, EVALUATING THE TRUSTWORTHINESS OF CHATGPT-GENERATED HEALTH INFORMATION AMONG FUTURE HEALTH CARE PROFESSIONALS. Georgian Med News. 2025(364–365):101–6. Alanazi A, editor. editor Assessing Clinicians’ Legal Concerns and the Need for a Regulatory Framework for AI in Healthcare: A Mixed-Methods Study. Healthcare: MDPI; 2025. Botha NN, Segbedzi CE, Dumahasi VK, Maneen S, Kodom RV, Tsedze IS, et al. Artificial intelligence in healthcare: a scoping review of perceived threats to patient rights and safety. Archives Public Health. 2024;82(1):188. Wong LP. Artificial intelligence and job automation: challenges for secondary students’ career development and life planning. Merits. 2024;4(4):370–99. Ahsan Z. Integrating artificial intelligence into medical education: a narrative systematic review of current applications, challenges, and future directions. BMC Med Educ. 2025;25(1):1187. Duan S, Liu C, Rong T, Zhao Y, Liu B. Integrating AI in medical education: a comprehensive study of medical students’ attitudes, concerns, and behavioral intentions. BMC Med Educ. 2025;25(1):599. St John A, Cooper L, Kavic SM. The role of artificial intelligence in surgery: what do general surgery residents think? Am Surgeon™. 2024;90(4):541–9. Oh S, Kim JH, Choi S-W, Lee HJ, Hong J, Kwon SH. Physician confidence in artificial intelligence: an online mobile survey. J Med Internet Res. 2019;21(3):e12422. Alomran AK, Alomar MF, Akhdher AA, Al Qanber AR, Albik AK, Alumran A, et al. Artificial intelligence awareness and perceptions among pediatric orthopedic surgeons: A cross-sectional observational study. World J Orthop. 2024;15(11):1023. Arboit L, Schneider DN, Collins T, Hashimoto DA, Perretta S, Dallemagne B et al. Surgeons Awareness, Expectations, and Involvement with Artificial Intelligence: a Survey Pre and Post the GPT Era. arXiv preprint arXiv:250608258. 2025. Temsah M-H, Aljamaan F, Malki KH, Alhasan K, Altamimi I, Aljarbou R, et al. editors. ChatGPT and the future of digital health: a study on healthcare workers’ perceptions and expectations. Healthcare: MDPI; 2023. Ghanem OA, Hagag AM, Kormod ME, El-Refaay MA, Khedr AM, Abozaid OM, et al. Medical students’ knowledge, attitudes, and practices toward generative artificial intelligence in Egypt 2024: a Cross-Sectional study. BMC Med Educ. 2025;25(1):790. Rahul C, Karunakaran S, Karthikeyan V, Varadharajan S, Chandru S. Knowledge, attitudes and perceptions of AI in medicine among Indian doctors. Bioinformation. 2025;21(5):1061. Allam RM, Abdelfatah D, Khalil MIM, Elsaieed MM, El Desouky ED. Medical students and house officers’ perception, attitude and potential barriers towards artificial intelligence in Egypt, cross sectional survey. BMC Med Educ. 2024;24(1):1244. Hart SN, Hoffman NG, Gershkovich P, Christenson C, McClintock DS, Miller LJ, et al. Organizational preparedness for the use of large language models in pathology informatics. J Pathol Inf. 2023;14:100338. Aldossari M. Participants’ perceptions of privacy and data sharing regarding health-related data using artificial intelligence. Asian J Sci Appl Technol. 2023;12(2):6–12. Banerjee M, Chiew D, Patel KT, Johns I, Chappell D, Linton N, et al. The impact of artificial intelligence on clinical education: perceptions of postgraduate trainee doctors in London (UK) and recommendations for trainers. BMC Med Educ. 2021;21(1):429. Kong Z, Kong D, Kong J, Xing Y, Liang P. The performance evaluation of the AI-assisted diagnostic system in China. BMC Health Serv Res. 2025;25(1):1179. Wang Z. Artificial intelligence in Chinese healthcare: a review of applications and future prospects. Biomed Eng Lett. 2025:1–8. Rincón EHH, Jimenez D, Aguilar LAC, Flórez JMP, Tapia ÁER, Peñuela CLJ. Mapping the use of artificial intelligence in medical education: a scoping review. BMC Med Educ. 2025;25(1):526. Abdelwanis M, Simsekler MCE, Gabor AF, Sleptchenko A, Omar M. Artificial intelligence adoption challenges from healthcare providers’ perspectives: A comprehensive review and future directions. Saf Sci. 2026;193:107028. Jung H, Kim Y, Choi H, Seo H, Kim M, Han J et al. Enhancing clinical efficiency through llm: Discharge note generation for cardiac patients. arXiv preprint arXiv:240405144. 2024. Biswas A, Talukdar W. Intelligent clinical documentation: Harnessing generative AI for patient-centric clinical note generation. arXiv preprint arXiv:240518346. 2024. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8394385","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":567136691,"identity":"869c04fb-8587-4f9f-9afc-9335576bf1ce","order_by":0,"name":"Saeed Zahmatkesh","email":"","orcid":"","institution":"Mashhad University of Medical Sciences","correspondingAuthor":false,"prefix":"","firstName":"Saeed","middleName":"","lastName":"Zahmatkesh","suffix":""},{"id":567136692,"identity":"a9d68572-f890-4d3b-a0c6-44c7b8ad69b9","order_by":1,"name":"Elaheh Hooshmand","email":"","orcid":"","institution":"Mashhad University of Medical Sciences","correspondingAuthor":false,"prefix":"","firstName":"Elaheh","middleName":"","lastName":"Hooshmand","suffix":""},{"id":567136693,"identity":"30b22191-3e06-4cc1-9b5f-f531ce258ea9","order_by":2,"name":"Marziyhe Meraji","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAyklEQVRIiWNgGAWjYLACCQMGHn44j5lILTKSDSRpAQIbgwPEukm3vffhB4uCbTzG186YbvzBYCfPwM77AK8WszPHjSUkDG7zmN3OMbvNw5Bs2MDMboBfy400BoQWoEcSGJjZ8DvM7P4z5h8gLcazc8xu/mCoJ0LLDTY2sC0G0jlmN3gYDhOh5UwamwVIi8TttDKgxuOGbQS1HD/GfFviz217/tnJ227+qKiW5+c/hl8LCDBLwJnAsCJgBwQwfiBG1SgYBaNgFIxcAABiwzl8e4X1LQAAAABJRU5ErkJggg==","orcid":"","institution":"Mashhad University of Medical Sciences","correspondingAuthor":true,"prefix":"","firstName":"Marziyhe","middleName":"","lastName":"Meraji","suffix":""}],"badges":[],"createdAt":"2025-12-18 10:38:17","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8394385/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8394385/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":99288146,"identity":"1c03ad2e-4f5a-42d9-811c-7555d49ef978","added_by":"auto","created_at":"2025-12-31 09:40:10","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":63694,"visible":true,"origin":"","legend":"","description":"","filename":"Re.docx","url":"https://assets-eu.researchsquare.com/files/rs-8394385/v1/1bc5eba40c1a4a0e250fbef2.docx"},{"id":99321231,"identity":"66f54817-9fbb-4283-a83d-8f8bdab1396c","added_by":"auto","created_at":"2025-12-31 16:39:17","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5698,"visible":true,"origin":"","legend":"","description":"","filename":"6accf4e4543e4e7cb5bb15d5a0a6cdd6.json","url":"https://assets-eu.researchsquare.com/files/rs-8394385/v1/6a58db7389dbc7683d349d48.json"},{"id":99319317,"identity":"8d75e858-3fd8-4309-9699-453656f35e49","added_by":"auto","created_at":"2025-12-31 16:36:55","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":110696,"visible":true,"origin":"","legend":"","description":"","filename":"6accf4e4543e4e7cb5bb15d5a0a6cdd61enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8394385/v1/642faad413c11ca3f03f8282.xml"},{"id":99319336,"identity":"544c3fc9-a849-474a-85c3-819ef2ba4740","added_by":"auto","created_at":"2025-12-31 16:36:57","extension":"xml","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":108397,"visible":true,"origin":"","legend":"","description":"","filename":"6accf4e4543e4e7cb5bb15d5a0a6cdd61structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8394385/v1/356c66ad10e3b909d22a2db4.xml"},{"id":99320380,"identity":"4b8c344d-1613-4113-81a9-69002c53f9c3","added_by":"auto","created_at":"2025-12-31 16:38:33","extension":"html","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":120272,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8394385/v1/cb35aa1dbb0805b2fa0f89e4.html"},{"id":100563526,"identity":"ed38c75b-899b-4f1d-b81b-3adcfc4fa564","added_by":"auto","created_at":"2026-01-19 08:46:48","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":799120,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8394385/v1/88779b24-9cb3-40a9-921c-f221b25ae84d.