Efficia DFM100 SMART Analysis

AED Algorithm
Application Note

Introduction
Efficia DFM100 automated external defibrillation accurately identifies features of morphology associated
(AED) function is designed to be used by a wide range of with shockable rhythms and successfully removes
clinicians on an unresponsive, pulseless, non-breathing artifacts. It also describes which rhythms are considered
patient of any age, including an infant or a child. This shockable and discusses the algorithm’s effectiveness.
application note explains how the AED algorithm
Efficia DFM100 SMART Analysis AED Algorithm

What is an Algorithm?
An algorithm is a sophisticated, mathematical process Algorithm performance is evaluated on two
of interpreting information. In AED mode, the Efficia criteria—sensitivity and specificity. Sensitivity refers to
DFM100 defibrillator/monitor uses an algorithm to the device’s ability to detect life-threatening ventricular
interpret electrical signals received from the patient’s arrhythmias. Specificity refers to the device’s ability to
heart, via multifunction defibrillation electrode pads. detect normal rhythms or arrhythmias that should not be
The algorithm determines if the patient has a shocked.
life-threatening arrhythmia and makes a shock / The Efficia DFM100 defibrillator/monitor uses an
no-shock decision. algorithm called “SMART Analysis.”
An algorithm is a crucial factor in the safety and SMART Analysis is a Philips Healthcare’s proprietary
performance of an AED. The algorithm must accurately analysis algorithm that analyzes a patient's ECG and
assess the cardiac rhythm of a patient and make an determines whether a shock should be delivered.
appropriate therapy recommendation.

Efficia DFM100 AED Algorithm Components

The SMART Analysis algorithm consists of three parts: Artifact Detection
 pad contact quality
Whenever any electrical signal (such as an ECG) is
 artifact detection
measured, there is always some electrical noise in the
 arrhythmia detection. environment that can interfere with an accurate
These three parts work together to read an ECG and measurement. Artifact detection is important in an ECG
evaluate available information to determine if a shock is analysis because it allows filtering out or compensation
appropriate. for this electrical noise. Any artifact that is undetected
can lead to incorrect decisions by the algorithm and can
cause incorrect or delayed treatment of the patient.
Pad Contact Quality
Among the artifact causes may be CPR, agonal
The analysis system continuously monitors the patient breathing, transportation, patient handling, and the
impedance to ensure that it remains within the presence of a pacemaker in the patient. The SMART
appropriate range. If the measured impedance is too Analysis’ action depends on how the artifact looks in
high, it may indicate that the pads are not properly relation to the ECG signal.
applied or that there may be a problem with the skin
To detect artifact, the Efficia DFM100 measures the
contact. Unless this is corrected, the SMART Analysis
transthoracic impedance, common mode current, and
will not be able to read the ECG effectively to determine
electrical potentials sensed by the pads and compares
whether a shock is advised. Poor pad connection can also
these values to the ECG signal. If these values correlate,
cause a problem with the delivery of current to the
then an artifact is detected, and appropriate voice prompt
patient. If the patient impedance is too high, then the
and display messages signal you to take appropriate
Efficia DFM100 issues voice prompts directing your
action. Otherwise analysis proceeds, and the Efficia
attention to the pads, announcing that pads contact is
DFM100 makes a shock / no-shock decision.
poor and instructing you to apply pads or to press the
pads firmly to correct the situation. If the amplitude of the underlying ECG signal is smaller
than an artifact signal, then the Efficia DFM100 may be
unable to accurately analyze the underlying ECG. Then

