Exploring the Utility of ChatGPT in Nursing Care Plan Development: A Qualitative Evaluation of Standardized Nursing Classifications Alignment

Abstract

Introduction: Generative artificial intelligence tools, such as ChatGPT, are being increasingly explored for their potential use in nursing education, particularly in clinical documentation and care planning. This study aimed to evaluate the alignment of ChatGPT-generated nursing care plans with standardized nursing classifications and to assess the educational utility of AI-assisted care planning in nursing student learning. Methods: A mixed-methods design was employed, combining quantitative analysis of care plan quality using a 3-point Likert scale with a qualitative expert review of nursing diagnoses and interventions based on alignment with NANDA-I nursing diagnoses and nursing interventions classification. Ten students used GPT-3.5 to generate ten care plans based on patient data. The development of prompts was through iterative testing. Results: Among the 30 nursing diagnoses evaluated, 43.3% were entirely appropriate, 40% partially appropriate, and 16.7% inappropriate. Nursing interventions demonstrated slightly better performance, with 50% being entirely appropriate, 36.7% partially appropriate, and 13.3% inappropriate. ChatGPT could generate nursing diagnoses and interventions but often misclassified priorities or used vague language. Discussion and Conclusion: ChatGPT can support nursing education when used as a supplementary tool; however, expert supervision is necessary to ensure safe and contextually valid care planning.