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"\u003cp\u003eHigh Awareness, Low Trust: Medical Residents’ Readiness to Use ChatGPT in Clinical Practice in an LMIC Context\u003c/p\u003e","fulltext":[{"header":"Introduction","content":"\u003cp\u003eRecent advances in artificial intelligence\u0026mdash;particularly the emergence of Large Language Models (LLMs) such as GPT-based systems\u0026mdash;have created unprecedented opportunities for transforming medical education, clinical decision-making, and health service delivery (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). These models can summarize complex clinical information, support diagnostic reasoning, enhance patient communication, and streamline administrative and educational tasks. As their capabilities rapidly expand, LLMs are increasingly viewed as a cornerstone of the next generation of digital health innovation (\u003cspan additionalcitationids=\"CR2 CR3\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e). However, despite their remarkable potential, their integration into real-world clinical environments remains challenging.\u003c/p\u003e \u003cp\u003eOne of the most critical concerns surrounding the use of LLMs in healthcare is the phenomenon of \u003cem\u003ehallucination\u003c/em\u003e, in which models generate plausible but factually incorrect or clinically unsafe information (\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e). Such errors can compromise patient safety, misguide clinical decisions, and undermine trust among healthcare providers. Equally important are the ethical and regulatory implications, including data privacy, security of electronic health records, and uncertainty regarding legal accountability in cases of AI-related harm (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e). International frameworks such as the GDPR in Europe and HIPAA in the United States have begun outlining boundaries for responsible AI adoption. However, in many low- and middle-income countries (LMICs), including Iran, clear regulatory infrastructures for medical AI are still emerging, creating additional uncertainty for users and policymakers.\u003c/p\u003e \u003cp\u003eImportantly, physicians\u0026rsquo; acceptance of AI-based tools is deeply influenced by their prior exposure, perceived usability, and confidence in the accuracy and safety of these systems (\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e). Among all healthcare providers, medical residents represent a uniquely critical population for investigation:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eThey form the frontline of patient care in teaching hospitals,\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eThey frequently interact with clinical documentation systems,\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eThey rely heavily on up-to-date medical knowledge, and\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eTheir perceptions often forecast the future trajectory of technology adoption within the health system.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eDespite the exponential global growth of LLM-related research, evidence from Middle Eastern health systems remains scarce (\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). Iran, in particular, is at an early stage of integrating AI technologies into clinical workflows, and very few empirical studies have examined healthcare workers\u0026rsquo; readiness, trust, concerns, or ethical apprehensions toward LLMs. This gap is even more pronounced among medical residents, who are both primary end-users and future decision-makers in the digital health ecosystem.\u003c/p\u003e \u003cp\u003eMashhad University of Medical Sciences (MUMS), as one of the largest academic medical centers in eastern Iran, offers a distinctive context for such an investigation. The university hosts a diverse body of residents across a wide range of specialties, manages some of the busiest tertiary hospitals in the region, and has begun exploring digital health innovations despite limited structural and regulatory frameworks. Studying this population therefore provides not only a representative view of residents' perceptions in Iran but also valuable insights for LMICs facing similar challenges in adopting LLMs safely and ethically.\u003c/p\u003e \u003cp\u003eGiven the rapid expansion of LLMs, the growing global debate on their risks and benefits, and the absence of region-specific evidence, this study aims to assess medical residents\u0026rsquo; awareness, attitudes, operational barriers, and privacy-related concerns regarding the use of LLMs in healthcare. Understanding these perspectives is essential for shaping evidence-based policies, designing targeted training programs, and developing ethical frameworks to guide responsible AI adoption in clinical settings.\u003c/p\u003e"},{"header":"Method","content":"\u003cp\u003eThe study population included all medical residents enrolled in specialty training programs at MUMS during the 2023\u0026ndash;2024 academic year. Eligibility criteria were:\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eActive enrollment in a residency program,\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eWillingness to participate voluntarily, and\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eAbility to complete a Persian-language online questionnaire.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003eResidents who declined participation, submitted incomplete questionnaires (defined as more than 20% missing responses), or withdrew before completion were excluded from the final analysis.\u003c/p\u003e \u003cp\u003eA stratified non-random sampling strategy was used to ensure proportional representation across major clinical specialties. The strata were defined based on clinical specialty, and proportional allocation was applied according to the distribution of residents within each department. The survey link was distributed through official residency communication platforms, hospital-based academic groups, and institutional mailing lists. A total of 651 residents completed the questionnaire and were included in the final analysis.\u003c/p\u003e \u003cp\u003eData were collected using a structured questionnaire adapted from the instrument originally developed by Nri-Ezedi et al. (\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e), designed to assess physicians\u0026rsquo; awareness, attitudes, perceived benefits, operational barriers, and privacy concerns regarding the clinical use of Large Language Models (LLMs).\u003c/p\u003e \u003cp\u003eBecause the original questionnaire was developed in English, a standardized forward\u0026ndash;backward translation procedure was implemented to ensure linguistic and conceptual equivalence. Two independent bilingual translators performed the forward translation into Persian. A reconciled version was subsequently back-translated into English by two independent translators who were blinded to the original instrument.\u003c/p\u003e \u003cp\u003eAn expert panel consisting of two specialists in Medical Informatics and one specialist in Medical Ethics evaluated the translated items for clarity, relevance, and cultural appropriateness. Quantitative content validity indices demonstrated excellent validity:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eAll items achieved a Content Validity Ratio (CVR) of 1.00,\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eThe Item-level Content Validity Index (I-CVI) ranged from 0.83 to 1.00, and\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eThe Scale-level CVI/Average (S-CVI/Ave) was 0.92.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eA pilot study was conducted with 30 medical residents to evaluate the internal consistency and comprehensibility of the Persian version of the questionnaire. Minor linguistic refinements were applied based on participant feedback. The final instrument demonstrated acceptable internal reliability with a Cronbach\u0026rsquo;s alpha coefficient of 0.76, indicating satisfactory internal consistency.