2
 Application Note

it prompts you not to touch the patient or to stop all Pacemaker Artifacts
motion, and informs you that the analysis has been
interrupted. If the patient has an internal pacemaker, then the Efficia
DFM100 AED algorithm filters attempt to remove the
If the amplitude of the ECG signal is sufficiently high pacemaker artifact and, if appropriate, advises to shock
relative to the artifact signal, or if the artifact does not the patient. The ECG shown on the display and stored in
correlate with the ECG signal, then the artifact may not the device memory still have the pacemaker spikes
interfere with the normal operation of the AED. In these represented, but the ECG used by the algorithm has the
cases, the Efficia DFM100 continues to make shock / spikes removed.
no-shock decisions and prompts you to press the flashing
Shock button if appropriate. The two segments in Figure 2 represent the Ventricular
Fibrillation (VF) ECG before and after the pacemaker
CPR Artifacts artifact is filtered out.
CPR rates significantly above 100 compressions per Figure 2 Pacemaker Artifact Removal
minute can cause incorrect or delayed analysis by the
Efficia DFM100. CPR performed with chest mV Before filtering:
compressions of rates over 135/minute can sometimes
mimic a shockable rhythm. If the Efficia DFM100
detects high CPR rates during ECG analysis, the AED
algorithm interrupts the rescuer doing CPR and instructs
to not touch the patient. Perform CPR at the rates sec
recommended by resuscitation guidelines to avoid
unnecessary interruptions of patient treatment.
mV After filtering:
Figure 1 shows an example of rapid CPR done in such a
way that it was not detected by the SMART Analysis.
The second segment shows the underlying asystole
present when CPR is stopped. The Efficia DFM100
AED continually monitors the ECG and looks for sec
changes in the rhythm, therefore the unit automatically
disarmed when CPR was discontinued, and no shock was
delivered to the patient. Even with a sophisticated artifact detection system, not
all artifact can be detected during the use of the AED,
Figure 1 CPR Artifact Interference therefore it is important to listen to the voice prompts
and observe the display messages given by the Efficia
DFM100 and to not touch the patient while the
algorithm is analyzing the ECG.
CAUTION: Due to differences between pacemaker
therapy designs, artifact removal cannot be guaranteed.
CPR artifact: underlying rhythm asystole
The AED algorithm is not designed to handle erratic
spiking problems caused by a properly or improperly
functioning pacemaker. In patients with cardiac
pacemakers, the Efficia DFM100 may have reduced
sensitivity and not detect all shockable rhythms.
Asystole is not considered a shockable rhythm.
Delivering a shock to a patient in asystole would not
return the heart to a normal rhythm and may actually
Post-CPR: underlying rhythm asystole prevent more appropriate therapies from being
successful.

3
Efficia DFM100 SMART Analysis AED Algorithm

Arrhythmia Detection Rate

The device’s ability to accurately assess the patient’s Rate measures how many times the heart beats per minute
cardiac rhythm is crucial for its safety and performance in (bpm). An adult heart beats approximately 60-100 bpm,
the AED mode. The Efficia DFM100 evaluates the but some normal rhythms can be very fast. Therefore, it is
cardiac rhythm by sensing electrical signals from the important to have additional indicators in the analysis
system of an AED. Rate is used along with the other
patient’s heart via multifunction pads and using a
parameters to help determine whether the rhythm is
software algorithm to interpret these signals and make a
shockable. The higher the rate, the more likely a rhythm is
shock / no-shock decision.
shockable. The lowest rate to be shocked is 135 bpm, and
SMART Analysis was developed and tested to ensure that this applies to those rhythms that are most disorganized,
its sensitivity (ability to detect shockable rhythms) and such as VF. The more organized a rhythm is, the higher the
the specificity (ability to detect non-shockable rhythms) rate must be in order to be shockable. However, rhythms
exceeds the AHA and AAMI DF80 guidelines. with narrow QRS complexes, such as Supraventricular
Tachycardia with Narrow QRS (SVT) will not be shocked,
To determine if a patient’s rhythm is shockable, the
regardless of the heart rate.
SMART Analysis algorithm evaluates these four
parameters of the ECG in 4.5-second segments: Figure 4 Rate in Arrhythmia Detection
 rate mV
 conduction (shape of the QRS complex)
 stability of the rhythm (regularity of the waveform
pattern) Better

Rate
 amplitude.
Figure 3 QRS Complex
sec
R-wave mV

Worse

Rate
Q S sec

QRS Complex

4
 Application Note

Conduction Figure 6 Stability in Arrhythmia Detection

Conduction is determined by examining the R-wave of the mV
QRS complex. Conduction is related to the propagation of
electrical impulses through the ventricles. In a healthy heart,
the ventricles contract in unison, which is reflected in the
Better
ECG by narrow QRS complexes with sharp transitions.
Non-perfusing rhythms are characterized by wide
complexes with smooth transitions. Therefore, a rhythm
with wide complexes and smooth transitions is more likely
sec
to be shocked. mV

