Interpretable machine learning for early neurological deterioration prediction in atrial fibrillation-related stroke

Seong-Hwan Kim,Yang Ha Hwang,Bum Joon Kim,Jay Chol Choi,Chi Kyung Kim,Kyungmi Oh,Yong-Jae Kim,Sungwook Yu,Eun‐Tae Jeon ,Man‐Seok Park ,Jong‐Won Chung ,Jeong‐Min Kim ,Tae‐Jin Song ,Woo‐Keun Seo ,Joon‐Tae Kim ,Kwang‐Yeol Park ,Kang‐Ho Choi ,Gyeong‐Moon Kim ,Sung Hyuk Heo,Jong-Ho Park,Oh Young Bang,Jin‐Man Jung

doi:10.1038/s41598-021-99920-7

Abstract

We aimed to develop a novel prediction model for early neurological deterioration (END) based on an interpretable machine learning (ML) algorithm for atrial fibrillation (AF)-related stroke and to evaluate the prediction accuracy and feature importance of ML models. Data from multicenter prospective stroke registries in South Korea were collected. After stepwise data preprocessing, we utilized logistic regression, support vector machine, extreme gradient boosting, light gradient boosting machine (LightGBM), and multilayer perceptron models. We used the Shapley additive explanation (SHAP) method to evaluate feature importance. Of the 3,213 stroke patients, the 2,363 who had arrived at the hospital within 24 h of symptom onset and had available information regarding END were included. Of these, 318 (13.5%) had END. The LightGBM model showed the highest area under the receiver operating characteristic curve (0.772; 95% confidence interval, 0.715–0.829). The feature importance analysis revealed that fasting glucose level and the National Institute of Health Stroke Scale score were the most influential factors. Among ML algorithms, the LightGBM model was particularly useful for predicting END, as it revealed new and diverse predictors. Additionally, the effects of the features on the predictive power of the model were individualized using the SHAP method.

Highlights

We aimed to develop a novel prediction model for early neurological deterioration (END) based on an interpretable machine learning (ML) algorithm for atrial fibrillation (AF)-related stroke and to evaluate the prediction accuracy and feature importance of ML models
We first demonstrated that integrated ML algorithms can be applied to predict END in AF-related stroke cases
Among the ML models investigated, LightGBM had the best performance, with an area under the receiver operating characteristic curve (AUROC) value of 0.772. This is a novel method with efficient computational power and wide scalability for processing categorical, multidimensional, and incredibly large d atasets[20], which makes it a suitable ML model in clinical settings

Summary

Introduction

We aimed to develop a novel prediction model for early neurological deterioration (END) based on an interpretable machine learning (ML) algorithm for atrial fibrillation (AF)-related stroke and to evaluate the prediction accuracy and feature importance of ML models. Study designs have been used, with some studies preferring to define END according to specific stroke subtypes (e.g., cardioembolism), making each predictor and recent nomograms difficult to use in real-world clinical practice[3,4,5,6] Those obstacles make it difficult to design a prospective early detection and early interventional study. Several markers, including clinical, radiological, and laboratory findings, have been associated with END in AFrelated stroke[9,10,11] In those studies, using a single marker had limited predictive power, since the diverse biomarkers and imaging markers relevant to END in AF-related stroke were not considered at the same time. The aim of our study was to develop an interpretable ML model that could predict END using the feature importance technique in AF-related stroke using a real-world multicenter cohort database

Objectives

Methods

Results

Conclusion