An accurate prognostic prediction for aneurysmal subarachnoid hemorrhage dedicated to patients after endovascular treatment

 

You do realize you're predicting  high fatality and permanent disability rather than doing the correct research to prevent that from happening. I'd fire the lot of you for not solving stroke.

An accurate prognostic prediction for aneurysmal subarachnoid hemorrhage dedicated to patients after endovascular treatment

Show all authors

Han Lu*, Gaici Xue*, Sisi Li*, ...

First Published June 1, 2022 Research Article 

https://doi.org/10.1177/17562864221099473

Article information

Abstract

Background:

Endovascular treatment for aneurysmal subarachnoid hemorrhage (aSAH) has high fatality and permanent disability rates. It remains unclear how the prognosis is determined by the complex interaction between clinical severity and aneurysm characteristics.

Objective:

This study aimed to design an accurate prognostic prediction model for aSAH patients after endovascular treatment and elucidate the interaction between clinical severity and aneurysm characteristics.

Methods:

We used a clinically homogeneous data set with 1029 aSAH patients who received endovascular treatment to develop prognostic models. Aneurysm characteristics were measured by variables, such as aneurysm size, neck size, and dome-to-neck ratio, while clinical severity on admission was measured by both comorbidities and neurological condition. In total, 18 clinical variables were used for prognostic prediction. Considering the imbalance between the favorable and the poor outcomes in this clinical population, both ensemble learning and deep reinforcement learning approaches were used for prediction.

Results:

The random forest (RF) model was selected as the best approach for the prognostic prediction for all patients and also for patients with good-grade aSAH. Using an independent test data set, the model made accurate predictions (AUC = 0.869 ± 0.036, sensitivity = 0.709 ± 0.087, specificity = 0.805 ± 0.034) with the clinical severity on admission as a leading contributor to the prediction. For patients with good-grade aSAH, the RF model performed the best (AUC = 0.805 ± 0.034, sensitivity = 0.620 ± 0.172, specificity = 0.696 ± 0.043) with aneurysm characteristics as leading contributors. The classic scoring systems failed in this patient group (AUC < 0.600; sensitivity = 0.000, specificity = 1.000).

Conclusion:

The proposed prognostic prediction model outperformed the classic scoring systems for patients with aSAH after endovascular treatment, especially when the classic scoring systems failed to make any informative prediction for patients with good-grade aSAH, who constitute the majority group (79%) of this clinical population.

Keywords

endovascular treatment, imbalanced data, machine learning, prognostic prediction, reinforcement learning

Introduction

Subarachnoid hemorrhage from ruptured intracranial aneurysms, a worldwide health burden, is characterized by its high fatality and permanent disability rates. Approximately one-third of all patients die owing to the severe brain injury with the initial weeks after aneurysmal subarachnoid hemorrhage (aSAH), and a large portion of survivors suffered from long-term disability or cognitive impairment.¹ Prognostic prediction model for patients after aSAH is critical not only to inform outcome expectations but also to identify modifiable contributors to a favorable prognosis. However, it remains unclear how the complex interaction between clinical severity and aneurysm characteristics jointly determines the prognosis after aSAH.

To date, a few clinical scoring systems can be used to inform the prognosis after aSAH, including the subarachnoid hemorrhage international trialists (SAHIT),² functional recovery expected after subarachnoid hemorrhage (FRESH),³ size of the aneurysm, age, Fisher grade, World Federation of Neurosurgical Societies after resuscitation (SAFIRE),⁴ and so on. However, these scoring systems were often built from clinically heterogeneous patient groups to maximize the overall sample size. For example, the patients in these studies were often treated with various methods, including surgical clipping, endovascular method, and even conservative treatment.^2–4 Among these different treatment approaches, the difference in prognosis had already been reported. A meta-analysis of the data from 11,568 patients showed that the coiling reduced the 1-year poor outcome rate (OR, 1.48) compared with clipping.⁵ Given the continuous surgical and material advances in the treatment of aSAH during the last two decades,⁶ the training data collected in the early 2000s for these scoring systems might make them less predictive in the latest clinical practice. Recently, the researches on the prognostic prediction models had begun to focus on the homogeneous groups of patients, especially the patients after aSAH treated with the endovascular approach only.^7–9 However, the sample sizes were often limited. To build an accurate and reliable prognostic prediction model, both large sample size and independent test data set are needed.

Another important limitation in literature is the lack of a prognostic prediction model for patients with good-grade aSAH on admission. As reviewed by a recent meta-analysis, five aSAH studies with a total of 2862 participants found that 2425 (84.7%) patients had the good-grade aSAH on admission, but among them 19.8% suffered poor outcomes.⁵ Therefore, an accurate prognostic model for this patient group could better inform the decision-making before the surgery. For example, when a poor outcome is predicted, alternative methods such as clipping may be considered. Furthermore, the identification of the key factors that contribute to this prognosis may provide novel opportunities toward better outcomes.

To address these limitations, we attempted to establish multivariate models for the prognostic prediction in patients after aSAH treated with the endovascular approach, both in the general patient population and in patients with good-grade aSAH on admission. We reviewed the data from the largest-to-date cohort of 1191 patients after aSAH who were treated with the endovascular approach at a single center between 2012 and 2018. Using the clinical information on admission, we proposed a few multivariate models and compared them with classic scoring systems to improve the prediction accuracy for 1-year prognoses of these patients and validated performances of these models using an independent test data set.

More at link.