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

Abstract

Intracranial aneurysms (IAs) remains 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 the data was randomly divided into a training set and a testing set. We developed ve 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 ve 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%; speci city: 70.4%). Our study rstly 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.

Global trend in research of intracranial aneurysm management with artificial intelligence technology: a bibliometric analysis

“…Notable contributions include Tanioka et al ’s creation of an AI model for predicting delayed cerebral ischemia ( 40 ), Kim et al ’s development of an AI-explainable predictive model for vasospasm prediction ( 41 ), and Bizjak et al ’s model to predict the growth of untreated IAs based on baseline aneurysm morphology ( 42 ). Additionally, some potential sub-fields, such as recurrence prediction for IAs after treatment ( 43 ) and effectiveness prediction after endovascular treatment ( 44 ), remain underexplored and warrant further investigation in the future.…”

A predictive model for the recurrence of intracranial aneurysms following coil embolization

“…D max has been demonstrated to have predictive value for ruptured aneurysms, and the rupture status can independently predict aneurysm recurrence. Based on clinical experience, neurosurgeons tend to regard IAs with a larger N as being associated with a high risk of recurrence ( 15 ). However, there have been limited studies regarding the significance of N in multivariate analysis.…”

“…Current models for predicting aneurysm recurrence are limited. Lin et al explored the short-term recurrence model in 6 months based on significant variables in the univariate analysis ( 15 ). We developed a novel predictive model based on independent risk factors, with an average follow-up of 12.53 months.…”

Machine learning model to predict 3-month nonunion(delayed union) of unstable distal clavicle fracture patients treated with open reduction and internal fixation

“…Recent research has demonstrated the growing application of ML models in the medical sector [11][12][13][14][15][16].ML models incorporate a variety of advanced algorithms to handle different data characteristics, including State-Action-Reward-State-Action , Random Forest Classi er, Deep Q Network, and Support Vector Machine [17].In comparison to traditional regression models, ML has been validated as an effective method for prognosis prediction in modeling due to its capacity to intricately analyze complex non-linear interrelations among variables [18,19], and to enhance prediction accuracy through its superior algorithms, particularly when analyzing large datasets with numerous variables [20,21]. Crucially, the widespread adoption of Electronic Patient Record systems and the extensive use of structured patient data have made sophisticated algorithmic modeling and bedside application feasible [22].…”