Skip to main content

Table 4 Top Predictors in the Random Survival Forest Model fitted to the Cohort of Norway Training Data

From: Identifying long-term and imminent suicide predictors in a general population and a clinical sample with machine learning

Combined Sexes (n=540)

Males Only (n=305)

Females Only (n=235)

Variable

Importance

p

Importance

p

Importance

p

Male

1.45

<0.01

—

—

—

—

Proportion of county with low income

0.66

<0.01

0.50

0.03

0.63

0.01

Lives with spouse/partner

0.56

<0.01

0.58

0.01

—

—

Mood symptoms

0.56

<0.01

0.62

0.01

0.69

0.01

Daily hours spent in smoke-filled rooms

0.50

<0.01

0.57

0.03

—

—

Daily smoking

0.44

<0.01

0.36

0.01

0.56

<0.01

Waist-hip ratio

0.32

0.02

—

—

—

—

Married

0.27

0.01

0.71

<0.01

—

—

Alcohol use

0.20

0.05

—

—

—

—

Takes blood pressure medications

–

–

0.14

0.05

—

—

  1. Note: Importance is determined by Altmann’s permutation method [57] in which the permuted values of a variable are compared with the true values. Greater decreases in prediction accuracy reflect higher importance.