I have to draw a ROC curve for my multivariate model. But in my multivariate model I have some variables having Odds ratio less than 1 e.g. 0.7, 0.1, 0.4 etc. But P values for all these variables are significant. Can I put these variables in ROC curve analysis?
You can still include those covariates in the model, if, based on your content expertise, they should be considered as covariates - even if they are not statistically significant on their own. In fact, you should consider interaction terms for covariates as well. In the end, for an ROC curve, you are interested in a model that generates a classification value that comes closest to the reference value. I am attaching a link to a paper I wrote on the topic several years ago, that you may find helpful.
Article Measuring diagnostic and predictive accuracy in disease mana...
Hello Mr.Hussain, recently we have developed an ROC model that is of multivariate in nature. If you wish you can interact with me.
- Badges
- Science topic
- Similar topics
- Chemistry
- Pharmacology
- Medical Pharmacology
- Clinical Pharmacology