Integrating NANDA-I Diagnoses with Functional Health Patterns: A Semantic Similarity Approach

This study proposes a data-driven method to map NANDA-I nursing diagnoses to Gordon's 11 Functional Health Patterns (FHP) using text similarity techniques. The goal is to support nurses in objectively identifying appropriate care needs without relying solely on individual clinical experience. The method leverages Sentence-BERT (SBERT) for contextual vectorization, WordNet for semantic expansion, and an individual threshold method for matching. Evaluation using 56 NANDA-I diagnosis instances revealed that the SBERT and WordNet configuration achieved the highest F1 score. SBERT significantly outperformed the classical TF-IDF method due to its ability to capture nuanced semantic relationships. While WordNet improved TF-IDF performance, its impact on SBERT was limited. The individual threshold method performed better than the weighted integration method, emphasizing the importance of threshold selection based on application goals. Despite challenges in mapping due to conceptual differences and linguistic ambiguity between NANDA-I and FHP, the study demonstrates that datadriven approaches can improve the accuracy and consistency of nursing assessments. Future work should explore larger datasets and diverse modeling techniques.