Multifractality in typing as a marker of fatigue
Multifractality is a property of biological signals that is informative about the effectiveness of behavioural coordination. Since fatigue is known to undermine the ability to coordinate behavioural resources, we hypothesise that multifractal signatures may function as effective markers of fatigue. To further this line of research we conduct a study to investigate the association between fatigue and multifractal signatures during typing. We observe that multifractality in keystroke timings is significantly lower after exposure to a mentally fatiguing task, and that multifractal properties correlate with the degree of self-reported fatigue. We discuss the implications of this research for the development of unobtrusive, low cost fatigue monitoring and interventions, noting that multifractal analysis requires only the capture of input behaviour for the task, and that multifractality has been observed in a range of task-directed behaviours, not only typing. This points to the potential to infer fatigue via existing commodity input hardware, in a range of tasks. • Multifractal properties have promise for inferring fatigue during office tasks. • Fatigued typing is found to be significantly less multifractal. • The approach is unobtrusive, easy to deploy, and privacy supportive. • Potential applications in a range of tasks beyond typing.
Authors
- Oussama Metatla (ORCID: https://orcid.org/0000-0002-3882-3243)
- Dan Bennett
- Anne Roudaut (ORCID: https://orcid.org/0000-0003-3082-4564)
Institutions
- University of Bristol (GB)
Publication Details
- Journal
- International Journal of Human-Computer Studies
- Published
- 2025-08-18
- DOI
- https://doi.org/10.1016/j.ijhcs.2025.103595
- Citations
- 1
- Primary Topic
- Engineering Diagnostics and Reliability
- Type
- article
- Field-Weighted Citation Impact
- 0.86
Funders
- UK Research and Innovation
- European Commission
- Engineering and Physical Sciences Research Council
- European Research Council
- HORIZON EUROPE European Research Council