Decision Trees for Managing Impaired Physical Mobility in Multiple Trauma Patients

To develop and validate decision trees using conditional probabilities to identify the predictors of mortality and morbidity deterioration in trauma patients.

A quasi-experimental longitudinal study conducted at a Level 1 Trauma Center in São Paulo, Brazil.

The study analysed 201 patient records using standardised nursing documentation (NANDA International and Nursing Outcomes Classification). Decision trees were constructed using the chi-squared automatic interaction detection (CHAID) algorithm and validated through K-fold cross-validation to ensure model reliability.

Decision trees identified key predictors of survival and mobility deterioration. Patients who did not require (NOC 0414) Cardiopulmonary Status but required (NOC 0210) Transfer Performance had a 97.4% survival rate. Conversely, those requiring (NOC 0414) Cardiopulmonary Status had a 25% risk of worsening mobility, compared to 9% for those who did not. K-fold cross-validation confirmed the model's predictive accuracy, reinforcing the robustness of the decision tree approach (Value).

Decision trees demonstrated strong predictive capabilities for mobility outcomes and mortality risk, offering a structured, data-driven framework for clinical decision-making. These findings underscore the importance of early mobilisation, tailored rehabilitation interventions and assistive devices in improving patient recovery. This study is among the first to apply decision trees in this context, highlighting its novelty and potential to enhance trauma critical care practices.

This study highlights the potential of decision trees, a supervised machine learning method, in nursing practice by providing clear, evidence-based guidance for clinical decision-making. By enabling early identification of high-risk patients, decision trees facilitate timely interventions, reduce complications and support personalised rehabilitation strategies that enhance patient safety and recovery.

This research addresses the challenge of improving outcomes for critically ill and trauma patients with impaired mobility by identifying effective strategies for early mobilisation and rehabilitation. The integration of artificial intelligence-driven decision trees strengthens evidence-based nursing practice, enhances patient education and informs scalable interventions that reduce trauma-related complications. These findings have implications for healthcare providers, rehabilitation specialists and policymakers seeking to optimise trauma care and improve long-term patient outcomes.

Patients provided authorisation for the collection of their clinical data from medical records during hospitalisation.