Artificial intelligence-supported revision of nursing care plans in a standardised newborn vignette: a one-group pretest–posttest study

Background: Nursing students often produce care plans that are plausible yet generic, with weak linkage among diagnoses, interventions and outcomes. Teacher-mediated, bounded generative AI may support revision while preserving clinical judgement. Aims: To assess whether an AI-supported Reflect–Compare–Revise workflow improves nursing care plan quality and diagnosis–intervention–outcome alignment using a standardised newborn vignette. Methods: A one-group pretest–posttest study involved 40 second-year nursing students at a public university in the Philippines. Students wrote an unaided care plan, then revised it using the structured workflow. Two blinded nurse-educator raters scored paired plans using the Quality of Diagnoses, Interventions and Outcomes instrument. Results: Post-revision documentation quality improved across all domains, with the largest raw gains in diagnostic formulation and nursing-sensitive outcomes. Interventions and outcomes remained the weaker downstream domains, with outcomes lowest at posttest. Diagnosis–intervention–outcome alignment and specificity increased, although explicit cross-linking remained inconsistent. Conclusion: This study adds preliminary evidence for nursing education: higher scores after guided AI-supported revision were associated with clearer and more coherent clinical reasoning in care planning. Policy recommendations are to permit bounded AI use with standardised prompts, revision logs/checklists, rubric-based feedback and verification safeguards to protect documentation integrity and patient safety.