A predictive nomogram of bleeding risk in patients with atrial fibrillation after drug-eluting stent implantation

Machine learning derived model for the prediction of bleeding in dual antiplatelet therapy patients

Objective

This study aimed to develop a predictive model for assessing bleeding risk in dual antiplatelet therapy (DAPT) patients.

Methods

A total of 18,408 DAPT patients were included. Data on patients' demographics, clinical features, underlying diseases, past history, and laboratory examinations were collected from Affiliated Dongyang Hospital of Wenzhou Medical University. The patients were randomly divided into two groups in a proportion of 7:3, with the most used for model development and the remaining for internal validation. LASSO regression, multivariate logistic regression, and six machine learning models, including random forest (RF), k-nearest neighbor imputing (KNN), decision tree (DT), extreme gradient boosting (XGBoost), light gradient boosting machine (LGBM), and Support Vector Machine (SVM), were used to develop prediction models. Model prediction performance was evaluated using area under the curve (AUC), calibration curves, decision curve analysis (DCA), clinical impact curve (CIC), and net reduction curve (NRC).

Results

The XGBoost model demonstrated the highest AUC. The model features were comprised of seven clinical variables, including: HGB, PLT, previous bleeding, cerebral infarction, sex, Surgical history, and hypertension. A nomogram was developed based on seven variables. The AUC of the model was 0.861 (95% CI 0.847-0.875) in the development cohort and 0.877 (95% CI 0.856-0.898) in the validation cohort, indicating that the model had good differential performance. The results of calibration curve analysis showed that the calibration curve of this nomogram model was close to the ideal curve. The clinical decision curve also showed good clinical net benefit of the nomogram model.

Conclusions

This study successfully developed a predictive model for estimating bleeding risk in DAPT patients. It has the potential to optimize treatment planning, improve patient outcomes, and enhance resource utilization.

LASSO-derived model for the prediction of bleeding in aspirin users

Aspirin is widely used for both primary and secondary prevention of panvascular diseases, such as stroke and coronary heart disease (CHD). The optimal balance between reducing panvascular disease events and the potential increase in bleeding risk remains unclear. This study aimed to develop a predictive model specifically designed to assess bleeding risk in individuals using aspirin. A total of 58,415 individuals treated with aspirin were included in this study. Detailed data regarding patient demographics, clinical characteristics, comorbidities, medical history, and laboratory test results were collected from the Affiliated Dongyang Hospital of Wenzhou Medical University. The patients were randomly divided into two groups at a ratio of 7:3. The larger group was used for model development, while the smaller group was used for internal validation. To develop the prediction model, we employed least absolute shrinkage and selection operator (LASSO) regression followed by multivariate logistic regression. The performance of the model was assessed through metrics such as the area under the receiver operating characteristic (ROC) curve (AUC), calibration curves, and decision curve analysis (DCA). The LASSO-derived model employed in this study incorporated six variables, namely, sex, operation, previous bleeding, hemoglobin, platelet count, and cerebral infarction. It demonstrated excellent performance at predicting bleeding risk among aspirin users, with a high AUC of 0.866 (95% CI 0.857-0.874) in the training dataset and 0.861 (95% CI 0.848-0.875) in the test dataset. At a cutoff value of 0.047, the model achieved moderate sensitivity (83.0%) and specificity (73.9%). The calibration curve analysis revealed that the nomogram closely approximated the ideal curve, indicating good calibration. The DCA curve demonstrated a favorable clinical net benefit associated with the nomogram model. Our developed LASSO-derived predictive model has potential as an alternative tool for predicting bleeding in clinical settings.

Assessing peptic ulcer risk with the HAMPROW score in the general Chinese population

The timely identification of individuals at high risk for peptic ulcers (PUs) is vital in preventing gastrointestinal bleeding after antiplatelet therapy. This study was designed to determine PU risk factors and develop a risk assessment model for PU detection in the general Chinese population. In a prospective dataset, clinical data from individuals undergoing gastroscopic evaluation between April 2019 and May 2022 were recorded. PUs were defined as mucosal defects exceeding 5 mm confirmed via gastroscopy. Participants were categorized into development (April 2019 to April 2021) and validation (May 2021 to May 2022) sets based on chronological order. LASSO-derived logistic regression analysis was employed to create a score, which was further validated via temporal validation. A total of 902 patients were ultimately enrolled, 204 (22.6%) of whom had PUs based on endoscopic findings. In the development cohort (n = 631), seven independent risk factors emerged: male sex (OR = 2.35, P = 0.002), white blood cell (WBC) count (OR = 1.16, P = 0.010), red blood cell (RBC) count (OR = 0.49, P < 0.001), globulin level (OR = 0.92, P = 0.004), albumin level (OR = 0.94, P = 0.020), pepsinogen I (PGI) level (OR = 1.01, P < 0.001), and positive Helicobacter pylori (HP) antibody (OR = 2.50, P < 0.001). Using these factors, a nomogram (HAMPROW score [hazard ratio (HP) antibody, albumin, male, PGI, RBC, globulin, and WBC]) was developed for individual PU prediction. The ability of the HAMPROW score to predict survival was confirmed with AUCs of 0.854 (95% CI 0.816-0.891) and 0.833 (95% CI 0.771-0.895) in the development and validation sets, respectively. In conclusion, the HAMPROW score can be used to screen for PUs effectively in the general Chinese population, facilitating personalized early detection of high risk of gastrointestinal bleeding before antiplatelet therapy.

A Practical Nomogram for Predicting the Bleeding Risk in Patients with a History of Myocardial Infarction Treating with Aspirin

BACKGROUND

Aspirin is a widely used antiplatelet medication to prevent blood clots, reducing the risk of cardiovascular event. Healthcare providers need to be mindful of the risk of aspirin-induced bleeding and carefully balancing its benefits against potential risks. The objective of this study was to create a practical nomogram for predicting bleeding risk in patients with a history of myocardial infarction treating with aspirin.

METHODS

A total of 2099 myocardial infarction patients with aspirin were enrolled. The patients were randomly divided into two groups, with a 7:3 ratio, for model development and internal validation. Boruta analysis was utilized to identify clinically significant features associated with bleeding. Logistic regression model based on independent bleeding risk factors was constructed and presented as a nomogram. Model performance was assessed from three aspects: identification, calibration, and clinical utility.

RESULTS

Boruta analysis identified eight clinical features from 25, and further multivariate logistic regression analysis selected four independent risk factors: hemoglobin, platelet count, previous bleeding, and sex. A visual nomogram was created based on these variables. The model achieved an area under the curve of 0.888 (95% CI: 0.845-0.931) in the training dataset and 0.888 (95% CI: 0.808-0.968) in the test dataset. Calibration curve analysis showed close approximation to the ideal curve. Decision curve analysis demonstrated favorable clinical net benefit for the model.

CONCLUSIONS

Our study focused on creating and validating a model to evaluate bleeding risk in patients with a history of myocardial infarction treated with aspirin, which demonstrated outstanding performance in discrimination, calibration, and net clinical benefit.