Remote Detection of Psychological
Physiological stress can be characterized as
dominance of the sympathetic branch of the autonomic
nervous system over the parasympathetic branch.
For example, when a person goes from a sitting
position to the more physically stressful act
of running, activity in the cardiac sympathetic
nerves increases, and activity in the parasympathetic
nerves decreases, leading to a faster heart rate
and increased blood pressure among other things.
Besides drastic changes in sympathetic/parasympathetic
balance as in this example, there may be more
subtle changes occurring on short timescales (seconds)
in response to psychological stress (for example,
not telling the truth while being questioned)
or pathology (for example, internal hemorrhage
of a patient in the intensive care unit).
Being able to track a person's physiological
state noninvasively and on short time-scales would
be beneficial from a basic science perspective
as well as possibly improving the clinical outcomes
of patients and in intelligence situations. Currently,
there is no way of noninvasively measuring sympathetic
or parasympathetic nervous tone, although several
methods for estimating it exist, like for example,
cardiovascular system identification (CSI). However,
the present estimation methods require minutes
of data, and are more effective at tracking general
state rather than the potentially more important
fluctuations on a beat-to-beat basis.
The goal of this project is to develop a noninvasive
method for estimation of physiological stress.
We aim to accomplish this using a few, easily
measured signals, like the sequence of heart beat
times and respiration. This data is routinely
measured in intensive care wards, where the application
of stress-monitoring may help predict unstable
or decompensated states before they manifest with
life-threatening symptoms. On the other hand,
the data could be collected in a non-contact manner
from subjects during a conversation, and stress-monitoring
could alert the questioner about potentially sensitive
topics or whether the subject is being deceitful
in responses to particular queries.