Integrated Protocol for Objective Pain Assessment

Maria Ghita

1,2

, Mihaela Ghita

1,2

, Clara Ionescu

1,2,3

and Dana Copot

1,2

Ghent University, Research Group of Dynamical Systems and Control, Technologiepark 914, Ghent 9052, Belgium

Core Lab EEDT, Flanders Make, Belgium

Technical University of Cluj Napoca, Department of Automatic Control, Memorandumului Street no 28, Cluj, Romania

Keywords: Non-Invasive Pain Measurement, Medical Devices, Residual Pain, Memory Effect, Nociceptor Stimulation,

Mechanical Pain, Measurement Protocol, Skin Impedance, Bioimpedance.

Abstract: In the absence of any standardized objective aid for measuring pain levels in human body, a manifold of

subjective tools have been developed to monitor chronic pain patients and intra-/post-operative analgesic

drug management. However, due to the subjective nature of the evaluation methods and tools, pain remains

a challenging phenomenon to be characterised for objective assessment and monitoring. In this paper we

briefly describe a protocol and methodology for non-invasive evaluation of pain as result of nociceptor

stimulation via skin impedance measurements. Both time-frequency domain analysis is performed,

providing interesting observations.

1 INTRODUCTION

Clinical literature, as well as biomedical engineering

literature, have identified the need of a non-invasive

medical device to measure the pain level in an

objective manner for patients. Pain is very important

phenomena in medicine and biology that includes

physiological, sensory, affective, cognitive,

behavioural and sociocultural aspects (Copot, 2018).

The subjective perception of pain is hard to quantify

and the most commonly used measures of pain

intensity are subjective methods, such as: numerical

rating scale (NRS), visual analogue scale (VAS) and

verbal rating scale (VRS) (Shieh, et al., 2018). All

tools currently available have a number of

limitations: i) they are not based on a mathematical;

ii) do not deliver an objective evaluation index, iii)

require the intervention by medical staff, iv) not

responsive to postoperative efforts of the patient, v)

not suitable for time-frequency domain dynamic

analysis, vi) do not provide continuous monitoring

and vii) they are often not reliable in all

measurement conditions (Shieh, et al., 2018).

Despite all those limitations, the perception of pain

is assessed in conscious awake patients from their

personal feedback information. The NRS is the most

commonly used pain scale, and patients are asked to

rate their pain level on a 0–10 scale.

Recommendations on pain management

strategies are based on the index provided by those

ratings and/or on caregiver’s opinion when patients

are not conscious or awake (e.g. infants, children,

anesthetized or delirious patients). Evaluating the

postoperative pain in intensive care units is a

necessary part of the overall treatment plan (Czaplik,

et al., 2012). According with recent studies, pain is

identified by the American Pain Society (APS) as

the fifth vital indicator in diseases and diagnosis

chart along with temperature, blood pressure, pulse

and respiration rates (Shieh, et al., 2018; Yang, et

al., 2017; Merboth and Barnason, 2000; McCaffery

and Pasero, 1997).

Ideally, a pain detection and evaluation device

should be non-invasive, applicable on any individual

and monitor changes in real time and in correlation

with the administered medication. To meet the

requirements of an objective pain assessment, the

concept of a continuous pain measurement by means

of non-invasive skin impedance measurements

enables clinicians to provide personalized and

effective pain management.

The scope of this paper, is to present and discuss

such a system. The ANSPEC-PRO prototype has

been validated in awake participants with self-

induced nociceptor excitation (Copot and Ionescu, in

print). Currently, it undergoes a clinical trial on post-

operatory awake patients in ICU at Ghent University

Hospital, Belgium (B670201734377).

Apart from the studies related to correlations to

NRS and other features enabled by such a device, it

Ghita, M., Ghita, M., Ionescu, C. and Copot, D.

Integrated Protocol for Objective Pain Assessment.

DOI: 10.5220/0006895300870092

In Proceedings of the 5th International Conference on Physiological Computing Systems (PhyCS 2018), pages 87-92

ISBN: 978-989-758-329-2

is interesting to investigate memory of pain as part

of extracellular tissue dynamics and latency in

perception process. This gives an in-depth

information that could explain changes in thresholds

for pain management decision makers, such as in

chronic pain patients.