Figure 5 Conduction in Arrhythmia Detection

mV Worse

Better
sec

sec
mV
Amplitude
Worse Amplitude is a measure of magnitude of the heart’s
electrical activity. A heart in asystole has a low-amplitude
ECG. Amplitude is dependent on the patient and pads
placement. Amplitude is the least important of the four
sec indicators.
SMART Analysis simultaneously measures the first three
indicators over 4.5 second segments of ECG, and then
Stability classifies each segment of ECG as shockable or not.
Stability refers to the regularity of the morphologies of the
Amplitude is used as a gating check to determine if a
various waves of the ECG complexes. The consistency of segment is considered shockable; i.e. the 4.5 second
both the shape of the complex and the period between segment of ECG must have at least a 100 mV
complexes also indicates whether a rhythm is perfusing. peak-to-peak median amplitude in order for a segment to
With a perfusing rhythm, the sequential complexes tend to be considered VF.
be similar in shape. A heart in ventricular fibrillation will Figure 7 Amplitude Measurement
have chaotic, unstable complexes.

4.5 second segment

5
Efficia DFM100 SMART Analysis AED Algorithm

Analysis Confirmation If the device detects one or, depending on artifact

presence, more consecutive segments that are
Before the Efficia DFM100 charges, the SMART non-shockable, it gives voice and visual prompt that no
Analysis algorithm must identify one or, depending on shock is advised and directs you to attend to the patient.
artifact presence, more ECG segments as shockable; then It also transitions into the No-Shock-Advised monitoring
the device must continue to see more shockable segments if so configured. The device continues to monitor the
in order to deliver a shock. The Efficia DFM100 ECG, but it gives minimal prompts until it identifies
continues to monitor the ECG even after a shock / another shockable segment; then it transitions back to
no-shock decision has been made and the unit has the AED mode, directs you to not touch the patient, and
charged; this means that the SMART Analysis would makes a decision to shock the patient if appropriate.
react to a change in the heart rhythm and disarm if the
rhythm becomes non-shockable after the device is
charged.

Efficia DFM100 AED Algorithm in Action

Analysis Examples Figure 9 Supraventricular Tachycardia

This section reviews four different ECG examples. Each
ECG is graphed based on its score for rate, conduction,
and stability to determine if the SMART Analysis
algorithm would or would not advise a shock.
In the Figures 8–12 below, the shock criteria surface is
drawn in grey. According to the algorithm, any dot above
the surface represents a shockable rhythm, and any dot SVT
Rate = 218 bpm
bpm
below is a non-shockable rhythm. Green dots indicate a
non-shockable rhythm for the Normal Sinus Rhythm
(NSR) and Supraventricular Tachycardia (SVT), and red
dots indicate a “shock advised” condition for the
polymorphic Ventricular Tachycardia and Fibrillation
(VT and VF) rhythms. No shock advised: excellent stability and conduction, high rate
Figure 8 Normal Sinus Rhythm
Figure 10 Polymorphic Ventricular Tachycardia

NSR
Polymorphic
Polymorphic
Rate = 76 bpm
bpm VT
Rate = 240 bpm
Rate bpm

No shock advised: excellent stability, conduction, and rate Shock advised: poor stability, very poor conduction, high rate