\u003c/p\u003e \u003cp\u003eData were collected electronically using Google Forms. The online questionnaire included an embedded informed consent form at the beginning of the survey. Participation was entirely voluntary, anonymous, and without financial or academic incentives. Participants were informed about the purpose of the study, confidentiality of the data, and their right to withdraw at any time without any consequences.\u003c/p\u003e \u003cp\u003eThe questionnaire assessed the following domains:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eDemographic characteristics (age, gender, specialty, years of experience, patient load)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eFamiliarity with artificial intelligence and ChatGPT\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eTrust in ChatGPT for clinical use\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eWillingness to use ChatGPT in healthcare practice\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003ePerceived benefits (e.g., access to information, efficiency)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003ePerceived risks (e.g., hallucination, privacy, legal responsibility)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eAppropriate clinical use-cases\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003ePrior experience and recommendation behavior toward AI tools\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eData analysis was performed using \u003cb\u003eSPSS version 22\u003c/b\u003e. Descriptive statistics, including frequencies and percentages, were used to summarize demographic characteristics and key study variables.\u003c/p\u003e \u003cp\u003eInferential analysis was conducted using:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eChi-square tests\u003c/b\u003e to examine associations between categorical variables, and\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eFisher\u0026rsquo;s Exact Test\u003c/b\u003e when expected cell counts were less than five.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eThe following associations were specifically tested:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eGender and trust in ChatGPT\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eSpecialty and familiarity with AI\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eClinical experience and concern about information accuracy\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eWillingness to use ChatGPT and belief in its potential to improve clinical outcomes\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eAll statistical tests were two-tailed, and statistical significance was defined as p\u0026thinsp;\u0026lt;\u0026thinsp;0.05.\u003c/p\u003e \u003cp\u003eAll ethical requirements for human research were strictly observed. Participation was voluntary, anonymity was maintained, and no identifiable personal data were collected. Participants were informed about the objectives of the study, data confidentiality, and their right to withdraw at any time.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eA total of \u003cb\u003e651 medical residents\u003c/b\u003e participated in the study. The demographic characteristics are presented in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eDemographic Characteristics of Participating Medical Residents (N\u0026thinsp;=\u0026thinsp;651)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVariable\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCategory\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eFrequency (%)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003e\u003cb\u003eAge\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e20\u0026ndash;29 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e267 (41.0%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e30\u0026ndash;39 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e225 (34.6%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e40\u0026ndash;49 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e134 (20.5%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026ge;\u0026thinsp;50 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e25 (3.8%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e\u003cb\u003eGender\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e376 (57.7%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e275 (42.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"4\" rowspan=\"5\"\u003e \u003cp\u003e\u003cb\u003eSpecialty\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSurgery\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e234 (35.9%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eInternal Medicine\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e167 (25.6%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePediatrics\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e116 (17.9%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eEmergency Medicine\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e92 (14.1%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOncology\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e42 (6.4%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003e\u003cb\u003eYears of Experience\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;5 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e275 (42.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5\u0026ndash;10 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e109 (16.7%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e10\u0026ndash;15 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e192 (29.5%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026gt;\u0026thinsp;15 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e75 (11.5%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003e\u003cb\u003eDaily Patient Visits\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;10 patients\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e75 (11.5%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e10\u0026ndash;20 patients\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e310 (47.7%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e20\u0026ndash;30 patients\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e183 (28.2%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026gt;\u0026thinsp;30 patients\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e83 (12.8%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe sample consisted predominantly of young residents, with more than 75% under the age of 40. This distribution indicates a workforce that is digitally exposed in personal life but may lack structured training in clinical AI. The strong presence of surgical and internal medicine residents provides insights across both procedural and cognitive specialties. Additionally, the substantial daily patient load observed among most participants highlights the potential relevance of time-saving digital tools in this context.\u003c/p\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e summarizes residents\u0026rsquo; awareness of and familiarity with artificial intelligence and ChatGPT.