Title : Development and preliminary evaluation of an explainable artificial intelligence driven digital twin nursing platform for perioperative management of patients undergoing oral and maxillofacial surgery: A mixed-methods study
Abstract:
To develop an explainable artificial intelligence (XAI)-driven Digital Twin Nursing Platform for patients undergoing oral and maxillofacial surgery and to evaluate its preliminary effectiveness in improving perioperative nursing outcomes, patient self-management, and nursing decision-making efficiency.
Design: A prospective mixed-methods pilot study.
Methods: An interdisciplinary team comprising oral and maxillofacial surgeons, specialist nurses, rehabilitation therapists, nutritionists, psychologists, and artificial intelligence engineers collaboratively developed the Digital Twin Nursing Platform using a human-centred co-design approach. The platform integrated electronic medical records, wearable physiological monitoring, patient-reported outcomes, and multimodal clinical data to generate individualized digital twins capable of dynamically predicting perioperative risks. Explainable machine-learning algorithms were embedded to provide transparent risk interpretation and personalized nursing recommendations. Between January and June 2026, 126 patients undergoing oral and maxillofacial surgery in a tertiary hospital were consecutively enrolled and allocated to an intervention group (n = 63) receiving Digital Twin-guided nursing care or a control group (n = 63) receiving routine perioperative nursing. Primary outcomes included postoperative complication incidence, pain intensity, rehabilitation adherence, and self-management ability. Secondary outcomes included nursing decision-making efficiency, patient satisfaction, and system usability.
Results: The Digital Twin prediction model demonstrated excellent discrimination, achieving an overall area under the receiver operating characteristic curve (AUC) of 0.92 (95% CI: 0.88–0.96), with a sensitivity of 89.7% and specificity of 85.4%. Compared with the control group, patients receiving Digital Twin-guided nursing experienced significantly fewer postoperative complications (6.3% vs. 19.0%, P = 0.031), lower postoperative pain scores on postoperative day 3 (2.8 ± 1.1 vs. 4.0 ± 1.4, P < 0.001), higher rehabilitation adherence (92.1% vs. 77.6%, P = 0.012), and greater self-management ability (85.7 ± 7.9 vs. 74.8 ± 8.6, P < 0.001). Patient satisfaction was also significantly higher in the intervention group (96.8% vs. 87.3%, P = 0.028). Among nurses, average clinical decision-making time decreased by 30.8%, while documentation efficiency improved by 27.5%. The system achieved excellent usability with a System Usability Scale score of 90.1 ± 4.6. Qualitative interviews identified three themes: Enhanced confidence in individualized nursing decisions, improved transparency of AI-assisted recommendations, and strengthened nurse patient communication.
Conclusion: The explainable AI-driven Digital Twin Nursing Platform demonstrated promising predictive performance and clinical applicability in perioperative oral and maxillofacial surgery nursing. By combining real-time individualized risk prediction with interpretable clinical recommendations, the platform improved patient outcomes, enhanced nursing efficiency, and facilitated personalized perioperative care. These findings support the feasibility of Digital Twin technology as an innovative strategy for advancing precision nursing and intelligent clinical decision support.
Keywords: Digital Twin; Explainable Artificial Intelligence; Precision Nursing; Oral and Maxillofacial Surgery; Perioperative Nursing; Digital Health.
