Using Stacking methods based Genetic Algorithm to predict the time between symptom onset and hospital arrival in stroke patients and its related factors

Firouz Amani; Jafar Abdollahi; Alireza Mohammadnia; Paniz Amani; Ghasem  Fattahzadeh-Ardalani

doi:10.18502/jbe.v8i1.10401

Firouz Amani Department of Biostatistics and Community Medicine, School of Medicine, Ardabil University of Medical Science, Ardabil, Iran.
Jafar Abdollahi Department of Computer Engineering, Ardabil Branch, Islamic Azad University, Ardabil, Iran.
Alireza Mohammadnia Department of Basic Science, School of Medicine, Ardabil University of Medical Science, Ardabil, Iran.
Paniz Amani Student in Electronic Engineering, Tabriz University, Tabriz, Iran.
Ghasem Fattahzadeh-Ardalani Department of Neurology, School of Medicine, Ardabil University of Medical Science, Ardabil, Iran.

DOI: https://doi.org/10.18502/jbe.v8i1.10401

Keywords: Stroke; Machine learning; Classification; Random Forest; Hospital.

Abstract

Introduction: The early arrival of patients with acute ischemic stroke to start of treatment by recombinant tissue plasminogen activator (rt-PA) within 4.5 hours after onset of stroke and its modeling by data mining methods is an important issue in care of stroke patients. In this paper, the aim was to provide methods to predict the time between symptom onset and hospital arrival in stroke patients and related factors, in addition to improve classification in minority class data, also to maintain the ability of classifying majority class data at an acceptable level.

Methods: We included 676 patients with ischemic stroke who referred to hospital of Ardabil city in the northwest of Iran in 2018. A new method using a combination of machine learning algorithms and genetic algorithms has been proposed to solve this problem. The performances were evaluated with accuracy, sensitivity, specificity, positive predictive value, and negative predictive value. Results: In this study, the stacking technique provides a better result (accuracy 99.51%, sensitivity 100%, and specificity 99.40%) among all other techniques.

Conclusion: Results of this study showed that this model can be used as a valuable tool for clinical decision making.