![](bg8.png)
Of course, the upside frequent data elicitation is
that the user gets support with managing his/her men-
tal resources during work, and this support becomes
more accurate the more data is fed into the system.
Another concern is intrusion to privacy when an-
alyzing webcam video streams of the user. It is,
however, important to note that only the screen co-
ordinates denoting the location of the user’s gaze are
logged. No other video data is analyzed or stored.
In general, the use of the software is entirely vol-
untary. The aim with colleting the data in the first
place is to support the user, and the software con-
tributes a major benefit for the user through the in-
dividually adapted advice that can be offered to users
to help them become more efficient at their work and
find a better balance between working and resting.
6 CONCLUSIONS
We have presented a tailored software suite for data
collection for machine learning. Using this software
data collection can be run unobtrusively as a back-
ground process on the user’s desktop or laptop. The
software continuously monitors and logs a vast array
of data, ranging from gaze, where the active window
is located on the computer screen, and key and mouse
actions. In addition, the software also queries the
user’s heartrate (with the user’s permission) from a
FitBit server.
In connection to data storage, a machine learning
operations flow has been set up, where data are pre-
processed and used for training individually adapted
user models, with the aim to predict the user’s cogni-
tive state.
The data collection software seems to be relatively
lightweight, so that it can be run on the user’s com-
puter without too much increase in CPU or memory
load. Also, the sensor part of the software has been di-
vided into multiple subprocesses, each of which can
be run on a separate CPU core, which allows for full
optimization on an operating system level.
A major outstanding question is if training of In-
dividual models, using only data from that user works
as expected. Specifically, it remains to be seen if the
data collected from each user is enough, that is the
data set is large enough, for learning a useful model
for each user.
ACKNOWLEDGEMENTS
This work was supported by the SmartWork project
(GA 826343), EU H2020, SC1-DTH-03-2018 -
Adaptive smart working and living environments sup-
porting active and healthy ageing.
REFERENCES
Cohen, S., Kamarck, T., and Mermelstein, R. (1983). A
global measure of perceived stress. Journal of health
and social behavior, pages 385–396.
Cohen, S., Kamarck, T., and Mermelstein, R. (1994). Per-
ceived stress scale. Measuring stress: A guide for
health and social scientists. 10:5.
Cole, S. R. (1999). Assessment of differential item func-
tioning in the Perceived Stress Scale-10. Journal
of Epidemiology and Community Health, 53(5):319–
320.
Hart, S. G. (2006). Nasa-Task Load Index (NASA-TLX);
20 Years Later. Proceedings of the Human Factors
and Ergonomics Society Annual Meeting, 50(9):904–
908.
Hjortskov, N., Riss
´
en, D., Blangsted, A. K., Fallentin, N.,
Lundberg, U., and Søgaard, K. (2004). The effect of
mental stress on heart rate variability and blood pres-
sure during computer work. European journal of ap-
plied physiology, 92(1-2):84–89.
Inoue, A., Kawakami, N., Shimomitsu, T., Tsutsumi, A.,
Haratani, T., Yoshikawa, T., Shimazu, A., and Oda-
giri, Y. (2014). Development of a Short Version of
the New Brief Job Stress Questionnaire. Industrial
Health, 52(6):535–540.
Mitchell, A. M., Crane, P. A., and Kim, Y. (2008). Perceived
stress in survivors of suicide: Psychometric properties
of the Perceived Stress Scale. Research in Nursing &
Health, 31(6):576–585.
Moreno-Esteva, E. G. and Hannula, M. S. (2015). Us-
ing gaze tracking technology to study student visual
attention during teacher’s presentation on board. In
Krainer, K. and Vondrov
´
a, N., editors, CERME 9 -
Ninth Congress of the European Society for Research
in Mathematics Education, Proceedings of the Ninth
Congress of the European Society for Research in
Mathematics Education, pages 1393–1399, Prague,
Czech Republic. Charles University in Prague, Fac-
ulty of Education and ERME.
Neuberger, G. B. (2003). Measures of fatigue: The Fa-
tigue Questionnaire, Fatigue Severity Scale, Multi-
dimensional Assessment of Fatigue Scale, and Short
Form-36 Vitality (Energy/Fatigue) Subscale of the
Short Form Health Survey. Arthritis Care & Research,
49(S5):S175–S183.
Phan, N. Q., Blome, C., Fritz, F., s, J. G., Reich, A., Ebata,
T., Augustin, M., Szepietowski, J. C., and St
¨
ander, S.
(2012). Assessment of pruritus intensity: Prospective
study on validity and reliability of the visual analogue
scale, numerical rating scale and verbal rating scale
in 471 patients with chronic pruritus. Acta dermato-
venereologica, 92(5):502–507.
Qian, S., Li, M., Li, G., Liu, K., Li, B., Jiang, Q., Li, L.,
Yang, Z., and Sun, G. (2015). Environmental heat
Collecting Data for Machine Learning on Office Workers’ Attention, Fatigue, Overload, and Stress during Computer Use
475