The paper is organized as follows. The available

measurement tools are briefly described in the

second section, along with the prototype developed

at Ghent University in our lab. Section 3 describes

the protocol and methodology proposed to be

followed in this study. The results are given in the

fourth section along with a short discussion of their

usefulness and limitations. A conclusion section

summarizes the main outcome of this work and

points to further steps.

2 AVAILABLE DEVICES

2.1 Commercial Devices

Objective tools developed for pain measurement

during consciousness and unconsciousness of the

patients in intensive care units (ICU) are one of the

main subject for further research. None of the

commercial devices can accurately measure pain

levels, despite the efforts to demonstrate the validity

and reliability of tested data. The following devices

for pain assessment have been developed in the last

fifteen years.

Med-Storm Pain Monitor is a medical device

intended to determine a patients’ sensitivity to pain

(http://www.med-storm.com/). The system uses real-

time data measurements to measure pain/nociceptive

stimuli and awakening during anaesthesia, intensive

care, in adults, children and infants. Hence, the

exosomatic electrodermal activity is measured in

terms of conductance. After different studies, there

was developed a standard index (not-standardized

method): The Skin Conductance Algesimeter index,

which is represented by the skin conductance

responses (SCR) per second. The company has

already delivered and sold the equipment for clinical

research projects or for diagnostic purposes, but is

not used by critical care clinicians because Med-

Storm Pain Monitor is not considered proper to be a

substitute for the medical stuff judgement and it

cannot be liable for the results obtained using it

(www.med-storm.com). This device is not suitable

for awake patients, nor for chronic pain patients.

AlgiScan monitors depth of analgesia in sedated

and unconsciousness patients using pupillary reflex

dilation (PRD). This method has been studied for the

evaluation of the level of sensibility to nociception

and in the prediction of the haemodynamic reactions

to nociceptive stimuli in volunteers and surgical

patients (www.medica.de). The pilot studies relate

that in anesthetized patients the pupil increase in size

due to an incision/tetanic electrical simulation,

measurements that can be highlighted by AlgiScan

device which indicates a pain pupillary index (PPI).

However, further research is required in order to use

AlgiScan as a standardised “objective “device for

pain measurement. This device is not suitable for

awake patients (discomfort due to blocked eyelid).

MEDASENSE is based on changes in

physiological parameters (heart rate, temperature,

skin conductance level and more) affected by pain

and analgesic medications. The technology

combines a non-invasive, finger-mounted probe for

collecting the physical data with artificial

intelligence algorithms that convert the data into a

Nociception Level Index (NOL). The pain-related

index is between 0 (no pain) and 100 (extreme pain).

This device is not available in Europe.

2.2 ANSPEC-PRO Prototype

ANSPEC – PRO device is a prototype developed

with the scope of continuously monitoring the pain

in patients who are conscious or not, by

measurements of changes in skin impedance

(Juchem and Ionescu, in review). An overview is

given in Figure 1.

Figure 1: ANSPEC-PRO prototype and afferent compo-

nents of the system.

The elements of the ANSPEC – PRO device are

listed here below.

- Disposable standard electrodes – are

interfacing the skin and the device and are

temporarily attached in the palm of the hand. This is

a three-electrode system, two current-carrying

electrodes and one pick-up electrode, which picks up

the voltage without carrying any currents for no

PhyCS 2018 - 5th International Conference on Physiological Computing Systems

polarization. The electrodes are disposable (single

use only).

- Data-acquisition circuit – essentially consists

of a power supply for the electrodes and interfaces

the microprocessor of the device with the signals

acquired by the electrodes. A carefully designed

voltage signal is sent to excite a part of the skin

using a National Instruments (Texas, USA) device

(cRIO9074 with NI9201- and NI9263-slots). The

current induced in the circuit by this voltage is

related to the bio-impedance of the human skin. A

voltage buffer limits the supplied current to +/-

20mA, well below the maximum allowed for in vivo

studies (5A).

- DELL Laptop computer – is used for capture,

save and display measured data in real time; is

interconnected through Ethernet with the data-

acquisition circuit. The laptop is a standard laptop

with the operating system Windows 7 Enterprise 64-

bit and a INTEL® Core ™ i7-6600U CPU@2.80

GHz processor.

- User interface – is developed in LabView.

