Table 3 Comparison of ML models’ performance.

From: The interpretable machine learning model for depression associated with heavy metals via EMR mining method

Characteristics

DNN

SVM

GNB

DT

XGB

GA-XGB

AAUC

0.657

(0.65,

0.664)

0.6

(0.594,

0.606)

0.608

(0.599,

0.616)

0.614

(0.608,

0.62)

0.686

(0.68,

0.69)

0.669

(0.663,

0.676)

BAUC

0.89

0.843

0.618

0.628

0.942

0.97

APS

0.06

0.047

0.044

0.038

0.062

0.068

Average recall

0.972

0.972

0.936

0.971

0.971

0.974

Average f1 score

0.958

0.958

0.941

0.957

0.957

0.964

Average accuracy

0.972

0.972

0.936

0.971

0.971

0.974

Average Brier score loss

0.028

0.029

0.064

0.029

0.027

0.026

Average cross-entropy loss

0.976

0.984

2.21

0.99

0.99

0.908

Average Jaccard index

0.944

0.944

0.91

0.943

0.944

0.949

Average Cohen’s kappa

0.031

0.011

0.04

0

0

0.178

  1. DNN: deep neural networks; SVM: support vector machine; GNB: Gaussian naive bayes; DT: decision tree classifier; XGB: extreme gradient boosting; AAUC: average area under the curve; BAUC: best area under the curve; APS: average precision score; NA: null.
Back to article page