Predicting Long-Term Outcomes After Poor-Grade Aneurysmal Subarachnoid Hemorrhage Using Decision Tree Modeling

Area of Science:

• Neurosurgery
• Neurology
• Medical Informatics

Background:

• Predicting long-term outcomes for poor-grade aneurysmal subarachnoid hemorrhage (aSAH) remains a significant clinical challenge despite treatment advancements.
• Accurate prognostication is essential for guiding treatment decisions and managing patient expectations in aSAH.

Purpose of the Study:

• To develop and validate a decision tree model for predicting long-term outcomes in patients with poor-grade aSAH.
• To compare the predictive performance of decision tree modeling against traditional logistic regression.

Main Methods:

• Retrospective analysis of a prospective multicenter registry of patients with World Federation of Neurosurgical Societies (WFNS) grade IV or V aSAH.
• Outcome assessment using the modified Rankin Scale (mRS) at 12 months; unfavorable outcome defined as mRS 4-5 or death.
• Development and external validation of prognostic models using logistic regression and decision tree algorithms.

Main Results:

• Older age, absent pupillary reactivity, lower Glasgow Coma Scale (GCS), and higher modified Fisher grade independently predicted unfavorable outcomes.
• The decision tree model demonstrated strong predictive performance with 0.833 accuracy, 0.821 sensitivity, 0.846 specificity, and 0.88 AUC in internal testing.
• Both decision tree and logistic regression models showed comparable predictive accuracy, achieving 0.895 overall accuracy in external validation.

Conclusions:

• Decision tree modeling provides a straightforward and effective tool for predicting long-term outcomes in poor-grade aSAH patients.
• The developed decision tree model can aid clinicians in treatment decision-making for aSAH management.