From Comprehension to Behaviour: A PROLIFERATE_AI Secondary Analysis of RAPIDx AI Adoption Across South Australian Emergency Departments

Background: While the RAPIDx AI RCT demonstrated the clinical performance of an AI-assisted pathway for chest pain assessment, real-world impact depends on sustained adoption by those who use it. Evidence from complex systems shows that comprehension of a tool’s function rarely ensures its continued use. This study applied PROLIFERATE_AI, a human-centred multimethod framework, to identify behavioural and contextual drivers of adoption across 12 South Australian emergency departments.

Methods: This secondary analysis used de-identified end-user data from clinicians, nurses, trainees, and technical staff participating in the RAPIDx AI RCT (ACTRN12620001319965; NHMRC GNT1191914; SACHREC OfR 272.20). The PROLIFERATE_AI survey measured five adoption constructs: Comprehension, Barriers, Motivation, Engagement, and Adoption. Quantitative analyses included correlation matrices, ridge regression, and role-based comparisons. Free-text responses were analysed using sentiment classification to contextualise quantitative findings.

Results: Adoption scores were similar across roles (mean ≈ 52–54 / 100). Technical staff showed the highest comprehension (63 ± 20), whereas nurses and registrars exhibited stronger engagement (56 ± 16). Comprehension showed negligible correlation with other constructs (r ≈ –0.01 to 0.30) and was excluded from the main model. Adoption correlated most strongly with Barriers (r = 0.64), Motivation (r = 0.59), and Engagement (r = 0.53), confirming that sustained use is primarily behavioural and contextual rather than cognitive.

Conclusions: Clinicians adopt AI in cardiac care when it fits workflows, fosters trust, and sustains engagement, not merely when understood. Adoption is thus behavioural and contextual. Embedding motivational design and co-created workflows can strengthen sustained integration. PROLIFERATE_AI offers a robust, human-centred framework to quantify and optimise adoption before scale-up, bridging pilot success and real-world cardiac impact.