Validation of clinical prediction models: what does the “calibration slope” really measure? | Litlas