This obviously requires followup research to figure why there was futile recanalization and come up with interventions to fix that.
Predicting futile recanalization, malignant cerebral edema, and cerebral herniation using intelligible ensemble machine learning following mechanical thrombectomy for acute ischemic stroke
- 1Department of Radiology, Nanfang Hospital, Southern Medical University, Guangzhou, China
- 2Department of Neurology, The Second Hospital of Jilin University, Changchun, China
- 3Department of Neurology, Nanfang Hospital, Southern Medical University, Guangzhou, China
- 4Hospital Office, Ganzhou People's Hospital, Ganzhou, China
- 5Hospital Office, Ganzhou Hospital-Nanfang Hospital, Southern Medical University, Ganzhou, China
Purpose: To establish an ensemble machine learning (ML) model for predicting the risk of futile recanalization, malignant cerebral edema (MCE), and cerebral herniation (CH) in patients with acute ischemic stroke (AIS) who underwent mechanical thrombectomy (MT) and recanalization.
Methods: This prospective study included 110 patients with premorbid mRS ≤ 2 who met the inclusion criteria. Futile recanalization was defined as a 90-day modified Rankin Scale score >2. Clinical and imaging data were used to construct five ML models that were fused into a logistic regression algorithm using the stacking method (LR-Stacking). We added the Shapley Additive Explanation method to display crucial factors and explain the decision process of models for each patient. Prediction performances were compared using area under the receiver operating characteristic curve (AUC), F1-score, and decision curve analysis (DCA).
Results: A total of 61 patients (55.5%) experienced futile recanalization, and 34 (30.9%) and 22 (20.0%) patients developed MCE and CH, respectively. In test set, the AUCs for the LR-Stacking model were 0.949, 0.885, and 0.904 for the three outcomes mentioned above. The F1-scores were 0.882, 0.895, and 0.909, respectively. The DCA showed that the LR-Stacking model provided more net benefits for predicting MCE and CH. The most important factors were the hypodensity volume and proportion in the corresponding vascular supply area.
Conclusion: Using the ensemble ML model to analyze the clinical and imaging data of AIS patients with successful recanalization at admission and within 24 h after MT allowed for accurately predicting the risks of futile recanalization, MCE, and CH.
Introduction
Stroke is a leading cause of mortality and disability worldwide. The global deaths caused by ischemic stroke increased by 60.68% over 30 years, from 2,049,670 in 1990 to 3,293,400 in 2019 (1). Acute ischemic stroke (AIS) is characterized by a sudden reduction or cessation of blood flow in a brain artery that results in ischemia and hypoxia of the brain tissue in the corresponding blood supply area. According to current international guidelines and related research, endovascular mechanical thrombectomy (MT) combined with recombinant tissue-type plasminogen activator (rt-PA) thrombolysis is the standard treatment in patients with AIS due to occlusion of the proximal anterior intracranial region, while MT is one of the most important forms of endovascular treatment (EVT) for large vessel occlusion (2–4).
However, despite recent improvements in MT procedure, futile recanalization, defined as a 90-day modified Rankin Scale (mRS-90) score >2 after adequate vessel recanalization, remains a serious clinical problem (5). The incidence of futile recanalization after MT is approximately 49–67% (5). The primary risk factors for patients with AIS include large infarct volume, poor collateral circulation, and high National Institutes of Health Stroke Scale (NIHSS) score (6–8). While the mRS and NIHSS scores are among the methods used to evaluate AIS functional outcomes, few studies have focused on the functional outcomes and potentially lethal complications in patients with AIS who have undergone an MT and for whom recanalization was achieved. Although computed tomography-angiography and magnetic resonance imaging (MRI) can be used to accurately evaluate the entire ischemic lesion (core and penumbra), non-contrast computed Tomography (NCCT) is common for patients with AIS after MT, due to its widespread availability, low cost, and rapid scanning speed (9).
Malignant cerebral edema (MCE) and cerebral herniation (CH) are relatively common and serious complications that lead to rapid deterioration of patient's condition, coma, poor prognosis, or even death. Therefore, being able to rapidly recognize which patients are at high risk for futile recanalization and potentially lethal complications after an MT can help clinicians make individualized treatment decisions.
The machine learning (ML) method can accurately process complex nonlinear relationships among a large number of variables, which is difficult to accomplish with traditional statistical models (10, 11). This technology has been applied to predict the outcomes of patients with AIS; however, a drawback of complex ML algorithms is its interpretability has limitations, which are commonly referred to as black-box models for clinicians. Previous researchers have attempted to solve this problem using simple ML algorithms, but more complex and improved models, such as the support vector machine (SVM), deep neural network, and ensemble ML algorithms, which may perform better in stroke-related tasks have not been fully utilized (12, 13). In addition, few studies have focused on the ability of applied complex ML methods to predict the occurrence of malignant complications in patients who undergo MT and recanalization.
Therefore, in this study, ensemble ML models were constructed to predict futile recanalization, MCE, and CH in patients with AIS treated with MT and in whom successful recanalization was achieved. The model we constructed can accurately identify and display the high-risk factors of each patient.
More at link.
No comments:
Post a Comment