#SignalQuality.tex#

\Medicine{Uncertainty due to signal quality}{Medical devices use a
variety of bio-sensors that measure record and analyze different
biometric signals. These signals vary in velocity, ranging from
low-velocity vital signs such as heart rate, oxygen saturation,
temperature and blood pressure, through high-velocity waveforms such
as ECG and EEG and medical imaging such as CT, X-ray, MRI and scanning
microscopes. Some signals, such as blood pressure or heart rate, can
be directly used in diagnosis.  Some other signals have to be
interpreted before they can be used in diagnosis. We use the acronym
HS (High Speed) to refer to those signals that require interpretation.

Interpreting HS is a significant fraction of the work of most
doctors. In addition, There are medical specialties such as Radiology
and Pathology that are devoted to interpreting HS. These so-called
``Pattern Doctors'' are predicted to be the early adopters of
AI~\cite{topol2019deep} or IA.  HS provides critical detailed
information about the patient's health. However, the richness of the
signal can make it susceptible to nuisance variability from noise,
limited resolution, operator error, etc. The number of nuisance
variables is very large and confounding.  In the next paragraph we
give an example of one nuisance variable: the placement of ECG leads
on the patient's body.
  
For ECG signals to be correctly and consistently interpreted, it is
important that the leads are placed correctly on the patient's
body. Several standards for placement have been published, for example, the standard 12 leads ECG system
\cite{goldberger2017clinical}, the EASI system \cite{Dower1988}, and the Frank lead system \cite{frank1956accurate}. 
It is not always possible to achieve a consistent and precise sensor
placement for biomedical signal collection due to various reasons, for example, the torso variation caused by gender, age and living styles.
This uncertainty might be tolerable for some clinical applications;
for example, an imprecise ECG sensor placement might not impact the
identification of some types of arrhythmia from the ECG signal, like
atrial fibrillation. However, this uncertainty might cause troubles in
identifying other types of arrhythmia, for example, premature atrial contraction.

}