Chronic diseases have continued to put immense pressure on healthcare facilities, particularly in the developing world, through inefficient management besides high incidence and prevalence. Modern healthcare systems do not have individualized strategies thus limiting preventative steps and management of chronic diseases. This leads to a very important topic – medication non-adherence. This research intended to improve disease prediction and management by creating an effective model that can predict the possibilities of non-compliance with the medication among chronic disease patients. Using the cross-sectional qualitative data from a developing country and the CRISP-DM method, the study employed Recurrent Neural Networks. The model fared better than other approaches like Naïve Bayes, Decision Trees and Support Vector Machines in terms of predicting chances of missed medication doses among chronic patients. The successful performance of the model validates the usefulness of the proposed method in this critical application domain. The research provides real life recommendations on how communication can be useful in the case of chronic diseases and compliance with medication, which is a concern in health systems around the world. The model offers a better solution in a situation where other chronic cases could have been managed properly and required less facilities and cost.