Comparison among random forest, logistic regression, and existing clinical risk scores for predicting outcomes in patients with atrial fibrillation: A report from the J-RHYTHM registry



Watanabe, Eiichi, Noyama, Shunsuke, Kiyono, Ken, Inoue, Hiroshi, Atarashi, Hirotsugu, Okumura, Ken, Yamashita, Takeshi, Lip, Gregory YH ORCID: 0000-0002-7566-1626, Kodani, Eitaro and Origasa, Hideki
(2021) Comparison among random forest, logistic regression, and existing clinical risk scores for predicting outcomes in patients with atrial fibrillation: A report from the J-RHYTHM registry. CLINICAL CARDIOLOGY, 44 (9). pp. 1305-1315.

[thumbnail of Comparison among random forest, logistic regression, and existing clinical risk scores for predicting outcomes in patients w.pdf] Text
Comparison among random forest, logistic regression, and existing clinical risk scores for predicting outcomes in patients w.pdf - Published version

Download (2MB) | Preview

Abstract

<h4>Background</h4>Machine learning (ML) has emerged as a promising tool for risk stratification. However, few studies have applied ML to risk assessment of patients with atrial fibrillation (AF).<h4>Hypothesis</h4>We aimed to compare the performance of random forest (RF), logistic regression (LR), and conventional risk schemes in predicting the outcomes of AF.<h4>Methods</h4>We analyzed data from 7406 nonvalvular AF patients (median age 71 years, female 29.2%) enrolled in a nationwide AF registry (J-RHYTHM Registry) and who were followed for 2 years. The endpoints were thromboembolisms, major bleeding, and all-cause mortality. Models were generated from potential predictors using an RF model, stepwise LR model, and the thromboembolism (CHADS<sub>2</sub> and CHA<sub>2</sub> DS<sub>2</sub> -VASc) and major bleeding (HAS-BLED, ORBIT, and ATRIA) scores.<h4>Results</h4>For thromboembolisms, the C-statistic of the RF model was significantly higher than that of the LR model (0.66 vs. 0.59, p = .03) or CHA<sub>2</sub> DS<sub>2</sub> -VASc score (0.61, p < .01). For major bleeding, the C-statistic of RF was comparable to the LR (0.69 vs. 0.66, p = .07) and outperformed the HAS-BLED (0.61, p < .01) and ATRIA (0.62, p < .01) but not the ORBIT (0.67, p = .07). The C-statistic of RF for all-cause mortality was comparable to the LR (0.78 vs. 0.79, p = .21). The calibration plot for the RF model was more aligned with the observed events for major bleeding and all-cause mortality.<h4>Conclusions</h4>The RF model performed as well as or better than the LR model or existing clinical risk scores for predicting clinical outcomes of AF.

Item Type: Article
Uncontrolled Keywords: arrhythmia, bleeding, machine learning, mortality, stroke, thrombosis
Divisions: Faculty of Health and Life Sciences
Faculty of Health and Life Sciences > Institute of Life Courses and Medical Sciences
Depositing User: Symplectic Admin
Date Deposited: 13 Dec 2021 16:15
Last Modified: 18 Jan 2023 21:19
DOI: 10.1002/clc.23688
Related URLs:
URI: https://livrepository.liverpool.ac.uk/id/eprint/3145305