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eAwareness and Familiarity with AI and ChatGPT (N\u0026thinsp;=\u0026thinsp;651)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVariable\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCategory\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eFrequency (%)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003e\u003cb\u003eAI Familiarity\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHighly familiar\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e100 (15.4%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSomewhat familiar\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e367 (56.4%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNot very familiar\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e167 (25.6%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNot familiar at all\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e17 (2.6%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eHeard of ChatGPT\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e651 (100%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e\u003cb\u003eDescription of ChatGPT\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAI system\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e501 (76.9%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eVirtual assistant\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e150 (23.1%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMedical chatbot\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0 (0%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eWhile familiarity with AI varied, awareness of ChatGPT was universal. Residents mostly perceived ChatGPT as a general AI system rather than a clinical tool, revealing a conceptual gap between exposure and specialized understanding. This pattern is typical in settings where formal AI education is limited.Levels of trust in ChatGPT and residents\u0026rsquo; willingness to use the tool in clinical care are presented in Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eTrust and Willingness to Use ChatGPT in Clinical Care (N\u0026thinsp;=\u0026thinsp;651)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVariable\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCategory\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eFrequency (%)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003eTrust in ChatGPT\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eVery confident\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0 (0%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eModerately confident\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e250 (38.5%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNot very confident\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e334 (51.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNot confident at all\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e67 (10.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003eWillingness to Use\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eStrongly willing\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e84 (12.8%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSomewhat willing\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e267 (41.0%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNot very willing\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e217 (33.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNot willing at all\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e83 (12.8%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eNo resident expressed complete trust in ChatGPT, and more than half were not very confident in its clinical reliability. Despite this, moderate willingness to use the tool suggests a cautious openness\u0026mdash;residents see potential utility but remain concerned about risks, accuracy, and patient safety. Table 4 summarizes residents\u0026rsquo; perceptions of the potential advantages of ChatGPT use in healthcare\u003c/p\u003e\u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePerceived Advantages of Using ChatGPT in Healthcare (N\u0026thinsp;=\u0026thinsp;651)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAdvantage\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFrequency (%)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eImproving access to information\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e284 (43.6%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEnhancing communication\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e8 (1.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIncreasing clinical efficiency\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e234 (35.9%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCombined advantages\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e125 (19.2%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eResidents viewed ChatGPT primarily as an information-support tool. The extremely low emphasis on communication enhancement implies that residents do not yet associate LLMs with patient\u0026ndash;provider relational improvements, consistent with early-phase AI adoption literature. Residents\u0026rsquo; perceived disadvantages and risks associated with the use of ChatGPT are presented in Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e5\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePerceived Disadvantages and Risks of ChatGPT (N\u0026thinsp;=\u0026thinsp;651)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDisadvantage\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFrequency (%)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMisinterpretation of information\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e267 (41.0%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDependency on technology\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e117 (17.9%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eReduced human interaction\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e75 (11.5%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCombined disadvantages\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e192 (29.5%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eConcerns centered heavily on information accuracy, reflecting global apprehensions regarding AI hallucinations. The prominence of multi-dimensional concerns indicates that skepticism is driven by complex and interconnected risks rather than a single barrier. Table\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e6\u003c/span\u003e summarizes residents\u0026rsquo; views on appropriate clinical applications of ChatGPT.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab6\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 6\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eAppropriate Clinical Use-Cases for ChatGPT (N\u0026thinsp;=\u0026thinsp;651)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUse-Case\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFrequency (%)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eProviding general health information\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e392 (60.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDiagnostic use alone\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0 (0%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedication management\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e67 (10.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCombined use-cases\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e192 (29.5%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eParticipants strongly supported low-risk applications such as patient education but unanimously rejected ChatGPT as an independent diagnostic tool. This reflects professional caution and aligns with international recommendations advising against AI-driven diagnosis without supervision.\u003c/p\u003e \u003cp\u003eResidents\u0026rsquo; concerns regarding accuracy, privacy, and legal responsibility related to the use of ChatGPT are presented in Table\u0026nbsp;\u003cspan refid=\"Tab7\" class=\"InternalRef\"\u003e7\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab7\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 7\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePrivacy, Accuracy, and Legal Concerns (N\u0026thinsp;=\u0026thinsp;651)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eConcern\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFrequency (%)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAccuracy\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e184 (28.2%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrivacy and security\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e159 (24.4%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLegal responsibility\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e50 (7.7%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCombined concerns\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e258 (39.6%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eAccuracy and privacy collectively accounted for most concerns, emphasizing the need for robust regulatory frameworks in early-stage AI adoption environments. The high rate of combined concerns underscores that residents view AI risk as multi-factorial. Table\u0026nbsp;\u003cspan refid=\"Tab8\" class=\"InternalRef\"\u003e8\u003c/span\u003e summarizes residents\u0026rsquo; prior experience with ChatGPT, recommendation to patients, and perceived importance of ChatGPT-related knowledge.