Diabetes Mel l it us Model (DMM): internal validation of a computer simulation model for type 1 and type 2 diabetes

Summary

The aim of the analysis described in this paper was to confirm the internal validity of the Diabetes Mellitus Model (DMM), which is an epidemiological simulation model for the prediction of short- and long-term complications of diabetes. For the validation, mean values and confidence intervals (CIs) of simulated event rates (ERs) were calculated. The expected ERs were derived from publications. Internal validity of the DMM was defined as an agreement (overlap) between the model outputs and the published data allowing for a range of 25% deviation from the original data.

The results of the validation process revealed coherence between mean simulated ERs and expected ERs for most of the examined events. A fit of the range of CIs within the range of expected ERs can be observed for macular oedema in type 1 and type 2 diabetes, for retinopathy in type 1 diabetes, and for amputation and diabetic foot syndrome in type 2 diabetes. With the exception of end-stage renal disease, the CIs and ranges of the other events overlap significantly, supporting the view that the model can be considered as internally valid.

These results substantiate the DMM as an internally valid diabetes model that predicts complication rates consistent with observed rates. The DMM is a valuable tool for medical decision-making. Further research is required to provide external validation of the model.

Key words: :

• computer simulation model
• diabetes
• internal validation