I have conducted a no-conditional logistic regression analysis to know the relationship between spine osteoarthritis and vertebral fracture in males over 50 years.
The validation of the model has been as follows:
- Classification table: overall -correct- percentage, 81,5%.
- Hosmer-Lemeshow test: chi-square 9,47, p=0,30.
- ROC curve: the best pair, SE=76,3%, SP=61,1%; AUC=0,728 (95%CI 0,68-0,76), p
Need to test the colinearity issue. Sensitivity test may also be needed.
Hello, Jason.
While I was developing the model, I was checking continuously the assumptions of regression, including the absence of a major colinearity.
Anyway, you´re right and it may be of interest to point out the test of colinearity in the validation, besides the ROC analysis and the maximum sensivity and specificity.
Regards.
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis