This paper reports the design of a deep neural network (DNN) focused to predict patterns of student dropout at the UPTC University of Colombia. We apply specialized artificial intelligence (AI) algorithms for the implementation of the DNN specifically were using a learning machine for classification and clustering task. Additionally, a dataset containing 17 attributes of 3000 academically active students was structured; the data set has been prepared to be trained as an entry to the neural network. The final result of the research reveals a predictive model trained through the DNN validated by several quality metrics that demonstrate reliability and accuracy of the results accomplished through the model.