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab8\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 8\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePrior Use of ChatGPT and Perceived Importance (N\u0026thinsp;=\u0026thinsp;651)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVariable\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCategory\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eFrequency (%)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e\u003cb\u003eRecommended AI to patients\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e67 (10.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e584 (89.7%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e\u003cb\u003eImportance of ChatGPT knowledge\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eVery important\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e100 (15.4%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eModerately important\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e343 (52.6%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNot important\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e208 (32.0%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e\u003cb\u003eBelief in improved outcomes\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e526 (80.8%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e125 (19.2%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eAlthough most residents believed ChatGPT could enhance healthcare outcomes, very few had ever recommended AI tools to patients. This gap indicates that theoretical acceptance has not yet translated into real-world behavior. Table\u0026nbsp;\u003cspan refid=\"Tab9\" class=\"InternalRef\"\u003e9\u003c/span\u003e presents the results of chi-square analyses assessing associations between demographic and attitudinal variables related to ChatGPT use.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab9\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 9\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eInferential Analysis of Key Associations (Chi-Square Tests, N\u0026thinsp;=\u0026thinsp;651)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eComparison\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eχ\u0026sup2;\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003edf\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ep-value\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eInterpretation\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGender \u0026times; Trust\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e8.71\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.033\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSignificant\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSpecialty \u0026times; AI familiarity\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e21.54\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.043\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSignificant\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eExperience \u0026times; Accuracy concern\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e14.62\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.101\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eNot significant\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWillingness \u0026times; Belief in improved outcomes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e19.83\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eHighly significant\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eTrust varied significantly by gender, and familiarity differed by specialty. Experience did not influence accuracy concerns, implying widespread skepticism across seniority levels. The strongest relationship was between willingness to use AI and belief in its benefits, emphasizing the central role of perceived value in adoption decisions.\u003c/p\u003e \u003cp\u003eTo identify independent predictors of \u003cb\u003eresidents\u0026rsquo;\u003c/b\u003e willingness to use ChatGPT in clinical practice, \u003cb\u003ea\u003c/b\u003e binary logistic regression analysis was performed. The dependent variable was defined as willingness to use ChatGPT (coded as 1\u0026thinsp;=\u0026thinsp;willing [strongly or somewhat willing], and 0\u0026thinsp;=\u0026thinsp;not willing [not very willing or not willing at all]).\u003c/p\u003e \u003cp\u003eIndependent variables entered into the model included:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eGender\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eSpecialty\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eLevel of familiarity with AI\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eTrust in ChatGPT\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eBelief in ChatGPT\u0026rsquo;s potential to improve clinical outcomes\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eThe results of the multivariable logistic regression analysis predicting residents\u0026rsquo; willingness to use ChatGPT in clinical practice are presented in Table\u0026nbsp;\u003cspan refid=\"Tab10\" class=\"InternalRef\"\u003e10\u003c/span\u003e.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab10\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 10\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eMultivariable Logistic Regression Predicting Willingness to Use ChatGPT (N\u0026thinsp;=\u0026thinsp;651)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003ePredictor\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAdjusted OR\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e95% CI\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003ep-value\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003eMale gender\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.28\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.91\u0026ndash;1.80\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.148\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003eSurgical specialty\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.94\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.31\u0026ndash;2.89\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003eHigh AI familiarity\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.63\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.74\u0026ndash;3.99\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003eModerate/High trust in ChatGPT\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e4.11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.88\u0026ndash;5.86\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003eBelief in improved outcomes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e5.27\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.46\u0026ndash;8.02\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"5\"\u003eModel fit:\u003c/td\u003e\u003c/tr\u003e \u003ctr\u003e\u003ctd colspan=\"5\"\u003eHosmer\u0026ndash;Lemeshow test: p\u0026thinsp;=\u0026thinsp;0.41\u003c/td\u003e\u003c/tr\u003e \u003ctr\u003e\u003ctd colspan=\"5\"\u003eNagelkerke R\u0026sup2; = 0.38\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eMultivariable logistic regression analysis demonstrated that trust in ChatGPT and belief in its potential clinical benefits were the strongest independent predictors of residents\u0026rsquo; willingness to use the tool in practice. Residents who expressed moderate to high trust in ChatGPT were 4.11 times more likely to report willingness to use it clinically (OR\u0026thinsp;=\u0026thinsp;4.11, 95% CI: 2.88\u0026ndash;5.86, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/p\u003e \u003cp\u003eSimilarly, those who believed that ChatGPT could improve clinical outcomes were more than five times as likely to be willing users (OR\u0026thinsp;=\u0026thinsp;5.27, 95% CI: 3.46\u0026ndash;8.02, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/p\u003e \u003cp\u003eHigher familiarity with AI also significantly increased willingness to use ChatGPT (OR\u0026thinsp;=\u0026thinsp;2.63, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Surgical residents were nearly twice as likely to report willingness compared with non-surgical specialties (OR\u0026thinsp;=\u0026thinsp;1.94, p\u0026thinsp;=\u0026thinsp;0.001). Gender was not a significant independent predictor in the adjusted model.