Determining the Need for Computed Tomography Scan Following Blunt Chest Trauma through Machine Learning Approaches

Area of Science:

• Emergency Medicine
• Machine Learning
• Medical Imaging

Background:

• Computed tomography (CT) scans are crucial for diagnosing trauma patients.
• Machine learning advances enable more accurate and rapid medical diagnoses.
• Reducing unnecessary CT scans in emergency medicine is a key objective.

Purpose of the Study:

• To develop a machine learning-based method to assist emergency physicians.
• To prevent the performance of unnecessary CT scans for chest trauma patients.

Main Methods:

• A dataset of 1000 samples was analyzed.
• Seven classification methods were evaluated: SVM, logistic regression, Naïve Bayes, decision tree, multilayer perceptron, random forest, and KNN.
• A decision tree approach was selected as the final model due to its interpretability.

Main Results:

• The decision tree algorithm demonstrated superior accuracy compared to other methods.
• The optimal tree depth was determined to be 7.
• The final model achieved 99.91% accuracy, 100% sensitivity, and 99.33% specificity.

Conclusions:

• The proposed machine learning model is highly reliable for determining the necessity of CT scans.
• High sensitivity of the model supports its clinical utility in reducing unnecessary procedures.