Academic Performance Patterns Detection Using Digital Phenotyping

In this article, some factors that influence poor academic performance are studied by analyzing the students’ digital phenotype. The study considers passive and ac-tive data obtained from 15 undergraduate volunteer students, collected over a two-week period. Although different problems arose with the operating system of mobile devices and other concerns, the developers solved the issues but redu-ced the participants from 31 to only 15. Data collection was carried out using a free software tool, storing the content in a database hosted in the cloud. Collected data was analyzed with propertary algorithms using machine learning libraries in Python. The obtained results show that people with high social interaction and those who had better sleep quality show better academic performance, while tho-se who showed greater mobility from school to their homes, exhibit lower per-formance. These results represent the usefulness of digital phenotyping to detect students with the possibility of presenting low academic performance