2
$\begingroup$

I am using the rmspackage to draw a nomogram to represent a logistic regression model fit.

ddist<-datadist(data)

options(datadist='ddist')

fit<-lrm(Outcome~Grade+CF+HF, x=TRUE, y=TRUE, data=data)

The model returns:

enter image description here

I plot the nomogram:

nom<-nomogram(fit, lp=FALSE)
print(nom)

I receive the following points for each level of my predictors:

enter image description here

My first question is about the point values assigned Grade. If I were drawing this nomogram by hand from the Beta coefficients in fit, my highest B coefficient is -4.1537(Grade=3). The next highest B coefficient is 2.9472 (CF=20).

2.9472/4.1537=0.71. 71 is the point value is equal to CF 20. This matches the points on the nomogram. I've manually added the red line to this value on the nomogram above.

To calculate the next point value, I select the third highest B coefficient, -2.5917 (Grade=2).

-2.5917/-4.154=.624. Instead of 62 being assigned to the Grade=2, the nomogram is assigned 38. I do not understand how this value was assigned instead of 62.

Grade is a factor in my dataframe data

> str(Grade)


Factor w/ 3 levels "1","2","3": 2 2 2 2 3 2 2 1 2 1 ...

My second question is about the Total Points calculation of 220. If I select the predictor values with the highest possible points Grade 1+CF 20+ HF 4 (or 100+71+28) my total points would be 199. Why does the total points scale go to 220 when there is no combination of values that equal 220 points?

Thank you for any help. I appreciate your time.

$\endgroup$
  • 1
    $\begingroup$ I think you might be better on R-help or StackOverflow as your main issue here seems to me to be programming rather than the statistical question. If you think you do have a predominantly statistical question perhaps you can edit to make it more salient. $\endgroup$ – mdewey Dec 6 '16 at 21:43
  • 3
    $\begingroup$ I think there is a statistical-ish question here about interpreting/understanding points-based nomograms. $\endgroup$ – Glen_b Dec 6 '16 at 23:14
1
$\begingroup$

The grade is a factor value. The highest beta coefficient is -4.1537(Grade=3). Please be aware the negative symbol.When you calculate the points for Grade=2, you should use (4.1537-2.5917)/4.1537=0.376. This points is in accordance of the nomograms your constructed.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.