Although modern advancements, such as the emergence of artificial intelligence (AI) tools, have enabled the integration of cutting-edge technologies into various studies, scientific inquiry and research cannot bypass established standards and values, as they are inherently bound by ethical principles. This study aimed to develop a model for publication ethics compliance in AI-driven research within Iranian universities of medical sciences.
This study employed an exploratory sequential mixed-methods design and was applied in terms of its purpose. In the qualitative phase, semi-structured interviews were conducted with 20 academic actors in the research system at both macro levels (directors, policymakers, and research vice-chancellors of the Ministry of Health and universities) and micro levels (editors-in-chief, peer reviewers, editorial board members, and researchers in the fields of research ethics and AI). Purposive sampling was performed using critical case and snowball strategies. Qualitative data were analyzed using thematic analysis in MAXQDA (version 20), resulting in the extraction of compliance dimensions across four categories: individual, organizational, environmental, and professional competencies. In the quantitative phase, the proposed model was operationalized into a researcher-developed questionnaire with 34 items and completed by 124 experts in research ethics and AI. The data were analyzed using partial least squares structural equation modeling (PLS-SEM) in SmartPLS (version 4) to evaluate the model fit.
The findings indicated that the factors influencing the ethical application of artificial intelligence tools in the publication of research results fall into four categories: individual (52 interpretive codes), organizational (145 interpretive codes), environmental (51 interpretive codes), and professional competencies (76 interpretive codes). The results also showed that the construct of adherence to publication ethics in AI‑based medical research has a positive and significant effect on all four dimensions. Among these, the organizational dimension demonstrated the strongest effect, with a path coefficient of 0.955 and a t‑value of 110.231, indicating that organizational factors play the most influential role in explaining adherence to publication ethics in AI‑based medical research. This was followed by the individual (0.953), environmental (0.882), and professional competencies (0.881) dimensions. Since the t‑values for all relationships were considerably higher than 1.96 and the p‑values were reported as 0.000, all paths in the model were statistically significant at the 95% confidence level. These findings indicate that the structural model possesses very strong explanatory power.
Publication ethics compliance in AI-driven medical research is a multidimensional phenomenon shaped by interrelated factors. The findings of this study can serve as an operational framework for ethical policymaking, the formulation of national guidelines, and the design of researcher empowerment programs in universities of medical sciences.