The HPI students Valentin Döring, Tobias Fiedler, Lucas Liebe, Leander Masopust, Kirill Postnov, Franz Sauerwald, Felix Treykorn, and Alexander Wischmann won a Machine Learning Challenge for the recognition of various activities at a Winter School on e-Health & Pervasive Technologies held in North Macedonia as part of their bachelor project at the chair of Digital Health - Connected Healthcare. The school was sponsored and hosted by the EU Horizon 2020 project WideHealth. The chair of Digital Health - Connected Healthcare is a partner.
As small, networked, body-worn computer systems, wearables like smartwatches and fitness trackers have become an integral part of everyday life. The inbuilt sensors generates a wide range of data on various health aspects such as sleep quality, mobility or calories burned. Further, the data stream can be used to recognize multiple activities with the help of suitable methods in machine learning. This field of research, known as activity recognition, is already an integral part of various wearables for specific sporting activities or detecting physical inactivity. This is made possible by identifying distinct movement patterns in the sensor data.
In this year's bachelor project, Sensor-based Nursing Activity App, at the chair Digital Health - Connected Healthcare, an app-based solution for automatically recognizing different nursing activities with wearables is being developed. The project is supervised by Orhan Konak and Prof. Dr. Bert Arnrich.As part of a Winter School of the EU project WideHealth, a machine learning competition for the recognition of 10 different activities with wearables was conducted at the Faculty of Electrical Engineering and Information Technologies of the University of Skopje in North Macedonia from 14th to 17th February 2022.
HPI students competed as a collaborative bachelor project team against teams from different countries. A total of 11 teams and 51 participants took part in the competition. The other teams consisted of bachelor's/master's students, doctoral students, and professionals with industry experience. Multiple classification results were allowed to be submitted per team per day. A total of 258 submissions were made, the most accurate was achieved by the HPI students. For their excellent performance, they were rewarded with a prize worth 300€.
