Predicting pneumonia during hospitalization in flail chest patients using machine learning approaches.

Area of Science:

• Medical Informatics
• Pulmonology
• Trauma Surgery

Background:

• Pneumonia is a significant complication in flail chest patients, associated with high morbidity and mortality.
• Current diagnostic methods for pneumonia lack accuracy, potentially delaying crucial antimicrobial therapy.
• Machine learning offers a promising approach to enhance early pneumonia detection using electronic medical records.

Purpose of the Study:

• To develop and validate a novel machine learning model for predicting pneumonia risk in patients with flail chest.
• To identify key clinical variables associated with pneumonia development in this patient population.

Main Methods:

• Retrospective analysis of 169 adult flail chest patients' electronic medical records from January 2011 to December 2021.
• Development and evaluation of seven machine learning models, including XGBoost, using a 7:3 training-test split.
• Variable selection via Fisher score and model performance assessment using Area Under the Receiver Operating Characteristic Curve (AUC).

Main Results:

• The XGBoost model demonstrated superior performance with an AUC of 0.895 (sensitivity: 84.3%; specificity: 80.0%).
• Pneumonia in flail chest patients was significantly associated with systolic blood pressure, pH value, blood transfusion, and Injury Severity Score (ISS).
• The XGBoost model, utilizing 32 variables, proved highly reliable in risk assessment.

Conclusions:

• The developed XGBoost model offers a reliable tool for assessing pneumonia risk in flail chest patients.
• The Shapley Additive exPlanations (SHAP) method effectively identified critical risk factors, enhancing the clinical interpretability of the model.