AUTHOR=Zhu Bin , Zhao Jianlei , Cao Mingnan , Du Wanliang , Yang Liuqing , Su Mingliang , Tian Yue , Wu Mingfen , Wu Tingxi , Wang Manxia , Zhao Xingquan , Zhao Zhigang 

TITLE=Predicting 1-Hour Thrombolysis Effect of r-tPA in Patients With Acute Ischemic Stroke Using Machine Learning Algorithm

JOURNAL=Frontiers in Pharmacology

VOLUME=Volume 12 - 2021

YEAR=2022

URL=https://www.frontiersin.org/journals/pharmacology/articles/10.3389/fphar.2021.759782

DOI=10.3389/fphar.2021.759782

ISSN=1663-9812

ABSTRACT=Background: Thrombolysis with r-tPA is recommended for patients after an acute ischemic stroke (AIS) within 4.5 h of symptom onset. However, only a few patients benefit from this therapeutic regimen. Thus, we aimed to develop an interpretable machine learning-based model to predict the thrombolysis effect of r-tPA at the super early stage.
Methods: A total of 353 patients with AIS were divided into training and test datasets. We then used six machine learning (ML) algorithms and a feature recursive elimination (FRE) method to explore the relationship among the clinical variables along with the NIH stroke scale (NIHSS) score 1 h after thrombolysis treatment. Shapley additive explanations (SHAP) and local interpretable model–agnostic explanation (LIME) algorithms were applied to interpret the ML models and determine the importance of the selected features.
Results: Altogether, 353 patients with an average age of 63.0 (56.0–71.0) years were enrolled in the study. Of them, 156 patients got favorable thrombolysis effect, and 197 patients showed unfavorable effect. A total of 14 variables were enrolled in the modeling, and 6 ML algorithms were used to predict the thrombolysis effect. After FRE screening, seven variables under the GBDT model (AUC = 0.81, specificity = 0.61, sensitivity = 0.9, and F1 score = 0.79) demonstrated the best performance. Of the seven variables, APTT (time), BNP, and FDP were the three most important clinical characteristics that might influence the r-tPA efficiency. 
Conclusion: This study demonstrated that the GBDT model with seven variables can better predict the early thrombolysis effect of r-tPA.