Repeated enflurane anaesthetics and model predictions: a study of the variability in the predictive performance measures.

We quantified the total variability (reproducibility) and the within-patient but between repeat anaesthetics variability (repeatability) in measures which are used to judge the predictive performance of our physiological model. We studied 14 patients who received enflurane closed-circuit anaesthesia on two occasions. The end-tidal concentrations measured and those predicted served to calculate the predictive performance measures of the model: root mean squared error (rmse = total error), bias (systematic error) and scatter (error around the bias). The overall results were: rmse 15 (7)%, bias 0 (14)% and scatter 9 (3)% (grand mean (total SD)). The within-patient SD values were smaller for the rmse (4%) and bias (10%), but not for scatter (3%). The repeat rmse values and biases were linked to the first results. This implies that these performance measures depended partly on the patient. As there was no association between the personal performance measures and age, sex, body weight, body surface area or body mass index, these characteristics cannot be used to further tune the model.