Correct and reliable annotations of ECG is core to modern AI and ML approaches to signal processing – the algorithms need the "ground truth" in order to learn. However, annotation of ECG and identification of heart arrhythmia episodes require detailed and intensive work by clinicians (ECG technicians, nurses, or cardiologists) who has to manually look at, and annotate long strips of ECG in 10 second bits. A very time consuming and tiresome process.
The idea of this project – or rather project family (there are room for more projects) – is to design, implement, and evaluate a gamification approach to ECG annotation. We envision creating a small mobile game, in which ECG specialists can compete against each other in doing annotations. This is in line with the trend on "citizen science" where people are crowd sourced to help in science.
The project would entail a number of different tasks, including:
- reviewing existing work on citizen science and crowd sourcing in science
- review of ECG annotation methods and best practice
- employ user-centred design methods for the design and usability evaluation of the mobile game
- implementation of the mobile game
- run a small study involving e.g. biomedical students for annotations
This project builds upon and will extend the CACHET Research Package, which has a tool for collection of ECG (mCardia) and a small annotation tool to take inspiration from. The Research Package is implemented in Flutter, which is a novel cross-platform (Android, iOS, web) user-interface framework.
Thesis type: Digital Health UX Design OR Software implementation with a small user study.
Technical skills: Requirement analysis; UX design, Mobile programming in Flutter; Usability evaluation.
Research skills: Software Engineering; Human-Computer Interaction; Biomedical signal analysis of ECG data
I samarbejde med
Copenhagen Center for Health Technology