6
 Application Note

Figure 11 Ventricular Fibrillation SMART Analysis Approach

In 1997, the American Heart Association published a
Scientific Statement that recommends a strategy for
evaluating the accuracy of the arrhythmia analysis
algorithms incorporated in AEDs.1 Following the process
described in this recommendation, over 3000 ECG
segments were collected into a database. This database
included both shockable and non-shockable rhythms,
VF and was used to design and validate the SMART Analysis
Rate = 240 bpm
bpm
system used in the Efficia DFM100.
Each segment was reviewed by a group of three
cardiologists to determine whether that segment should
be considered shockable or non-shockable. If there was
disagreement on a particular segment, the cardiologists
Shock advised: very poor stability and conduction, high rate were asked to discuss the segment and come to a
consensus on how to classify the segment. By far, the
most disagreements resulted from ventricular tachycardia
Sensitivity and Specificity (VT) segments and were related to whether it was
Definitions appropriate for an AED to shock this type of VT.
Figure 12 shows the diagram of evaluated ECGs shock /
The following four parameters are used to assess the
no-shock decisions against the SMART Analysis
algorithm’s performance:
parameters. In this diagram, each of the 3000 segments is
A true positive (A) is a shockable rhythm associated with plotted according to their stability, conduction, and rate,
cardiac arrest that is classified as a shockable as in the “Analysis Examples” on page 6. If the dot is red,
rhythm/condition. the cardiologists considered it a shockable rhythm; if it is
A false positive (B) is an organized or perfusing green, it was considered a non-shockable rhythm.
non-shockable rhythm that has been incorrectly classified Figure 12 Evaluated ECGs Plot
as a shockable rhythm/condition.
A false negative (C) is a shockable rhythm associated with
cardiac arrest that has been incorrectly classified as a
non-shockable rhythm/condition.
A true negative (D) is any non-shockable rhythm that is
classified as a non-shockable rhythm/condition.
The sensitivity of the device is the number of true
positive shockable rhythms, expressed as a percentage of
the total number of shockable rhythms:
A
sensitivity = --------------  100 %
A+C

The specificity is the number of organized or perfusing

rhythms that have been correctly classified as
non-shockable rhythms/conditions by the algorithm, and The SMART Analysis algorithm is designed to make
is expressed as a percentage of the total number of aggressive decisions about shocking VF rhythms and
non-shockable rhythms/conditions: conservative decisions about shocking VT rhythms that
may have an associated pulse. Figure 12 shows only red
D
specificity = --------------  100% dots above the shock-criteria surface, indicating that a
B+D shock will be advised only if it is needed.

