Application of Explainable AI Models for Real-Time Pediatric Nursing Assessment: Improving Clinical Accuracy and Family-Centered Care

Background and Aim: Accurate pediatric nursing assessment is critical for timely diagnosis and effective management of childhood illnesses. Recent advances in Artificial Intelligence (AI), particularly Explainable AI (XAI), offer novel opportunities to enhance clinical accuracy while maintaining transparency and trust in healthcare decisions. Despite growing interest, the application of XAI in real-time pediatric nursing assessment remains underexplored. This systematic review aims to evaluate the current evidence on the use of explainable AI models in pediatric nursing assessments, focusing on their impact on clinical accuracy and family-centered care. Methods: A comprehensive literature search was performed in PubMed, Scopus, Web of Science, IEEE Xplore, Cochrane Library, and Embase databases for articles published up to August ۲۰۲۵. Search terms included combinations of "explainable AI," "XAI," "artificial intelligence," "pediatric nursing assessment," "clinical accuracy," and "family-centered care." Inclusion criteria comprised original research studies applying XAI models in pediatric nursing assessment (ages ۰-۱۸), reporting clinical performance metrics and/or effects on family engagement. Exclusion criteria were studies on adults, non-clinical AI applications, reviews, editorials, and studies without explainability components. Two independent reviewers screened ۱,۴۸۹ records, with ۲۴ studies meeting eligibility for qualitative synthesis. Results: The ۲۴ included studies utilized a variety of machine learning models integrated with XAI techniques such as SHAP, LIME, and attention mechanisms to provide interpretable outputs during nursing assessments. Sample sizes ranged from ۵۰ to ۱,۲۰۰ pediatric patients across diverse clinical settings including intensive care, outpatient, and emergency departments. Reported clinical accuracy metrics showed an area under the curve (AUC) between ۰.۷۵ and ۰.۹۴, indicating strong predictive performance. Multiple studies demonstrated that XAI models improved nurse decision-making confidence and facilitated family engagement by explaining assessment outcomes in understandable terms. However, only ۷ studies performed external validation, and fewer addressed long-term clinical impact or ethical considerations. Conclusion: Explainable AI models represent a promising advancement in pediatric nursing assessment by enhancing clinical accuracy and supporting family-centered care through transparent decision support. Despite encouraging results, limited external validation and heterogeneity in methodologies warrant cautious interpretation. Future research should prioritize large-scale, multicenter prospective studies, standardized explainability metrics, and integration of ethical frameworks to fully realize the benefits of XAI in pediatric nursing practice.

کلیدواژه ها:

Explainable AI ، Pediatric Nursing Assessment ، Clinical Accuracy ، Family-Centered Care ، Machine Learning