Machine learning techniques for classification problems related to therapies in diabetes patients

The growing use of technology and cyber tools has been embraced by the healthcare sector in many ways. An interesting and currently not completely exploited field of application is "patient engagement" . This thesis tackles the problem of classifying diabetes patients, with the use of machine learning, based on the therapy they are following in: patients that are following the correct therapy and patients that are not following the therapy, or for which the therapy is not correct.