In short, ANSPEC-PRO device is a non-invasive

method for continuously measurements of changes

in skin impedance caused by an applied stimulus

(pain). The changes in skin impedance reflect

changes in the extracellular fluid matrix composition

which facilitates the electro-chemical channel

communication for pain signalling pathway.

Electrical variability in the electrical carrier

throughout the signalling pathway, originated by

mechanical nociceptor stimulation, affect the

response of the skin related in impedance values.

The device measures the current i(t) coming from

the skin. Also, it acquires the measured signal v(t)

with a 15KHz sampling frequency, f

and sends it to

an analogue output port, using zero-order hold

protocol for digital signal processing (Copot, 2018).

As part of the signal conditioning step, the current is

transformed to a voltage, using a transimpedance

amplifier (TIA), which can be then interpreted by

the algorithms, as in the Figure 2.

Figure 2: Block diagram with the skin impedance Z(s,k) as

black box system.

The bio-electrical-impedance is the Laplace

transform equivalent model Z(s), with s=d/dt the

Laplace operator, of the ratio of the Laplace

transformed voltage V(s) and current I(s) signals.

The impedance can be evaluated using spectral

identification methods, and further parametric model

identification as given in (Ionescu and De Keyser,

2003). The impedance is further evaluated using

moving averaged windows over time, such that it

depicts a both time- and frequency- domain

variability index Z(s,k), with k denoting the

discretized time (Pintelon and Schoukens, 2012).

3 PROTOCOL AND

METHODOLOGY

3.1 Hypothesis

In this paper, we discuss the two major hypothesis

for our study:

1. that a latency exist which implies a minimum

time elapsed for nociception stimulation to fade

under the threshold for pain pathway to be open

– if measurement time between nociception

stimulation time intervals is not adequately

chosen, one gets residual pain/memory effect.

The clinical implication of this is over-dosing.

2. that the impedance of extracellular fluid

changes is not dependent on sensor location.

The clinical relevance of this hypothesis is that

the location of the electrodes may be chosen

freely and does not affect the impedance values.

Both hypotheses are made under the further

assumption that no other device/sensor/monitor is

present on the patient/volunteer at the time of

testing.

3.2 Protocol

The participants were the authors of this paper. The

biometric information of two volunteers are:

Volunteer #1 – weight 70 kg, height 1.80 cm,

age 24 years, female

Volunteer #2 – weight 66 kg, height 1.79 cm,

age 24 years, female.

The volunteers approved with the protocol and

procedures prior to data collection. Both participants

were eligible and reliable for this study, especially

because their biometric data are similar and it is

expected to have virtually no effect on results.

Subjects are clinically healthy, awake and without

prior pain or related medications.

The protocol has been designed for 36 minutes

and was conducted indoors, as follows.

Integrated Protocol for Objective Pain Assessment

Case A: participants were asked to sit on a chair

and act normally without affecting the sensors

attached to the left hand. Data acquisition starts with

a reference range of

- 2 minutes when no pain is applied (NP1).

The activity continues with pain/no pain

alternation:

- 1 minute nociceptor stimulation applied

with a clip on the right hand (P1),

- 1 minute no pain applied (NP2),

- 1 minute nociceptor stimulation applied

with a clip on the left hand – same location

with the sensors (P2),

- 1 minute no pain applied (NP3),

- 1 minute nociceptor stimulation applied

with a clip on the right ear – totally

different location with the sensors (P3),

- 2 minutes no pain applied (NP4).

The total period of time for following the

procedure in Case A is 9 minutes.

Case B: participants were asked to take the same

sitting position as in case A. The measuring session

starts (NP1) and ends (NP2) also with a period of 2

minutes when no pain is applied to have a reference

for the measurements.

Between the reference range of measurements,

the procedure has been realised continuously:

- 1 minute nociceptor stimulation applied

with a clip on the right hand (P1),

- 1 minute nociceptor stimulation applied

with a clip on the left hand – same location

with the sensors (P2),

- 1 minute nociceptor stimulation applied

with a clip on the right ear – totally

different location with the sensors (P3).

The total period of time for following the

procedure in Case B is 7 minutes.

The time interval elapsed between the two cases

for measurement on the same individual was 20

minutes.

