skin conductance rises or the characteristic shape of hotflash events, others have
used more sophisticated algorithms to classify hotflashes. For example, Thurston
et al. showed high sensitivity and specificity of Support Vector Machine approaches
to classifying hotflashes (Thurston et al.2009a,2011a). Others have also employed
a pattern-matching algorithm to identify hot flashes in data from the Bahr
Monitor™(Stefanopoulou and Hunter 2013 ). The investigation into the optimal
algorithm to classify hotflashes from skin conductance is ongoing.
The use of sternal skin conductance to detect hotflashes has several advantages,
but some notable disadvantages. One of the primary advantages is that this measure
does not rely upon the subjective reporting of hotflashes which can be influenced
by the range of factors, such as emotions, physical state, and adherence to reporting.
Skin conductance monitoring for hot flashes can record events continuously,
including during sleep. It does not manifest the placebo response common with
self-reported hotflashes (Carpenter et al. 2007 ). However, there are a number of
key limitations to SCL measures of hotflashes, such as artifact in signals created by
physical activity and other types of sweating, issues with electrode adhesive and
gels, participant burden and discomfort, and the limitations of current hotflash
detection algorithms.
Concordance/Discordance Between Physiologic
and Subjective Hot Flash Measures
Physiologic hotflashes measures have typically been validated using participant
self-report of hotflashes as the referent in controlled laboratory settings. Early
studies showed very high concordance between subjective report and SCL-detected
hotflashes (Freedman 1989 ; Tataryn et al. 1981 ). A meta-analysis of hotflash
studies utilizing physiologic hotflash measures was performed analyzing the data
from studies that assessed peri- and postmenopausal hotflashes via skin conduc-
tance in controlled laboratory and ambulatory environments (Mann and Hunter
2011 ). Analyses revealed that in laboratory studies, overall sensitivity across all
studies was 0.69 and specificity was 0.97, suggesting that sternal skin conductance
is a reasonable indicator of hotflashes in a controlled environment (Fig.11.1).
However, in ambulatory settings, the concordance between SCL measures of hot
flashes and self-reported hotflashes has been consistently lower (Carpenter et al.
2012 ; Thurston et al. 2009a; Carpenter and Rand 2008 ; Otte et al. 2009 ).
Specifically, SCL measures show lower average sensitivity in ambulatory studies
(39%) than they do in laboratory studies (69%) (Mann and Hunter 2011 ). In
contrast, average specificity of SCL measures is consistently high (97%). As these
studies employed participant self-report as the referent, a range of factors may
account for the lower sensitivity in ambulatory settings including increased artifact
in skin conductance measures, distractions away from bodily sensations, and
competing activities that may decrease adherence to hotflash reporting.
240 W.I. Fisher and R.C. Thurston