\u003c/p\u003e \u003cp\u003eThe Hosmer\u0026ndash;Lemeshow test indicated acceptable model fit (p\u0026thinsp;=\u0026thinsp;0.41), and the model explained approximately 38% of the variance in willingness to use ChatGPT.\u003c/p\u003e \u003cp\u003eAll predictors were entered simultaneously using the Enter method. The goodness-of-fit of the model was assessed using the Hosmer\u0026ndash;Lemeshow test, and adjusted odds ratios (ORs) with 95% confidence intervals (CIs) were calculated. A p-value of \u0026lt;\u0026thinsp;0.05 was considered statistically significant.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eMedical residents\u0026rsquo; perceptions of large language models, particularly ChatGPT, reflect a complex and transitional stage between widespread awareness and cautious clinical engagement. Although awareness of ChatGPT was universal in this study, both descriptive and regression findings demonstrate that awareness alone is insufficient to drive adoption. Trust and perceived clinical usefulness emerged as the most influential determinants of willingness to use ChatGPT, underscoring that psychological confidence and perceived performance enhancement remain the central pillars of technology acceptance in clinical environments (\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe multivariable regression model provides important insights beyond bivariate associations by demonstrating that belief in ChatGPT\u0026rsquo;s potential to improve clinical outcomes was the strongest independent predictor of willingness to use the technology. Residents holding this belief were more than five times as likely to express willingness for clinical use. This strong predictive effect aligns closely with the core assumptions of the Technology Acceptance Model and related behavioral frameworks, which posit perceived usefulness as the primary driver of adoption behavior. These findings mirror international evidence showing that clinicians\u0026rsquo; acceptance of generative AI is shaped far more by demonstrated clinical value than by novelty or general technological enthusiasm (\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eTrust in ChatGPT also remained a dominant predictor in the adjusted model, with residents who reported moderate to high trust being more than four times as likely to report willingness to use the tool. This finding reinforces concerns surrounding hallucination, misinformation, and clinical safety that have been widely reported in previous studies (\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e). In environments where accountability structures for AI-supported decisions remain poorly defined, trust becomes not only a technical issue but also a legal and ethical one. Persistent uncertainty regarding medical liability and professional responsibility in cases of AI-related harm continues to undermine clinicians\u0026rsquo; confidence, particularly in low- and middle-income health systems where regulatory guidance is still evolving (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe independent effect of AI familiarity on willingness to use ChatGPT further highlights the critical role of education and structured exposure. Residents with higher familiarity were significantly more likely to express behavioral intention to use AI in practice. This finding aligns with evidence from medical education research showing that formal AI training markedly improves confidence, interpretability skills, and appropriate clinical use of intelligent systems (\u003cspan additionalcitationids=\"CR20\" citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e). In contrast, the predominance of informal exposure in the Iranian context appears insufficient to establish durable clinical trust or responsible adoption practices.\u003c/p\u003e \u003cp\u003eSpecialty-based differences also persisted in the adjusted model, with surgical residents demonstrating significantly higher odds of willingness to use ChatGPT. This likely reflects greater exposure to technology-driven clinical environments, algorithm-supported workflows, and decision-support systems in procedural disciplines (\u003cspan additionalcitationids=\"CR23 CR24\" citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e). The absence of a significant independent effect of gender in the regression model suggests that previously observed gender-based differences in trust may largely be mediated by familiarity and perceived usefulness rather than inherent attitudinal differences.\u003c/p\u003e \u003cp\u003eDespite widespread belief in the potential benefits of ChatGPT, actual clinical recommendation behavior remained strikingly low. Only a small minority of residents reported ever recommending AI tools to patients. This behavioral gap reflects a broader pattern observed in other LMIC settings, where high conceptual acceptance coexists with limited real-world implementation due to regulatory ambiguity, institutional caution, and fear of ethical or legal repercussions (\u003cspan additionalcitationids=\"CR27 CR28\" citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e). This discrepancy underscores that favorable attitudes do not automatically translate into clinical behavior in the absence of formal governance and institutional endorsement.\u003c/p\u003e \u003cp\u003e Residents\u0026rsquo; strong preference for low-risk use cases such as general health information and educational support, coupled with unanimous rejection of independent diagnostic use, further reflects professional caution and adherence to prevailing ethical standards. These attitudes closely align with international recommendations cautioning against unsupervised use of generative AI in high-stakes medical decision-making (\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e). The dominance of concerns related to accuracy, privacy, and legal responsibility in this study further reinforces that residents perceive AI-related risks as multidimensional and interconnected rather than isolated technical issues (\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eCompared with high-income settings where formal AI curricula and regulatory pathways are increasingly embedded within medical education systems, the Iranian context remains at an early stage of institutional integration. Studies from technologically advanced countries have demonstrated substantially higher trust and acceptance among trainees who are exposed to structured educational interventions and supervised clinical AI applications (\u003cspan additionalcitationids=\"CR33\" citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e). The absence of such frameworks in Iran likely explains the coexistence of optimism and deep skepticism observed among residents in this study.\u003c/p\u003e \u003cp\u003eFrom a policy and educational perspective, these findings indicate that successful integration of LLMs into clinical practice cannot rely solely on technological availability. Structured AI literacy programs, clearly defined institutional policies, and national-level regulatory frameworks are essential to transform willingness into safe, ethical, and sustainable clinical use (\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e, \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e). Supervised pilot implementations in low-risk domains such as discharge summaries, guideline synthesis, and patient education may serve as controlled entry points for building experiential trust while minimizing patient safety risks (\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eSeveral limitations should be considered when interpreting these findings. The cross-sectional design precludes causal inference, and attitudes were assessed using self-reported measures that may be influenced by social desirability or recall bias. The single-center nature of the study may limit generalizability to other training institutions with different technological infrastructures. Nonetheless, the large and diverse sample, rigorous instrument validation, and integration of both descriptive and predictive analyses substantially strengthen the credibility of the findings.