Using multiclass machine learning model to improve outcome prediction of acute ischemic stroke patients after reperfusion therapy

摘要

This study tempts to develop a multiclass algorithm-based prediction model on 3-month outcome of acute ischemic stroke patients with reperfusion therapy. Patients of Acute ischemic stroke requiring reperfusion therapy were recruited from January 2016 to December 2019 in Kaohsiung Chang Gung memorial Hospital. Multiclass machine learning model include Logistic Regression, Supportive Vector Machine, Random Forest, extreme Gradient Boosting were used for training in compare with DRAGON score, a traditional clinical score based on statistics model, on 3-month clinical outcome. Compare with DRAGON score, multiclass machine learning approach is associate with better accuracy, along with better precision, recall and specificity on predicting 90-day functional outcome classifications of acute ischemic stroke patients requiring reperfusion therapy.