In order to investigate the existence of a memory

effect of pain or residual pain, the protocol

procedures enables to observe differences in data

between Case A and Case B (sensors placement is

on the left hand, pain location is maintained: right

hand (P1), left hand (P2), right ear (P3)).

3.3 Analysis Tools

The recorded data were post-processes and analysed

in MATLAB (The MathWorks, Inc. USA) version

R2017b (9.3).

Figure 3: Electrical Scheme Analogy.

The system has 2 inputs: the multisine excitation

signal U(s) and the nociceptor stimulus N(s):





(



)

()

=()



(



)

()



(1)

with V(s) and I(s) the measured signals. Define now

the vectors:





=









, and 



=











(2)

containing the cross-power spectra 



(



)

between

two distinct signals and auto-power-spectra 



(



)

of a signal. It follows that:





(





)

=

(





)





(



)

(3)

If the nociceptor stimulation signal is not

correlated with the multisine excitation signal, then

the impedance can be directly estimated from (3).

Every 60 sec, the impedance is calculated and

plotted against frequency, by means of its real and

imaginary parts. The complex impedance is then

normalized and analysed per interval of pain (P) or

no pain (NP), as the response of the nociceptor

excitations.

The variability within individual is observed

with ANOVA method, using absolute individual

impedance values. Boxplot analysis is the procedure

used for determining whether variation in the

response variable arises within the same individual,

for both protocols.

One way anova has been used to compare among

the group of values. The function ANOVA1 has

been used in Matlab which returns box plots of the

observations in data y, by group. Box plots provide a

visual comparison of the group location parameters.

If y is a vector, then the plot shows one box for each

value of group. If y is a matrix then the plot shows

one box for each column of y. On each box, the

central mark is the median and the edges of the box

are the 25th and 75th percentiles (1st and 3rd

quantiles). The whiskers extend to the most extreme

PhyCS 2018 - 5th International Conference on Physiological Computing Systems

data points that are not considered outliers. The

outliers are plotted individually. The interval

endpoints are the extremes of the notches. The

extremes correspond to:

q2±1.57(q3–q1)

√



(4)

where q2 is the median (50th percentile), q1 and q3

are the 25th and 75th percentiles, respectively, and n

is the number of observations (excludes NaN

values).

Confidence intervals have been calculated at

95\%, and significant differences defined for p-

values<0.05. The function TTest in Matlab has been

used.

4 RESULTS

4.1 Bio-electrical-impedance as

Function of Frequency

The frequency response of the bio-electrical-

impedance for every protocol interval in Case A is

depicted in Figure 4, using experimental data from

volunteer #1.

Figure 4: Individual frequency response of the normalized

impedance, evaluated for “pain”/”no pain applied”

intervals for case A.

It is observed that the bold lines that denote the

second (P2) and third (P3) pain interval responses

overlap the corresponding non-pain intervals: NP2

and NP3. This suggests that NP2 and NP3 indicate

the presence of pain latency (i.e. memory pain).

Hence, even in absence of nociceptor stimulation,

the impedance indicates presence of pain pathways

because of the pain memory effect. Also, since the

nociceptor stimulus is applied in different locations

and still detected with our non-invasive

measurement device, we conclude that the device is

sensitive to any stimulation through the

physiological pathway of pain.

At this point, the first hypothesis of our study is

demonstrated.

In the protocol for Case B, the pain is applied

continuously to different places on the volunteer #1

and the responses are evaluated per interval, as

depicted in Figure 5.

Figure 5: Individual frequency response of the normalized

impedance, evaluated for “pain”/”no pain applied”

intervals for case B.

From the three locations of the nociceptor

stimulation tested, it can be seen that the first two

pain responses (P1 and P2) seem to give the same

result. This relates to the left and right hand,

respectively. The third one (P3) suggests some

differences (the bold line with circle marks) – on the

ear. Further analysis will clarify whether or not the

location on the ear provides biased results due to

electrical activity of other nearby sources (e.g.

brain).

Therefore, the bio-impedance is sensitive to any

nociceptor stimulation location. While the

impedance has different values for each interval, the

amplitude value cannot be correlated to the stimuli

location.

The second hypothesis of our study is also

demonstrated.

4.2 Variability within Individual

From the two protocols analysed above, the

variability within the same individual is described by

means of boxplot in Figure 6. Instead of complex

(real and imaginary parts), we now introduced the

absolute values of the impedance |Z| obtained for

each nociceptor stimulation interval.

