IEEE Consumer Electronics Magazine - July/August 2022 - 40

Home Healthcare Technologies & Services
Indeed, the cut-off point (CP) was found to be equal
to 111.50 (112), that is the lower possible limit of
normality reported in the work of Martin.38
A more in-depth look reveals that this ROC
Figure 7. ECG versus PPG and ROC curve.
normal in comparison to data of Table 1, could
imply a bradycardia.
It is essential though to understand that an
embryo's heart starts to work at the very first time
at about 21 to 22 days after the conception, while
arrhythmias could emerge during the second trimester
of pregnancy. Furthermore, fetus arrhythmias
usually appear during night hours before
pregnant goes to sleep and consequently not during
the regular ultrasound examination. Therefore,
the test procedures involved in our system have
been conducted even at night hours. All findings
have been examined from qualified physicians for
further repeated checks and, finally, all four pregnant
women gave birth to healthy babies.
Nonetheless, another statistical test had already
been conducted, namely the so-called independentsamples
t-test, based on which the group measurements
of Pregn3 and Pregn4 have been declared as
abnormal, whereas measurements of Pregn1 and
Pregn2 have been considered normal. From the
observation of Table 3 (taking into consideration
the rejection of the null-hypothesis) it is evident
that a mean-value of 130 pulses (with a standard
deviation of almost 11.80) has been observed in the
normal range, whereas, a mean value of 104.12
pulses (with a standard deviation of almost 13.32)
was found abnormal. This is based on the fact of
assuring statistically significantly differences
between the mean values. This evidence, empowered
our position relevant to the fact that Pregn3
and Pregn4 needed further screening processes
before their babies' birth.
In order to shed light to the previous findings
and for validation purposes, we have conducted a
ROCanalysis. Byaccepting that heart-beatmeasurements
from Pregn1 and Pregn2 are inside normality
and in-turn by assuming that data were retrieved
from Pregn3 and Pregn4 are within the abnormal
area of heart behavior (as previously discussed),
then the cut-off point should be calculated as the
lower possible limit of normal heart-beat behavior.
40
point has a sensitivity of 97.7%, revealing that there
is the same possibility that the point (heart beat) is
within the normal region (positive). This point has
also a 1-specificity equal to 31.3%, revealing that
the abnormal results (negatives) exhibit the latter
possibility to be falsely interpreted as positive
ones (false positives). Furthermore, the area under
curve (AUC) was found to be equal to 91.5%, which
in turn is interpreted as an increased validity. Note
that AUC gives an aggregate performance measure
throughout all possible classification thresholds
and it ranges from0 to 1. Results close to 0.0 denote
that all predictions are almost 100% false and those
close to 1 (as in our case) then there is a strong possibility
that our prediction is true (CP112 bpms),
as also being confirmed from Martin.38
Many problems occurring during pregnancy can
be efficiently addressed using the proposed system
configuration. Poor oxygenation of the fetus may
lead to alterations of the heart rate, mainly bradycardia
and fluctuations of the fetal heart rate. In
addition, numerous other conditions, either already
known or underlying, namely maternal thyroid dysfunction,
maternal infection, chorioamnionitis, etc.,
can also report fetal heart alterations. Therefore,
the proposed systemmay have lifesaving effects on
fetal well-being. Moreover, the proposed FHR monitoring
implementation has a broad scope of influence,
with the potential to be efficiently used by
pregnant women during COVID-19 home quarantine.
Specifically, it canbe used i) to detect andautomatically
alert for emergency intervention, and ii) to
detect fetus and mother underreporting physiological
metrics which could be proved valuable for
fetus and mother health when a proper automatic
measuring system can alert on the aforementioned
findings. Consequently, this project may be useful
for all those directly involved in the healthcare
industry apart from the pregnantwomen.
Also, the field of telemedicine and live monitoring
will be greatly benefited by using edge
technologies, such as 5G. Therefore, the benefits
of the application are numerous:
1) It is a cheap and reliable system. A reliable cardiographer
can cost about 1500 euros while the
proposed project provides efficient health monitoring
at a production cost of 30 euros only.
2) Easy development and implementation.
IEEE Consumer Electronics Magazine
IEEE Consumer Electronics Magazine - July/August 2022

Table of Contents for the Digital Edition of IEEE Consumer Electronics Magazine - July/August 2022