\u003c/p\u003e \u003cp\u003eIn summary, medical residents in Iran appear to occupy an intermediate stage of AI adoption characterized by high awareness, cautious optimism, and persistent structural and ethical uncertainty. The regression findings clearly demonstrate that trust, perceived clinical benefit, and structured familiarity are the primary forces shaping adoption behavior. Without parallel investments in education, governance, and institutional safeguards, the clinical potential of large language models is unlikely to be realized safely or equitably within the Iranian health system or similar LMIC contexts.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThis study provides timely evidence on Iranian medical residents\u0026rsquo; early engagement with large language models such as ChatGPT in clinical settings. Although awareness was universal and most residents acknowledged potential benefits, trust remained limited and actual clinical use cautious. The findings confirm that awareness alone is insufficient to drive meaningful adoption.\u003c/p\u003e \u003cp\u003eMultivariable regression analysis showed that trust in ChatGPT, belief in its potential to improve clinical outcomes, and familiarity with AI were the strongest independent predictors of willingness to use the technology. These results indicate that adoption is primarily driven by perceived value and professional confidence rather than by mere exposure.\u003c/p\u003e \u003cp\u003eThe persistence of concerns related to accuracy, privacy, and legal responsibility explains the observed gap between positive attitudes and real-world clinical recommendation. Overall, the safe and sustainable integration of large language models into clinical training in Iran requires structured AI education, formal institutional policies, and clear national regulatory frameworks to translate cautious optimism into responsible clinical practice.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e:\u0026nbsp;This study was conducted in accordance with the principles of the \u003cstrong\u003eDeclaration of Helsinki\u003c/strong\u003e. This research is approved by Ethics Committee of Mashhad University of Medical Sciences (IR.MUMS.FHMPM.REC.1403.253)\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e: The authors declare their Consent for publication\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and material\u003c/strong\u003e: The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e: The authors declare that they have no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e: this research is part of master thesis in field of health care management supported by Mashhad University of medical sciences. (Gant NO. 4032023)\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe research team would like to thank all Resident who participated in this research despite all the limitations\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026rsquo; Contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSZ\u003c/strong\u003e and \u003cstrong\u003eEH\u003c/strong\u003e contributed equally to the conception and design of the study. \u003cstrong\u003eSZ\u003c/strong\u003e was involved in data collection and initial drafting of the manuscript. \u003cstrong\u003eEH\u003c/strong\u003e led the methodological design, statistical analysis, interpretation of results, and critical revision of the manuscript. \u003cstrong\u003eMM\u003c/strong\u003e supervised the study, contributed to questionnaire adaptation and validation, and provided substantial intellectual input throughout the research process. All authors reviewed and approved the final version of the manuscript and agree to be accountable for all aspects of the work.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eWang D, Zhang S. Large language models in medical and healthcare fields: applications, advances, and challenges. Artif Intell Rev. 2024;57(11):299.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJung KH. Large Language Models in Medicine: Clinical Applications, Technical Challenges, and Ethical Considerations. Healthc Inf Res. 2025;31(2):114\u0026ndash;24.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNazi ZA, Peng W. Large Language Models in Healthcare and Medical Domain: A Review. Informatics. 2024;11(3):57.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChow JCL, Li K. Large Language Models in Medical Chatbots: Opportunities, Challenges, and the Need to Address AI Risks. Information. 2025;16(7):549.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJonnagaddala J, Wong ZS-Y. Privacy preserving strategies for electronic health records in the era of large language models. npj Digit Med. 2025;8(1):34.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang J, Zhang Z-m. Ethics and governance of trustworthy medical artificial intelligence. BMC Med Inf Decis Mak. 2023;23(1):7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChang Y, Liu H, Lu C, Zhang N, SoK. Security and Privacy Risks of Healthcare AI. arXiv preprint arXiv:240907415. 2024.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTopol EJ. High-performance medicine: the convergence of human and artificial intelligence. Nat Med. 2019;25(1):44\u0026ndash;56.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJiang F, Jiang Y, Zhi H, Dong Y, Li H, Ma S et al. Artificial intelligence in healthcare: past, present and future. Stroke Vascular Neurol. 2017;2(4).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBehzadifar M, Azari S, Sajedimehr N, Aalipour A, Nematkhah M, Teli BD, et al. Challenges of using artificial intelligence in Iran's health system: a qualitative study. J Prev Med Hyg. 2025;66(3):E331\u0026ndash;40.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDehnavieh R, Inayatullah S, Yousefi F, Nadali M. Artificial Intelligence (AI) and the future of Iran\u0026rsquo;s Primary Health Care (PHC) system. BMC Prim Care. 2025;26(1):75.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNri-Ezedi CA, Ezidiegwu US, Ohuche IO, Musa S, Chikani UN, Okonkwo IO, et al. Integration of large language models in Nigeria's healthcare: doctors early perspectives, operational barriers, and privacy concerns. Discover Health Syst. 2024;3(1):104.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhou J, Li H, Chen S, Chen Z, Han Z, Gao X. Large language models in biomedicine and healthcare. npj Artif Intell. 2025;1(1):44.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDSouza H, Mohammed A, Almuntashiri AA, Kumar VVH, Mulla M, Khader AA, et al. Knowledge, awareness and attitude of healthcare professionals toward the use of ChatGpt: A cross-sectional study. J Educ health promotion. 2025;14(1):304.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDai Q, Li M, Yang M, Shi S, Wang Z, Liao J, et al. Attitudes, Perceptions, and Factors Influencing the Adoption of AI in Health Care Among Medical Staff: Nationwide Cross-Sectional Survey Study. J Med Internet Res. 2025;27:e75343.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlhur A, Sendi D, AlZahrani M, Abusharha L, Abudaak R, Alsinan R, EVALUATING THE TRUSTWORTHINESS OF CHATGPT-GENERATED HEALTH INFORMATION AMONG FUTURE HEALTH CARE PROFESSIONALS. Georgian Med News. 2025(364\u0026ndash;365):101\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlanazi A, editor. editor Assessing Clinicians\u0026rsquo; Legal Concerns and the Need for a Regulatory Framework for AI in Healthcare: A Mixed-Methods Study. Healthcare: MDPI; 2025.