Calibration Basics
Calibration Basics
TECHNOLOGY MANAGEMENT
Calibration
Basics
Calibration
compares
an
instrument’s
output
against
an
accurate
standard.
Once
an
error
is
known,
any
deviations
can
be
corrected.
For
example,
a
thermometer
could
be
calibrated
and
adjusted
to
show
the
true
temperature
in
Celsius.
For
medical
devices,
any
output
that
will
reach
the
patient
(fluid
flow,
pressure,
electrical
current,
etc)
or
measurement
from
the
patient
(temperature,
heart
rate,
blood
pressure,
etc)
needs
to
be
calibrated
periodically.
Calibration
is
important
for
patient
safety
and
accurate
diagnosis
and
treatment.
When
to
perform
calibration
• with
a
new
instrument
• for
preventative
maintenance
• if
an
instrument
has
had
a
shock
or
vibration
which
may
have
put
it
out
of
calibration
• when
there
is
a
change
in
weather
• whenever
observations
appear
questionable
Procedure
1. Decide
on
what
parameter
to
calibrate.
This
parameter
is
the
thing
that
the
machine
is
measuring.
For
example,
an
ECG
measures
a
physiological
parameter,
heart
rate,
while
a
ventilator
measures
a
physical
parameter,
pressure.
In
some
cases,
a
machine
may
measure
more
than
one
parameter.
For
example,
an
infant
incubator
may
be
designed
to
control
the
temperature,
O2
concentration,
and
humidity
inside
the
chamber,
and
it
may
be
necessary
to
calibrate
all
of
these
parameters
independently.
Identify
a
standard
you
can
use
to
compare
the
parameter.
A
perfect
standard
can
reliably
and
correctly
measure
the
parameter.
,
We
compare
the
device
being
calibrated
to
the
standard.
The
appropriate
standard
depends
on
what
you
are
measuring.
A
perfect
standard
may
not
be
reliable
in
all
situations,
but
there
are
different
techniques
that
can
be
used
in
these
cases
to
allow
us
to
be
reasonably
confident
that
the
machine
in
question
is
properly
calibrated.
Here
are
some
general
types
of
standards
to
consider:
• Use
an
equipment-‐specific
test
device.
For
example,
electrosurgical
unit
tester
for
an
electrosurgical
unit.
This
method
is
often
the
most
accurate.
However,
test
equipment
may
be
expensive.
• Use
physics
principle
such
as
water
density
or
pressure.
For
example,
liquid
water
has
a
known
density
of
1000
kg/m3.
Use
a
known
volume
of
water
to
calibrate
a
scale.
This
method
works
well.
However,
not
all
equipment
types
have
a
suitable
standard.
• Use
a
known
working
machine
and
a
test
patient.
For
example,
find
a
glucometer
that
the
clinical
staff
is
satisfied
with.
Measure
a
test
patient’s
blood
glucose
level
with
this
working
glucometer.
Then
immediately
measure
the
same
patient’s
glucose
level
with
the
glucometer
under
inspection.
This
method
is
often
easy.
However,
this
method
requires
a
working
machine
for
comparison.
Also,
this
method
assumes
the
“working”
machine
as
the
standard.
If
the
working
machine
is
inaccurate,
this
method
may
be
very
inaccurate.
• Test
relative
behavior.
For
example,
if
you
hold
your
breath,
the
pulse
oximeter
oxygen
saturation
reading
should
go
down.
This
method
does
not
provide
accurate
results,
just
a
low-‐level
check
that
the
machine
is
functioning
Below
are
examples
of
potential
standards:
Standard
Medical
Device,
Equipment-‐ Physics
or
Working
Test
Relative
(Output
to
be
specific
Test
Physiology
Machine
&
Test
Behavior
calibrated)
Device
Principles
Patient
Aneroid
Water
column
Known
working
See
if
Sphygmomanome or
mercury
machine
on
manometer
same
patient.
holds
pressure
ter
Sphygmomano
(Pressure)
meter.
Known
working
Breathe
fast
or
Respiratory
rate
Count
breaths
machine
on
slow.
monitor
over
a
specific
same
patient.
(Breathing
Rate)
time.
Hold
your
Apnea
Monitor
breath
(Alarm)
Shows
ECG
ECG
Patient
reading
(Electrical
Activity)
Simulator
Liquid
volume
ment
Infusion
Pump
Treat
delivered
over
(Flow
Rate)
time
Freshly
killed
Defibrillator
Defibrillator
pig
(animal
will
(Current)
Tester
jump)
Electrosurgery
Electrosurgery
Orange,
bar
of
Unit
Unit
Tester
soap,
piece
of
(Current)
chicken
Infant
Incubator
(Chamber
Thermometer
Temperature)
Infant
Incubator
Wet
Bulb
Test
(Humidity)
Infant
Incubator
Candle
Test
for
(Oxygen
O2
sensor
O2
Concentration)
Concentrator
2. Take
measurements
from
the
machine
and
from
the
standard.
Record
your
results.
(Use
clinically
relevant
points)
Table
1:
Heart
Rate
Monitor
Calibration
Table
1
shows
a
calibration
table
for
a
heart
rate
monitor.
Standard,
in
Measured,
The
standard
data
are
from
manual
pulse
readings
bpm
in
bpm
(number
of
pulses
counted
over
1
minute).
The
measured
72
73
data
are
the
heart
rate
monitor
readings.
90
93
100
98
3. Calculate
the
Room
Mean
Square
(RMS)
Error.
The
RMS
98
95
error
is
the
error
of
the
dataset.
This
error
represents
how
80
82
different
the
measured
data
is
from
the
output.
81
78
In
order
to
calculate
the
RMS
error,
you
first
need
to
calculate
the
residuals.
Residuals
are
the
difference
between
the
measured
values and
the
standard
values .
Residuals
can
be
positive
or
negative.
Positive
residuals
signal
that
the
machine
overestimated
the
value.
Negative
residuals
mean
the
machine
underestimated
the
value.
Next,
you
need
to
calculate
the
residuals
squared.
The
results
are
show
in
table
2.
Table
2:
Heart
Rate
Calibration
with
residuals
2
Measured
Standard
Residuals
Residuals
73
72
1
1
93
90
3
9
98
100
-‐2
4
95
98
-‐3
9
82
80
2
4
78
81
-‐3
9
Total
36
Next,
average
the
squares
of
the
residuals.
The
square
root
of
this
average
is
the
RMS
error.
4. Keep
a
copy
of
your
data,
calculations
and
RMS
records
for
future
reference.
5. Determine
the
acceptable
error
range.
• If
possible,
locate
the
service
manual.
The
manual
should
include
a
range
of
acceptable
accuracy
for
the
output.
This
range
may
be
listed
under
“technical
specification”
or
“output
range.”
• If
the
manual
is
not
available,
discuss
this
error
with
the
clinical
staff.
In
the
heart
rate
example,
ask
the
clinical
staff
if
an
error
of
2.5
beats
per
minute
is
acceptable.
6. If
your
calculated
RMS
error
is
outside
of
the
acceptable
range,
the
machine
is
broken.
Diagnose
and
repair
the
machine.
The
machine
should
not
be
placed
back
into
service
on
patients
until
it
is
repaired
and
has
passed
a
calibration
test.
If
your
calculated
RMS
error
is
within
the
acceptable
range,
you
are
done!