Machine Learning Approaches-Driven for Mortality Prediction for Patients Undergoing Craniotomy in ICU

Area of Science:

• Neurosurgery
• Medical Informatics
• Machine Learning in Healthcare

Background:

• Craniotomy patients in the intensive care unit (ICU) face significant mortality risks.
• Predictive modeling can aid in identifying high-risk individuals for targeted interventions.

Purpose of the Study:

• To develop and validate a machine learning model for predicting mortality in post-craniotomy ICU patients.
• To identify key factors contributing to mortality in this patient population.

Main Methods:

• Retrospective analysis of a surgical intensive care unit (ICU) database.
• Application of five machine learning algorithms for mortality prediction.
• Evaluation of model performance using accuracy, precision, recall, F1-score, and Area Under the Curve (AUC).

Main Results:

• eXtreme Gradient Boosting (XGBoost) demonstrated superior performance with an AUC of 0.84.
• Local Interpretable Model-agnostic Explanations (LIME) analysis identified significant high-risk factors for mortality.
• The study successfully established a mortality predictive model for ICU patients post-craniotomy.

Conclusions:

• The developed XGBoost model provides a reliable tool for predicting mortality in ICU patients after craniotomy.
• Identification of high-risk factors can support clinical decision-making and improve patient outcomes.
• This predictive model has the potential to enhance patient care and resource allocation in neurosurgical ICUs.