7
Efficia DFM100 SMART Analysis AED Algorithm

Figure 12 shows some red dots that fall below the shock SMART Analysis is designed to be conservative for stable
criteria surface. In these instances, the algorithm does not monomorphic tachycardias. The rate threshold for a
advise a shock, but the cardiologists concluded that a shockable tachycardia varies from a minimum of about
shock should be advised. These rhythms are typically 160 bpm for rhythms with very slow ventricular-like
intermediate VT that may have some perfusion conduction to a maximum threshold of 600 bpm for
associated with them. If they are non-perfusing rhythms, rhythms with healthy normal conduction. Thus,
they quickly degrade to the point that they migrate above rhythms with normal conduction are not shocked
the shock-criteria surface and SMART Analysis advises a regardless of the rate.
shock. If the shock criteria were changed so that the The AHA has issued a Scientific Statement that identifies
surface was shifted to try to catch more of the shockable SVT as a non-shockable rhythm, and requires a
rhythms below the surface, the algorithm would also minimum algorithm specificity of 95% for this rhythm.1
advise a shock for a greater number of non-shockable This high-specificity requirement assumes that a
rhythms. The SMART Analysis algorithm is designed to high-quality assessment of perfusion status has been
be conservative in this respect in order to increase the made, thereby eliminating many SVTs from
specificity of Efficia DFM100 AED. consideration for shock. The Efficia DFM100 AED is
While rate is a key factor, it is not the only factor. The designed to issue a no-shock recommendation for
more normal the conduction and stability of the QRS rhythms of supraventricular origin regardless of their
complexes, the greater the possibility of perfusion, and rate, and has demonstrated 100% specificity when tested
the less likely is the SMART Analysis to recommend a against a database containing 100 samples of SVT with
shock. For example, if an infant patient with a fast rates as high as 255 bpm.
normal sinus rhythm has a heart rate of 250 bpm with For rhythms that have poorer morphological stability,
excellent conduction and stability, the SMART Analysis such as polymorphic VT and VF, the rate threshold varies
would correctly not advise a shock. in a similar manner. As morphological stability degrades,
Shockable Rhythms the rate threshold is progressively reduced, approaching a
minimum rate threshold of about 135 bpm.
The SMART Analysis algorithm is designed to shock
This adaptive design allows the rate threshold to be
these most common rhythms associated with sudden
varied from a minimum level for the most lethal VF
cardiac arrest:
rhythms, providing very high sensitivity, to increasingly
 ventricular fibrillation (VF) higher rate thresholds as the stability or conduction
 ventricular flutter characteristics approach normal, providing very high
 polymorphic ventricular tachycardia (VT). specificity. Borderline rhythms, such as monomorphic
In addition, it is designed to avoid shocking rhythms that tachycardias are treated conservatively, with the
are commonly accompanied by a pulse or rhythms that expectation that if they are hemodynamically unstable,
would not benefit from an electrical shock. The AHA then the rhythm will soon exhibit shockable
states that rhythms accompanied by a pulse should not characteristics.
be defibrillated because no benefit will follow, and Two samples of monomorphic tachycardia are shown in
deterioration in rhythm may result.1 Figure 13 as examples of borderline rhythms that do not
The shock / no-shock decision made by Efficia DFM100 require shocks. Both of these rhythms are of
AED may be different from a decision a clinician may supraventricular origin, with one known to be
make. AED is designed to be used by rescuers trained in accompanied by a pulse. The SMART Analysis algorithm
Basic Life Support, and the Manual mode is designed to gives a no-shock recommendation for both of these
be used by qualified medical personnel trained in rhythms.
Advanced Cardiac Life Support (ACLS). The main AHA guidelines recommend synchronized cardioversion
difference is that the algorithm in an AED tries to for hemodynamically unstable monomorphic tachycardia,
differentiate between VT that has a pulse and VT that but allows unsynchronized cardioversion shocks if the
has not, therefore the Efficia DFM100 AED is more synchronized one is not available. If SMART Analysis
conservative in shocking intermediate rhythms such as does not recommend a shock for monomorphic
fine VF and VT that do not meet all criteria for inclusion tachycardia, then consider using the Manual mode if a
in the shockable VT rhythm category. shock is desired and an ACLS clinician is present.

8
 Application Note

Figure 13 Monomorphic Tachycardia The samples shown in Figure 14 are examples of

polymorphic VT and flutter. These rhythms represent
ECGs that are not associated with a pulse and are
considered shockable forms of VT.
The Figure 15 example of VF would not be considered a
shockable rhythm because of its low frequency. Besides
Rate = 192 bpm No Shock Advised the possibility of artifact generating this type of rhythm,
there are studies that indicate that a fine VF such as this
would benefit from a minute or two of CPR before a
shock is attempted. CPR tends to oxygenate the
myocardium and increase the electrical activity of the
Rate = 144 bpm No Shock Advised heart, making it more susceptible to defibrillation.
Figure 15 Low-Frequency VF
Figure 14 Polymorphic VT and Flutter

