Description

Dong et al reportetd the EVendo score to identify a patient with cirrhosis who may have high-risk esophageal varices. This was derive using machine learning. The authors are from University of California Los Angeles, Arizona State University and Oliver View-UCLA Medical Center.


Patient selection: cirrhosis

 

Parameters:

(1) INR

(2) platelet count in 10^9/L

(3) serum BUN in mg/dL

(4) serum AST in U/L

(5) hemoglobin in g/dL

(6) ascites

 

Parameter

Finding

Points

INR

 

(INR)

serum AST

 

(AST) / 35

platelet count

 

(platelet count) / 150

serum BUN

 

(BUN) / 20

hemoglobin

 

(hemoglobin) / 15

ascites

absent

0

 

present

1

 

where:

• The continuous variables were normalized to the upper limit of normal with the exception of the platelet count which was normalized to the lower limit of normal. This reflects the denominators in the point column.

 

points for numerator =

= (points for INR) + (points for AST) + 8.5

 

points for denominator =

= (points for platelet count) + (points for BUN) + (points for hemoglobin)

 

score =

= ((numerator) / (denominator)) + (points for ascites

 

Interpretation:

• A score <= 3.90 is associated with a low probability of high-risk esophageal varices.

 

Performance:

• The area under the ROC curve depends on the population and ranges from 0.74 to 0.84.


To read more or access our algorithms and calculators, please log in or register.