Artificial Intelligence Readiness and Technology Adoption among Healthcare Professionals in Bangladesh: Evidence from Partial Least Squares Structural Equation Modeling

Artificial intelligence (AI) presents possibilities of enhancing healthcare information processing, documentation, professional decision support, and coordination within an organization. However, the presence of technology is not always associated with readiness or long-term professional application. This paper looks into the associations between perceived AI literacy, organizational support, inadequacy of technological infrastructure, perceived usefulness, perceived risk, AI readiness, and self-reported AI technology adoption behavior among healthcare professionals in Bangladesh. The quantitative cross-sectional survey design was modelled through analysing partial least squares structural equations modelling (PLS-SEM). The results of the analysis include 460 respondents, seven reflective constructs, and 28 measurement indicators. The measurement model has reasonable internal consistency, convergent validity and discriminant validity. The reported estimates support all the seven specified structural hypotheses. AI readiness is positively related to perceived AI literacy (β = 0.257), organizational support (β = 0.219), technological infrastructure (β = 0.186), and perceived usefulness (β = 0.224) and is negatively related to perceived risk (β = 0.222). There exists a positive relationship between AI readiness and technology adoption (β = 0.489), and a direct correlation between the adoption and perceived usefulness (β = 0.270). The structural model describes 45.1 % of difference in preparedness and 42.9 % in the adoption behavior. There exist five statistically significant indirect relationships between readiness and organizational, individual, and technological perceptions and reported professional AI use. Its findings indicate that responsible adoption needs to be supported by concomitant investments in workforce capabilities, implementation support, infrastructure, and risk management. Self-reported data at one point in time restricts the ability to make causal interpretations and must be generalized carefully.