For the first pain interval (P1), there are no

statistical significant differences within individual

per protocol (p<0.7). By contrast, the second (P2)

and third (P3) pain interval, significantly differences

are observed (p<0.05). Despite the fact that all

nociceptor stimulation amplitudes are equal, the P2

and P3 data are clearly higher in amplitude for case

B than for case A. This is due to overlapping of

electrochemical ions channel activity in case B,

Real Part ( )

Imaginary Part ( )

Real Part ( )Imaginary Part ( )

Integrated Protocol for Objective Pain Assessment

since in the proposed protocol the no-pain intervals

are not performed.

Figure 6: Absolute values of frequency response complex

impedance in one individual per protocol (Case A and

Case B). Each figure depicts boxplot analysis in all pain

intervals during both protocols.

5 CONCLUSIONS

This work describes the first steps towards a novel

non-invasive pain detection, evaluation and

monitoring in chronic pain patients. Ideally, the

same index would be valid also for analgesic drug

dose management in general anaesthesia. The

protocol and biosignal processing methodology

proposed here lead to results to support the claim

that latency of pain pathway exists (i.e. memory

pain). Additionally, the tests indicated the technical

soundness of the measurements, by accurate

detection of nociceptor stimulation intervals through

skin impedance evaluation. The location of

nociceptor stimulation has no effect to the the

accuracy of the device.

The evaluation of the ANSPEC-PRO prototype

in clinical environment for patients experiencing

post-operative pain is currently ongoing. Major

challenges are expected by evaluating a “pain index”

that can be correlated with patient information, in

order to make ANSPEC -PRO clinically useful.

ACKNOWLEDGEMENTS

The authors appreciate the support given by Martine

Neckebroek for the useful discussions and Jasper

Juchem for the technical assistance.

This work has been financially supported by

Flanders research centre grant nr 1501517N and

G026514N.

REFERENCES

Copot, D., 2018. PhD thesis "Fractional calculus based

methods and models to evaluate diffusion in human

body", Ghent University.

Copot, D., Ionescu, C., 2018. Prototype and Methodology

for Non-Invasive Nociception Stimulation and Related

Pain Assessment in Awake Individuals accepted to

IEEE Transactions on Biomedical Engineering

Czaplik, M., Hubner, C., Kony, M., Kaliciak, J., Kezze, F.,

Leonhardt, S., Rossaint, R., 2012. Acute pain therapy

in postanesthesia care unit directed by skin

conductance: a randomized controlled trial. PLoS

ONE, July, Volume 7, Issue 7, e41758.

Ionescu, C. M. and Keyser, R. D., 2003. A novel

parametric model for the human respiratory system.

Proc. of the IASTED International Conference on

Modelling and simulation, Acta Press, pp. 246-251.

Juchem, J., Ionescu, C., Bioimpedance Measurement

Device for Nociceptor Stimulation Detection,

submitted to IEEE Trans on Biomed Circ and Syst

McCaffery, M. and Pasero, C. L., 1997. Pain ratings: The

fifth vital sign. The American Journal of Nurses,

Volume 97, pp. 15-16.

Merboth, M. K. and Barnason, S., 2000. Managing pain:

The fifth vital sign. Nursing Clinics of North America,

Volume 35, pp. 375-383.

Pintelon, R. and Schoukens, J., 2012. System

Identification: A Frequency Domain Approach, Wiley-

IEEE Press. New Jersey, 2nd Edition.

Shieh, J.-S., Dai, C.-Y., Wen, Y.-R., Sun, W.-Z., 2018. A

novel fuzzy pain demand index derived from patient-

controlled analgesia for postoperative pain. IEEE

Transactions on Biomedical Engineering, January,

Volume 54, Issue 12, pp. 2123-2132.

Yang, G., Jiang, M., Ouyang, W., Ji, G., Rahmani, A.,

Liljeberg, P., Tenhunen, H., 2017. IoT-based remote

pain monitoring system: from device to cloud

platform. IEEE Journal of Biomedical and Health

Informatics, DOI: 10.1109/JBHI.2017.2776351.

PhyCS 2018 - 5th International Conference on Physiological Computing Systems