AWARE_AI, a Machine‐Learning–Based Reproducible Workflow Modelling the Fundamentals of Care Framework and Patient‐Reported Wellbeing: A Cross‐Sectional Feasibility Study

Maria Alejandra Piñero de Plaza, Alison Kitson, Ana María Porcel‐Gálvez, Renuka Visvanathan, Alline Beleigoli, Carlos Javier Barrera Causil, Fernando Marmolejo Ramos, Regina Allande‐Cussó
Journal of Advanced Nursing · Wiley · 2026-11-06 · DOI: 10.1111/jan.70800

Abstract

ABSTRACT Aims To examine whether theory‐ordered Fundamentals of Care domain associations are present in patient‐reported data and to assess the feasibility of a reproducible machine‐learning‐based workflow for modelling these associations and wellbeing in an Australian pilot sample. Design Secondary cross‐sectional feasibility analysis. Methods We analysed de‐identified survey data from 36 adults who had received healthcare in the previous 12 months using a frozen Python workflow. The workflow estimated theory‐ordered bivariate associations among Context of Care, Relationship, Integration of Care, and a 12‐item study‐specific wellbeing index, assessed out‐of‐sample prediction and explored item‐level patterns using penalised regression. Feasibility was assessed across process, analytical, predictive, and reproducibility dimensions. Results Theory‐ordered bivariate associations were positive and consistent with the proposed Fundamentals of Care ordering. Context of Care was associated with Relationship, Relationship with Integration of Care and Integration of Care with wellbeing. Their descriptive serial product was positive but does not support mediation or causal interpretation. Process, analytical and reproducibility feasibility criteria were met, whereas predictive feasibility was not. Out‐of‐sample prediction was negligible and unstable. Associations were strongest for evaluative wellbeing indicators and weaker for affective and health‐status indicators. Conclusion Care context, relational quality, integrative care and patient‐reported wellbeing co‐varied in a pattern consistent with the FoC framework. The cross‐sectional design does not establish a conditional pathway, causation or individual‐level prediction. AWARE_AI provides a transparent foundation for larger longitudinal validation. Impact Clinicians and leaders should treat item‐level patterns as hypotheses for future testing, not actionable determinants. Researchers can reuse the workflow in larger longitudinal studies with wellbeing measured before and after defined care episodes. Reporting Method We followed STROBE and TRIPOD+AI. Patient or Public Contribution This secondary analysis did not include patient or public involvement in study design, conduct, analysis, interpretation or manuscript preparation. Patients contributed data as survey respondents.

Read at publisher →

Similar articles