Fine VF No Shock Advised

Rate = 240 bpm Shock Advised

Rate = 288 bpm Shock Advised

9
Efficia DFM100 SMART Analysis AED Algorithm

SMART Analysis Algorithm Validation

Algorithm performance is evaluated by its sensitivity and For example, in preparation for introducing the pediatric
specificity. The SMART Analysis algorithm provides an defibrillation electrodes, a database was assembled that
exceptional level of sensitivity and specificity, and its included 696 pediatric arrhythmias.
validation results exceed AHA recommendations and
international standards for adult defibrillation 1,2,3,
including the requirements for the one-sided lower Specific Concerns for Advanced
confidence limit (LCL). The studies cited above and Users of Efficia DFM100 AED
Table 1 data are the result of testing the extremely
challenging rhythm collection from the Philips Simulator Issues with SMART Analysis
Healthcare ECG rhythm databases that deliberately test
the limits of AEDs. ECG simulators are designed to train people to recognize
different heart rhythms based on a visual analysis of the
Table 1 AED Validation Results a
data and cannot be used to verify defibrillator analysis
Rhythm Class AAMI DF80 Observed algorithms. Simulators contain simulated waveforms and
requirement1 Performance typically have only one example of each type of rhythm.
In addition, these devices only store a few seconds of
Shockable Sensitivity > 90% 97% (n=300)
Rhythm: VF ECG signal that is repeated over and over again. This
Shockable Sensitivity > 75% 78% (n=100)
apparent stability can cause the Efficia DFM100 to not
Rhythm: VT advise a shock even though the simulator-generated
Non-shockable Specificity > 99% 100% (n=300) rhythm may appear shockable to the user.
Rhythm: NSR The conduction and stability characteristics of a
NSR: Asystole Specificity > 95% 100% (n=100) simulated VT waveform frequently appear to be high and
NSR: All other Specificity > 95% includes: repeatable. Also, the shape of the simulator's QRS
non-shockable SVT (R>100), SVD (R100), 100% (n=450) complexes may be fairly sharp, indicating possible
rhythms VEB, Idioventricular, and
Bradycardia perfusion and causing the SMART Analysis algorithm to
determine that the rhythm is not shockable. A
a. Additional details are available upon request. A non-disclosure monomorphic VT must have a relatively high rate and
agreement may be required. poor conduction to be considered shockable by the
SMART Analysis. Polymorphic VTs are considered
In the original out-of-hospital study involving 100 shockable at lower rates because there is variation in the
patients,4 the SMART Analysis system correctly shape of the QRS complexes.
identified all patients in VF (100% sensitivity, no false
negatives) and correctly identified and did not shock all Most simulated VF signals are interpreted as shockable
patients in non-VF rhythms (100% specificity, no false by Efficia DFM100. However, most VT rhythms found
positives). Borderline rhythms are reviewed periodically in simulators are monomorphic VT and are not
to determine if the algorithm should be fine-tuned for considered shockable because the shape and regularity of
new products. the waveform indicate that there may be a pulse
associated with it.

10
 Application Note

Efficia DFM100 AED Features References

1 American Heart Association (AHA) AED Task Force,
Therapy Delivery Speed: Maximum time from
Subcommittee on AED Safety & Efficacy. Automatic
initiation of rhythm analysis to readiness for discharge of
External Defibrillators for Public Access Use:
AED with a fully charged battery is no more than 20 Recommendations for Specifying and Reporting
seconds (up to a maximum of 17 s for analysis and up to Arrhythmia Analysis Algorithm Performance,
3 s for charge). Incorporation of New Waveforms, and Enhancing Safety.
Circulation, 1997; 95:1677-1682.
Configurable Resuscitation Protocols: You have the
2 ANSI/AAMI DF80:2003/(R)2010, Association for the
flexibility to configure your Efficia DFM100 to match
Advancement of Medical Instrumentation, 2010
your institution’s resuscitation protocols.
3 IEC 60601-2-4 International Standard, 2010
 Customize the device for the number of shocks (1-4) 4 Jeanne Poole, M.D., et al. Low-energy
in a series. impedance-compensating biphasic waveforms terminate
ventricular fibrillation at high rates in victims of
 Select the energy setting within a given shock series. out-of-hospital cardiac arrest, Journal of Cardiovascular
Default: 150 J for adult; Electrophysiology, December 1997.
50 J for infant/child (non-configurable).
 Set the CPR pause interval 1-3 minutes.

11
Efficia DFM100 SMART Analysis AED Algorithm

© 2014
Koninklijke Philips N.V.
Philips Healthcare is part All rights are reserved.
of Royal Philips Reproduction in whole or in part
is prohibited without the prior
written consent of the copyright
On the web
holder.
[Link]
Philips Healthcare reserves the right
By e-mail to make changes in specifications or
healthcare@[Link] to discontinue any product at any
time without notice or obligation
By postal service and will not be liable for any
Philips Healthcare consequences resulting from the use
3000 Minuteman Road of this publication.
Andover, MA 01810-1085
Published March 2014, Edition 1
Europe, Asia, and Africa
Tel: +49 7031 463 2254 453564491231

Latin America *453564491231*

Tel: +55 11 2125 0744 *1.0*

12