Skip to main content

Table 3 Test sets result of machine learning models and the Logistic model on 90-day stroke outcome prediction

From: Machine learning is an effective method to predict the 90-day prognosis of patients with transient ischemic attack and minor stroke

Model

Auc(95%CI)

Accuracy(95%CI)

PPV(95%CI)

NPV(95%CI)

F1-score(95%CI)

Brier score(95%CI)

CB

0.839(0.823,0.854)

0.942(0.938,0.947)

0.660(0.605, 0.716)

0.951(0.948,0.954)

0.404(0.382,0.427)

0.047(0.044,0.050)

XGB

0.838(0.822,0.853)

0.943(0.939,0.947)

0.664(0.595,0.734)

0.952(0.949,0.955)

0.423(0.394,0.452)

0.047(0.044,0.050)

GBDT

0.835(0.820,0.850)

0.942(0.938,0.946)

0.648(0.589,0.707)

0.951(0.948,0.954)

0.403(0.377,0.428)

0.047(0.044,0.050)

RF

0.832(0.815,0.849)

0.940(0.937,0.943)

0.659(0.595,0.723)

0.946(0.944,0.949)

0.326(0.303,0.348)

0.048(0.045,0.051)

Ada

0.823(0.810,0.837)

0.941(0.938,0.945)

0.636(0.570,0.702)

0.951(0.949,0.953)

0.395(0.366,0.424)

0.159(0.157,0.161)

LRa

0.822(0.813,0.831)

0.941(0.938,0.945)

0.685(0.635,0.735)

0.947(0.944,0.951)

0.348(0.320,0.376)

0.048(0.046,0.051)

  1. aLR Logistic regression model