Abstract:Bone non-union is among the most severe complications associated with trauma surgery, occurring in 10-30% of cases after long bone fractures. Treating non-unions requires a high level of surgical expertise and often involves multiple revision surgeries, sometimes even leading to amputation. Thus, more accurate prognosis is crucial for patient well-being. Recent advances in machine learning (ML) hold promise for developing models to predict non-union healing, even when working with smaller datasets, a commonly encountered challenge in clinical domains. To demonstrate the effectiveness of ML in identifying candidates at risk of failed non-union healing, we applied three ML models (logistic regression, support vector machine, and XGBoost) to the clinical dataset TRUFFLE, which includes 797 patients with long bone non-union. The models provided prediction results with 70% sensitivity, and the specificities of 66% (XGBoost), 49% (support vector machine), and 43% (logistic regression). These findings offer valuable clinical insights because they enable early identification of patients at risk of failed non-union healing after the initial surgical revision treatment protocol.

What problem does this paper attempt to address?

The problem this paper attempts to address is: how to predict whether patients will experience nonunion (i.e., failure to heal) after non-union fracture surgery using machine learning models. Nonunion is a serious complication of trauma surgery, with an incidence rate of approximately 10% to 30%. It requires a high level of surgical skill to treat and often necessitates multiple revision surgeries, sometimes even leading to amputation. Therefore, accurately predicting which patients may develop nonunion is of great significance for improving patients' quality of life and reducing the medical burden. In the paper, the authors used three machine learning models (logistic regression, support vector machine, and XGBoost) to analyze the TRUFFLE clinical dataset, which includes 797 patients, to evaluate the performance of these models in predicting nonunion after non-union fracture surgery. The study results indicate that the XGBoost model performed the best in this task, with a sensitivity of 70% and a specificity of 66%. These findings provide clinicians with valuable early identification tools, helping to take timely personalized interventions and improve patients' chances of recovery.

Predictive Model Development to Identify Failed Healing in Patients after Non-Union Fracture Surgery