A 13-state homogeneous semi-Markov model for predicting the HIV disease evolution: a case study