Prediction of symptomatic intracranial hemorrhage before mechanical thrombectomy using machine learning in patients with anterior circulation large vessel occlusion

Introduction: Symptomatic intracranial hemorrhage (sICH) after mechanical thrombectomy (MT) is associated with worse outcomes. We sought to develop and internally validate a machine learning (ML) model to predict sICH prior to MT in patients with anterior circulation large vessel occlusion.

Methods: Consecutive adults who underwent MT for ICA, M1, or M2 occlusions at a single comprehensive stroke center from 2015 – 2020 were reviewed. sICH was defined as any ICH after MT with an increase in NIHSS of ≥ 4. The data was split into 80% training and 20% hold-out test sets. 9 ML models were screened using repeated 5-fold cross validation (CV) and the top 3 were selected for hyperparameter tuning with nested CV. The top performing ML model was evaluated on the test set. SHapley Additive exPlanations (SHAP) analysis was used to identify the most predictive features in the ML models.

Results: 497 patients (256 males, 241 females, mean age 64 years) met inclusion criteria. After model screening and hyperparameter tuning the top performing ML model was XGBoost. The AUC for the ML model on the test set was 0.79 (95% CI 0.67 – 0.89), which was significantly higher (p < 0.001) than the LR model (0.54 [95% CI 0.33 – 0.76]). The ML model also performed significantly better than the TAG (TICI, ASPECTS and blood glucose) score (0.69 [95% CI 0.55 – 0.85], p < 0.001) and STBA score (0.45 [95% CI 0.30 – 0.60], p < 0.001). Based on SHAP values, the most predictive features of sICH in the ML model were lower ASPECTS score, lower collateral score, and higher presenting NIHSS.

Conclusion : A ML model accurately predicted sICH prior to MT. It performed better than a standard statistical model and previously described clinical prediction models. Future studies with a larger sample size will be required for external validation of this model.