Optimising Temporal Segmentation of Multi-Modal Non-EEGSignals for Human Stress Analysis
This project tackles the challenge of analysing human stress levels by optimising how we divide time segments in data collected from various sensors beyond electroencephalography (EEG). The key question lies in how to best segment this multi-modal data over time. The project aims to find the optimal temporal segmentation strategies that effectively capture the dynamic changes in these diverse signals, ultimately improving the accuracy of stress analysis.
Publication Details
- R. Udhayakumar, S. Rahman, S. Gopakumar and C. Karmakar, “Nonlinear Features from Multi-Modal Signals for Continuous Stress Monitoring,” 2023 45th Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), Sydney, Australia, 2023
- Udhayakumar, R., Rahman, S., Buxi, D., Macefield, V. G., Dawood, T., Mellor, N., & Karmakar, C. (2023). Measurement of stress-induced sympathetic nervous activity using multi-wavelength PPG. Royal Society Open Science, 10(8), 221382.
Proposed Future Work Details
Future work involves investigation into the following avenues:
- Investigate methods for personalising the temporal segmentation based on individual characteristics or stress response patterns.
- Applying explainable deep learning methods to investigate stress predictors in complex multimodal signals.
- Translate the research findings into practical applications like stress management apps, workplace intervention programs, or mental health monitoring tools.