Key result
Machine learning models outperform standard logistic regression in predicting dabigatran-related vascular and bleeding risks.
Why the study?
The study aimed to develop suggestive utilization rules for dabigatran using an integrated machine learning decision-tree model.
Does an integrated machine learning decision-tree model improve the prediction of vascular events and bleeding compared to logistic regression in patients with non-valvular atrial fibrillation taking dabigatran 110 mg or 150 mg twice daily?
Observational (n=11,802)
Yes
Does an integrated machine learning decision-tree model improve the prediction of vascular events and bleeding compared to logistic regression in patients with non-valvular atrial fibrillation taking dabigatran 110 mg or 150 mg twice daily?
An integrated machine learning feature-selection and CART decision-tree model outperforms logistic regression in risk stratification and generates interpretable clinical rules to guide personalized dabigatran dosing (110 mg vs 150 mg) in patients with non-valvular atrial fibrillation.
Answers are grounded in this paper and cite their sources.
ML models may refine dabigatran risk prediction in AF; hypothesis-generating and requires prospective validation before clinical use.
The new generation of nonvitamin K antagonists are broadly applied for stroke prevention due to their notable efficacy and safety. Our study aimed to develop a suggestive utilization of dabigatran through an integrated machine learning (ML) decision-tree model. Participants taking different doses of dabigatran in the Randomized Evaluation of Long-Term Anticoagulant Therapy trial were included in our analysis and defined as the 110 mg and 150 mg groups. The proposed scheme integrated ML methods, namely naive Bayes, random forest (RF), classification and regression tree (CART), and extreme gradient boosting (XGBoost), which were used to identify the essential variables for predicting vascular events in the 110 mg group and bleeding in the 150 mg group. RF (0.764 for 110 mg; 0.747 for 150 mg) and XGBoost (0.708 for 110 mg; 0.761 for 150 mg) had better area under the receiver operating characteristic curve (AUC) values than logistic regression (benchmark model; 0.683 for 110 mg; 0.739 for 150 mg). We then selected the top ten important variables as internal nodes of the CART decision tree. The two best CART models with ten important variables output tree-shaped rules for predicting vascular events in the 110 mg group and bleeding in the 150 mg group. Our model can be used to provide more visualized and interpretable suggestive rules to clinicians managing NVAF patients who are taking dabigatran.
No takes yet. Share an insight, caveat, or question.
Huang et al. (2023) conducted an observational in Non-valvular atrial fibrillation (NVAF) (n=11,802). Integrated machine learning decision-tree model (naive Bayes, random forest, CART, and XGBoost) for dabigatran 110 mg and 150 mg risk evaluation vs. Logistic regression (benchmark model) was evaluated on Prediction (AUC) of vascular events (stroke, systemic embolism, and vascular death) in the 110 mg group and bleeding in the 150 mg group within the first year of follow-up (RF AUC 0.764 (110 mg) and 0.747 (150 mg); XGBoost AUC 0.708 (110 mg) and 0.761 (150 mg); reconstructed 10-variable CART AUC 0.624 (110 mg) and 0.717 (150 mg)). Random forest and XGBoost outperformed logistic regression in predicting dabigatran risks.
Synapse has enriched 10 related papers. Consider them for comparative context: