Hello,
I recently stumbled on a paper in the field of neurosurgery. The paper analyzes imaging parameters that predict injury related to the spinal cord (Posterior Ligamentous Complex Injuries). I have two concerns related to the paper and I need assistance from a statistician
1) Is that possible that they mention that one parameter is associated with positive association (OR > 1) with PLC injury in axial section but not associated with PLC injury in sagittal section (OR < 1)? This didn't make sense because it is the same parameter but once viewed in axial view and another time viewed in sagittal view.
2) My second concern is that whether it is possible to use the parameter that was associated with negative association (OR < 1) to create ROC curve and measure its predictive power (sensitivity, specifity, etc)? This this didn't make sense again because how come that one parameter is associated with negative odds ratio can be used to predict the dependent variable and be used to create a cut-off value that predict this dependent variable?
Can you supply your results with references?
Thanks in advance.
- Badges
- Science topic
- Similar topics
- Filing