Accurately predicting the risk of unfavorable outcomes after endovascular coil therapy in patients with aneurysmal subarachnoid hemorrhage: an interpretable machine learning model.

BACKGROUND: Despite endovascular coiling as a valid modality in treatment of aneurysmal subarachnoid hemorrhage (aSAH), there is a risk of poor prognosis. However, the clinical utility of previously proposed early prediction tools remains limited. We aimed to develop a clinically generalizable machine learning (ML) models for accurately predicting unfavorable outcomes in aSAH patients after endovascular coiling. METHODS: Functional outcomes at 6 months after endovascular coiling were assessed via the modified Rankin Scale (mRS) and unfavorable outcomes were defined as mRS 3-6. Five ML algorithms (logistic regression, random forest, support vector machine, deep neural network, and extreme gradient boosting) were used for model development. The area under precision-recall curve (AUPRC) and receiver operating characteristic curve (AUROC) was used as main indices of model evaluation. SHapley Additive exPlanations (SHAP) method was applied to interpret the best-performing ML model. RESULTS: A total of 371 patients were eventually included into this study, and 85.4% of them had favorable outcomes. Among the five models, the DNN model had a better performance with AUPRC of 0.645 (AUROC of 0.905). Postoperative GCS score, size of aneurysm, and age were the top three powerful predictors. The further analysis of five random cases presented the good interpretability of the DNN model. CONCLUSION: Interpretable clinical prediction models based on different ML algorithms have been successfully constructed and validated, which would serve as reliable tools in optimizing the treatment decision-making of aSAH. Our DNN model had better performance to predict the unfavorable outcomes at 6 months in aSAH patients compared with Yan's nomogram model.

A visualized MAC nomogram online predicts the risk of three-month mortality in Chinese elderly aneurysmal subarachnoid hemorrhage patients undergoing endovascular coiling.

OBJECTIVE: Aneurysmal subarachnoid hemorrhage (aSAH) is an aggressive disease with higher mortality rate in the elderly population. Unfortunately, the previous models for predicting clinical prognosis are still not accurate enough. Therefore, we aimed to construct and validate a visualized nomogram model to predict online the 3-month mortality in elderly aSAH patients undergoing endovascular coiling. METHOD: We conducted a retrospective analysis of 209 elderly aSAH patients at People's Hospital of Hunan Province, China. A nomogram was developed based on multivariate logistic regression and forward stepwise regression analysis, then validated using the bootstrap validation method (n = 1000). In addition, the performance of the nomogram was evaluated by various indicators to prove its clinical value. RESULT: Morbid pupillary reflex, age, and using a breathing machine were independent predictors of 3-month mortality. The AUC of the nomogram was 0.901 (95% CI: 0.853-0.950), and the Hosmer-Lemeshow goodness-of-fit test showed good calibration of the nomogram (p = 0.4328). Besides, the bootstrap validation method internally validated the nomogram with an area under the curve of the receiver operator characteristic (AUROC) of 0.896 (95% CI: 0.846-0.945). Decision curve analysis (DCA) and clinical impact curve (CIC) indicated the nomogram's excellent clinical utility and applicability. CONCLUSION: An easily applied visualized nomogram model named MAC (morbid pupillary reflex-age-breathing machine) based on three accessible factors has been successfully developed. The MAC nomogram is an accurate and complementary tool to support individualized decision-making and emphasizes that patients with higher risk of mortality may require closer monitoring. Furthermore, a web-based online version of the risk calculator would greatly contribute to the spread of the model in this field.

Development and External Validation of a Dynamic Nomogram With Potential for Risk Assessment of Ruptured Multiple Intracranial Aneurysms.

BACKGROUND AND PURPOSE: About 20.1% of intracranial aneurysms (IAs) carriers are multiple intracranial aneurysms (MIAs) patients with higher rupture risk and worse prognosis. A prediction model may bring some potential benefits. This study attempted to develop and externally validate a dynamic nomogram to assess the rupture risk of each IA among patients with MIA. METHOD: We retrospectively analyzed the data of 262 patients with 611 IAs admitted to the Hunan Provincial People's Hospital between November 2015 and November 2021. Multivariable logistic regression (MLR) was applied to select the risk factors and derive a nomogram model for the assessment of IA rupture risk in MIA patients. To externally validate the nomogram, data of 35 patients with 78 IAs were collected from another independent center between December 2009 and May 2021. The performance of the nomogram was assessed in terms of discrimination, calibration, and clinical utility. RESULT: Size, location, irregular shape, diabetes history, and neck width were independently associated with IA rupture. The nomogram showed a good discriminative ability for ruptured and unruptured IAs in the derivation cohort (AUC = 0.81; 95% CI, 0.774-0.847) and was successfully generalized in the external validation cohort (AUC = 0.744; 95% CI, 0.627-0.862). The nomogram was calibrated well, and the decision curve analysis showed that it would generate more net benefit in identifying IA rupture than the "treat all" or "treat none" strategies at the threshold probabilities ranging from 10 to 60% both in the derivation and external validation set. The web-based dynamic nomogram calculator was accessible on https://wfs666.shinyapps.io/onlinecalculator/. CONCLUSION: External validation has shown that the model was the potential to assist clinical identification of dangerous aneurysms after longitudinal data evaluation. Size, neck width, and location are the primary risk factors for ruptured IAs.

Development and assessment of machine learning models for predicting recurrence risk after endovascular treatment in patients with intracranial aneurysms.

Intracranial aneurysms (IAs) remain a major public health concern and endovascular treatment (EVT) has become a major tool for managing IAs. However, the recurrence rate of IAs after EVT is relatively high, which may lead to the risk for aneurysm re-rupture and re-bleed. Thus, we aimed to develop and assess prediction models based on machine learning (ML) algorithms to predict recurrence risk among patients with IAs after EVT in 6 months. Patient population included patients with IAs after EVT between January 2016 and August 2019 in Hunan Provincial People's Hospital, and an adaptive synthetic (ADASYN) sampling approach was applied for the entire imbalanced dataset. We developed five ML models and assessed the models. In addition, we used SHapley Additive exPlanations (SHAP) and local interpretable model-agnostic explanation (LIME) algorithms to determine the importance of the selected features and interpret the ML models. A total of 425 IAs were enrolled into this study, and 66 (15.5%) of which recurred in 6 months. Among the five ML models, gradient boosting decision tree (GBDT) model performed best. The area under curve (AUC) of the GBDT model on the testing set was 0.842 (sensitivity: 81.2%; specificity: 70.4%). Our study firstly demonstrated that ML-based models can serve as a reliable tool for predicting recurrence risk in patients with IAs after EVT in 6 months, and the GBDT model showed the optimal prediction performance.

A novel visual dynamic nomogram to online predict the risk of unfavorable outcome in elderly aSAH patients after endovascular coiling: A retrospective study.

Background: Aneurysmal subarachnoid hemorrhage (aSAH) is a significant cause of morbidity and mortality throughout the world. Dynamic nomogram to predict the prognosis of elderly aSAH patients after endovascular coiling has not been reported. Thus, we aimed to develop a clinically useful dynamic nomogram to predict the risk of 6-month unfavorable outcome in elderly aSAH patients after endovascular coiling. Methods: We conducted a retrospective study including 209 elderly patients admitted to the People's Hospital of Hunan Province for aSAH from January 2016 to June 2021. The main outcome measure was 6-month unfavorable outcome (mRS ≥ 3). We used multivariable logistic regression analysis and forwarded stepwise regression to select variables to generate the nomogram. We assessed the discriminative performance using the area under the curve (AUC) of receiver-operating characteristic and the risk prediction model's calibration using the Hosmer-Lemeshow goodness-of-fit test. The decision curve analysis (DCA) and the clinical impact curve (CIC) were used to measure the clinical utility of the nomogram. Results: The cohort's median age was 70 (interquartile range: 68-74) years and 133 (36.4%) had unfavorable outcomes. Age, using a ventilator, white blood cell count, and complicated with cerebral infarction were predictors of 6-month unfavorable outcome. The AUC of the nomogram was 0.882 and the Hosmer-Lemeshow goodness-of-fit test showed good calibration of the nomogram (p = 0.3717). Besides, the excellent clinical utility and applicability of the nomogram had been indicated by DCA and CIC. The eventual value of unfavorable outcome risk could be calculated through the dynamic nomogram. Conclusion: This study is the first visual dynamic online nomogram that accurately predicts the risk of 6-month unfavorable outcome in elderly aSAH patients after endovascular coiling. Clinicians can effectively improve interventions by taking targeted interventions based on the scores of different items on the nomogram for each variable.