Comparative effectiveness of conventional nursing versus data-driven predictive nursing models in orthopedic trauma patients: a systematic review and meta-analysis

Orthopedic injuries have necessitated the adoption of appropriate nursing interventions to promote favorable healing and minimize risks and complications associated with these conditions. In recent times, predictive nursing models that use artificial intelligence have generated considerable interest, particularly for comparison with traditional nursing interventions. To determine the comparative effectiveness of traditional nursing interventions versus data-driven predictive nursing models in orthopedic injury cases. A systematic review of 33 studies, including randomized controlled trials, cohort studies, quasi-experiments, and machine-learning-based prediction studies, was conducted. Pooled odds ratios (OR) with 95% confidence intervals (CI) were computed by employing a random-effects inverse-variance meta-analysis approach. Heterogeneity between studies was measured with I-squared statistics, while publication bias was detected with Egger’s test and funnel plot analyses. There was a statistically significant improvement in all intervention categories compared to the control group. Traditional and perioperative nursing interventions had an overall OR of 2.13 (95% CI 1.74–2.61), whereas postoperative rehabilitation interventions had an OR of 1.97 (95% CI 1.37–2.84). The machine learning-based predictive models had an OR of 2.26 (1.89–2.71), whereas the predictive nursing decision-support systems had an OR of 2.14 (1.70–2.68). Prognostic studies also had OR 2.27 (1.82–2.83). The heterogeneity was at most moderate, and there was no publication bias except in prognostic studies. Predictive and conventional nursing strategies both significantly improve orthopedic trauma outcomes, with predictive models offering greater potential for individualized care.