Table 4 Areas under the receiver operating characteristic curve for various prediction models in the Northwestern Medicine Enterprise Data Warehouse Heart Failure Cohort.

From: A machine learning model for identifying patients at risk for wild-type transthyretin amyloid cardiomyopathy

Model

N

AUROC

ATTRwt-CM RF model

39,654

0.80

ATTRwt-CM RF model, age > 70 years

23,570

0.82

Age only

39,624

0.54

Age + sex

39,618

0.62

Age + sex + ethnicitya

39,203

0.70

Age + sex + ethnicity + logBNPb

20,419

0.73

Age + sex + ethnicity + logBNP + abnormal troponin-Ic

15,046

0.73

ATTRwt-CM RF model + age + sex + ethnicity

39,203

0.83

ATTRwt-CM RF model + age + sex + ethnicity + total number of encounters

38,337

0.83

  1. ATTRwt-CM amyloidogenic transthyretin (wild-type), AUROC area under the receiver operating characteristic curve, RF Random Forest, BNP B-type natriuretic peptide.
  2. aEthnicity categories: non-Hispanic White, non-Hispanic Black, Hispanic, Asian, others.
  3. bHighest BNP value in the electronic health record, log-transformed.
  4. cBased on the highest troponin-I in the electronic health record (abnormal defined as >0.04 ng/ml).
Back to article page