Predicting diabetes complications through dynamic bayesian networks