Skip to main content

Table 4 Feature importance of ML predictive models of 10 years and lifetime ASCVD risk (random forest algorithm)

From: Cardiovascular risk assessment using ASCVD risk score in fibromyalgia: a single-centre, retrospective study using “traditional” case control methodology and “novel” machine learning

 

Feature importance in 10-year ASCVD risk ML predictive model

Feature importance in lifetime ASCVD risk ML predictive model

Age

0.150

0.074

Height

0.063

0.051

Weight

0.078

0.035

BMI

0.065

0.039

Systolic blood pressure

0.043

0.069

Diastolic blood pressure

0.030

0.037

Diabetes mellitus

0.041

0.007

Hypertension

0.006

0.015

Dyslipidaemia

0.012

0.009

Hypothyroidism

0.023

0.013

Smoker

0.000

0.000

Disease duration

0.036

0.043

FIQR score

0.071

0.062

BPI total score

0.038

0.029

BPI average score

0.066

0.047

ESR

0.036

0.044

CRP

0.040

0.050

Serum total cholesterol

0.077

0.179

Serum LDL

0.041

0.083

Serum HDL

0.022

0.051

Serum triglycerides

0.070

0.055

  1. Feature importance is the degree of contribution in the random forest ML predictive model. Seven parameters of high feature importance are emboldened in each ML classifier model
  2. ML, Machine learning; ASCVD, atherosclerotic cardiovascular disease; BMI, body mass index; FIQR, revised fibromyalgia impact questionnaire; BPI, brief pain inventory; ESR, erythrocyte sedimentation rate; CRP, C-reactive protein; LDL, low-density lipoprotein; HDL, high-density lipoprotein