- Ana Bogdanovic, M. Sc. student in Biomedical Engineering
- Lorenzo Gecchelin, M. Sc. student in Design & Engineering
- Anisia Lauditi, M. Sc. student in Biomedical Engineering
- Noemi Gozzi, M. Sc. student in Biomedical Engineering
- Armando Bellante, M. Sc. student in Computer Science & Engineering
- Letizia Bergamasco, M. Sc. student in ICT for Smart Societies
- Moaad Khamlich, M. Sc. student in Computational Engineering
Stress is a psycho-physical response to very different loads, of an emotional, cognitive or social nature, which is perceived as excessive thus having severe implications on wellbeing both in the short and in the long term. Different physiological manifestations occur during stressful events which, if detected promptly, can help in managing the situation. Therefore, the objective of this project is to develop a small portable device for psychological stress detection. This includes design of machine learning framework for stress detection and a prototype of a low-cost portable device for recording the physiological data. The ML framework is including the model together with the heuristic and knowledge based feature engineering from physiological time series. As a result EMoCy system is achieving accuracy of 97.2 ± 2% on stress/baseline binary classification task.
6. Statistics
17%
Communicable, maternal, perinatal
and nutritional conditions
6%
Injuries
31%
Mental, neurological
and substance use
disorders
14%
Musculoskeletal
diseases
31%
Other non-
communicable diseases
(e.g. CVD, cancer,
diabetes, respiratory
diseases)
10% Depression
4% Anxiety disorder
s
4% Alcohol use disorders
14% Other disorders
Global distribution
of non-fatal disease
burden of disease
(years lived with disability
)
Mnookin, S. (2016). Out of the Shadows: Making Mental Health a Global Development Priority. [Report] World Health Organization and World Bank
Group. Available at: https://www.who.int/mental_health/advocacy/wb_background_paper.pdf?ua=1 [Accessed 7 Nov. 2019].
6
14. CONSEQUENCESRESULTS
less computational power
reduced power consumption
less sophisticated sensors
better accuracy (97.2%)
lower sampling frequency (64 Hz)
smaller windows (60s)
fewer signals (EDA, RESP, PPG)
14
Emocy - So Far
15. better accuracy (97.2%)
lower sampling frequency (64 Hz)
smaller windows (60s)
fewer signals (EDA, RESP, PPG)
CONSEQUENCESRESULTS
less computational power
reduced power consumption
less sophisticated sensors
WEARABLE DEVICE
15
Emocy - So Far