Skip to main content

Table 3 Performance of the machine learning methods in terms of average (+/− std. dev.) sensitivity, specificity and area under the receiver operating characteristic (AUC), applied to the dataset with different missing value imputation techniques (complete cases, categorical variable encoding the missingness, mean/mode imputation, and random forest imputation)

From: Combining macula clinical signs and patient characteristics for age-related macular degeneration diagnosis: a machine learning approach

Type of imputation on the dataset (N = 444)

Performance

function

One-rule

Decision tree

Logistic regression

Random forest

AdaBoost

Support vector machine

Complete cases

AUC

0.74+/−0.05

0.90+/−0.03

0.93+/−0.04

0.94+/−0.01

0.92+/−0.02

0.92+/−0.03

Sensitivity

0.87+/−0.10

0.88+/−0.07

0.92+/−0.03

0.90+/−0.03

0.91+/−0.02

0.94+/−0.03

Specificity

0.60+/−0.18

0.74+/−0.15

0.70+/−0.08

0.78+/−0.07

0.71+/−0.06

0.67+/−0.07

Categorical variable encoding the missingness

AUC

0.73+/−0.04

0.88+/−0.02

0.91+/−0.01

0.92+/−0.02

0.90+/−0.01

0.89+/−0.03

Sensitivity

0.92+/−0.07

0.88+/−0.07

0.92+/−0.03

0.91+/−0.02

0.91+/−0.03

0.93+/−0.03

Specificity

0.42+/−0.05

0.61+/−0.18

0.60+/−0.07

0.68+/−0.06

0.60+/−0.06

0.51+/−0.07

Mean/mode

AUC

0.69+/−0.05

0.85+/−0.02

0.88+/−0.02

0.87+/−0.02

0.87+/−0.02

0.86+/−0.04

Sensitivity

0.94+/−0.05

0.92+/−0.04

0.94+/−0.02

0.93+/−0.02

0.93+/−0.02

0.96+/−0.02

Specificity

0.31+/−0.12

0.56+/−0.10

0.54+/−0.05

0.56+/−0.05

0.53+/−0.05

0.47+/−0.06

Random forest

AUC

0.79+/−0.02

0.95+/−0.02

0.96+/−0.01

0.96+/−0.01

0.96+/−0.01

0.94+/−0.03

Sensitivity

0.97+/−0.04

0.94+/−0.04

0.96+/−0.02

0.94+/−0.02

0.95+/−0.01

0.96+/−0.01

Specificity

0.60+/−0.06

0.78+/−0.09

0.75+/−0.05

0.81+/−0.04

0.76+/−0.04

0.75+/−0.05

  1. Results are calculated on 50 bootstrap tests, using out-of-bag predictions (in bold the best performance for each characteristic).