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBotha NN, Segbedzi CE, Dumahasi VK, Maneen S, Kodom RV, Tsedze IS, et al. Artificial intelligence in healthcare: a scoping review of perceived threats to patient rights and safety. Archives Public Health. 2024;82(1):188.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWong LP. Artificial intelligence and job automation: challenges for secondary students\u0026rsquo; career development and life planning. Merits. 2024;4(4):370\u0026ndash;99.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAhsan Z. Integrating artificial intelligence into medical education: a narrative systematic review of current applications, challenges, and future directions. BMC Med Educ. 2025;25(1):1187.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDuan S, Liu C, Rong T, Zhao Y, Liu B. Integrating AI in medical education: a comprehensive study of medical students\u0026rsquo; attitudes, concerns, and behavioral intentions. BMC Med Educ. 2025;25(1):599.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSt John A, Cooper L, Kavic SM. The role of artificial intelligence in surgery: what do general surgery residents think? Am Surgeon\u0026trade;. 2024;90(4):541\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOh S, Kim JH, Choi S-W, Lee HJ, Hong J, Kwon SH. Physician confidence in artificial intelligence: an online mobile survey. J Med Internet Res. 2019;21(3):e12422.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlomran AK, Alomar MF, Akhdher AA, Al Qanber AR, Albik AK, Alumran A, et al. Artificial intelligence awareness and perceptions among pediatric orthopedic surgeons: A cross-sectional observational study. World J Orthop. 2024;15(11):1023.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eArboit L, Schneider DN, Collins T, Hashimoto DA, Perretta S, Dallemagne B et al. Surgeons Awareness, Expectations, and Involvement with Artificial Intelligence: a Survey Pre and Post the GPT Era. arXiv preprint arXiv:250608258. 2025.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTemsah M-H, Aljamaan F, Malki KH, Alhasan K, Altamimi I, Aljarbou R, et al. editors. ChatGPT and the future of digital health: a study on healthcare workers\u0026rsquo; perceptions and expectations. Healthcare: MDPI; 2023.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGhanem OA, Hagag AM, Kormod ME, El-Refaay MA, Khedr AM, Abozaid OM, et al. Medical students\u0026rsquo; knowledge, attitudes, and practices toward generative artificial intelligence in Egypt 2024: a Cross-Sectional study. BMC Med Educ. 2025;25(1):790.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRahul C, Karunakaran S, Karthikeyan V, Varadharajan S, Chandru S. Knowledge, attitudes and perceptions of AI in medicine among Indian doctors. Bioinformation. 2025;21(5):1061.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAllam RM, Abdelfatah D, Khalil MIM, Elsaieed MM, El Desouky ED. Medical students and house officers\u0026rsquo; perception, attitude and potential barriers towards artificial intelligence in Egypt, cross sectional survey. BMC Med Educ. 2024;24(1):1244.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHart SN, Hoffman NG, Gershkovich P, Christenson C, McClintock DS, Miller LJ, et al. Organizational preparedness for the use of large language models in pathology informatics. J Pathol Inf. 2023;14:100338.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAldossari M. Participants\u0026rsquo; perceptions of privacy and data sharing regarding health-related data using artificial intelligence. Asian J Sci Appl Technol. 2023;12(2):6\u0026ndash;12.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBanerjee M, Chiew D, Patel KT, Johns I, Chappell D, Linton N, et al. The impact of artificial intelligence on clinical education: perceptions of postgraduate trainee doctors in London (UK) and recommendations for trainers. BMC Med Educ. 2021;21(1):429.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKong Z, Kong D, Kong J, Xing Y, Liang P. The performance evaluation of the AI-assisted diagnostic system in China. BMC Health Serv Res. 2025;25(1):1179.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang Z. Artificial intelligence in Chinese healthcare: a review of applications and future prospects. Biomed Eng Lett. 2025:1\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRinc\u0026oacute;n EHH, Jimenez D, Aguilar LAC, Fl\u0026oacute;rez JMP, Tapia \u0026Aacute;ER, Pe\u0026ntilde;uela CLJ. Mapping the use of artificial intelligence in medical education: a scoping review. BMC Med Educ. 2025;25(1):526.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAbdelwanis M, Simsekler MCE, Gabor AF, Sleptchenko A, Omar M. Artificial intelligence adoption challenges from healthcare providers\u0026rsquo; perspectives: A comprehensive review and future directions. Saf Sci. 2026;193:107028.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJung H, Kim Y, Choi H, Seo H, Kim M, Han J et al. Enhancing clinical efficiency through llm: Discharge note generation for cardiac patients. arXiv preprint arXiv:240405144. 2024.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBiswas A, Talukdar W. Intelligent clinical documentation: Harnessing generative AI for patient-centric clinical note generation. arXiv preprint arXiv:240518346. 2024.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"ChatGPT, Large Language Models, Medical residents, Trust, AI acceptance, Digital health, Iran, Clinical education","lastPublishedDoi":"10.21203/rs.3.rs-8394385/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8394385/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eBackground: Large language models such as ChatGPT are rapidly entering medical education and clinical practice, yet evidence on clinicians’ readiness to adopt these tools in low- and middle-income countries remains limited. This study aimed to assess medical residents’ awareness, trust, perceptions, and predictors of willingness to use ChatGPT in clinical settings in Iran.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eMethods: A cross-sectional survey was conducted among 651 medical residents at Mashhad University of Medical Sciences between December 2023 and March 2024. A validated Persian questionnaire assessed familiarity with AI, trust in ChatGPT, perceived benefits and risks, and willingness to use the tool. Descriptive statistics, Chi-square tests, and multivariable logistic regression were used for data analysis.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eResults: Awareness of ChatGPT was universal; however, trust remained limited. Most residents supported the use of ChatGPT for low-risk tasks such as general health information, while none endorsed its independent diagnostic use. Multivariable regression showed that trust in ChatGPT, belief in its potential to improve clinical outcomes, and higher AI familiarity were the strongest independent predictors of willingness to use the technology. Concerns related to accuracy, privacy, and legal responsibility remained the dominant barriers to clinical adoption.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eConclusion: Iranian medical residents demonstrate cautious optimism toward ChatGPT, but limited trust and regulatory uncertainty restrict real-world use. Strengthening AI education, establishing institutional policies, and developing national regulatory frameworks are essential to enable the safe and effective integration of large language models into clinical training and practice.\u003c/p\u003e","manuscriptTitle":"High Awareness, Low Trust: Medical Residents’ Readiness to Use ChatGPT in Clinical Practice in an LMIC Context","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-12-31 09:40:04","doi":"10.21203/rs.3.rs-8394385/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"f5865a6c-cba7-440d-85c5-687068b3b7dd","owner":[],"postedDate":"December 31st, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-01-19T08:44:00+00:00","versionOfRecord":[],"versionCreatedAt":"2025-12-31 09:40:04","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8394385","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8394385","identity":"rs-8394385","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}