e-Inclusion aims to provide the benefits of digital technology for every member of society. Digital skills and their meaningful use are a prerequisite for everyone to be e-included. The improvement of learning outputs of online and blended courses on digital skills is therefore an important aspect of ensuring an e-included society. Due to the use of learning management systems and their ability to collect data on students, different types of student data become available for analysis. We proposed the data-driven approach which uses student data and machine learning algorithms to predict learning outcomes. The goal of this article is to present the conceptual architecture and prototype of the e-inclusion prediction system which is based on a combination of several algorithms and uses a machine learning approach.