Wearable Wireless Inertial Sensors for Long-Time Monitoring of

Specific Motor Symptoms in Parkinson’s Disease

P. Lorenzi

, R. Rao

, A. Suppa

A. Kita

, R. Parisi

, G. Romano

and F. Irrera

Department of Information Engineering, Electronics and Communication, Sapienza University of Rome,

Via Eudossiana 18, Rome, Italy

Department of Neurology and Psychiatry, Sapienza University of Rome, Viale dell’Università 30, Rome, Italy

Keywords: Biosensors, Wearable Wireless Inertial Sensors, Electronic-Health, Long-Time Home Monitoring,

Parkinson’s Disease.

Abstract: It is proposed an electronic system for the long-time monitoring of specific motor symptoms in patients

affected by Parkinson’s Disease while being at home and making their usual daily activity. The system is

made of a network of non-invasive wireless inertial sensors fixed on the patient body. The muscles activity

is contemporarily analysed through the integration of a circuit for the surface electromyography. Post-

processing algorithms quantify movements in terms of amplitude and power spectrum. Data are

electronically elaborated and wireless transmitted to a receiver in the patient home, to be accessed remotely

by doctors. The challenge is the automatic distinction between specific parkinsonian symptoms including

resting tremor and freezing of gait and patient’s voluntary movements made in daily life. To the aim, the

contemporarily analysis of muscle activity becomes necessary in specific situations, as in the case of

freezing of gait, where accelerometers signals may be misleading. Goal of this research is the

comprehension of all the possible environmental and individual factors which favor worsening of gait

disorders during the patient daily life and the customization of the drug therapy, aiming to preventing

catastrophic events such as falls. Results shown here refer to upper limb tremor and freezing of gait.

1 INTRODUCTION

The analysis of human movement is of major

interest since when, in the last decade, integrated

electronic technologies have finally allowed the

detection, sort and quantification of kinetic

components using the fusion of inertial sensors.

Integrated inertial sensor chips are today

commercially available with a few dollars cost. They

include accelerometers and gyroscopes, an

embedded Micro-Controller Unit (MCU), a non-

volatile memory, a transmission module, in addition

to a battery. Some products devoted to sport training

and rehabilitation and consisting in one sensor with

a dedicated software are commercial. On the

contrary, nothing is available on the market for the

long-time home monitoring of the Parkinson’s

Disease (PD) motor symptoms, although in the last

few years research groups have published papers on

related topics (Patel, 2009; Pantelopoulos, 2010;

Zwartjes, 2010; Patel, 2010; Bächlin, 2010;

Schepers, 2010; Gouwanda, 2011; Becq, 2011;

Taraldsen, 2011; Niazmand, 2011; Sama, 2012;

Caldara, 2014). PD is a chronic neurodegenerative

disorder affecting about 2% of the worldwide

population over 70. Typical PD motor symptoms

include resting tremor, muscle rigidity and

bradykinesia (slowness of movements) (Berardelli,

2001). In PD, motor symptoms manifest when

dopaminergic denervation induces functional

abnormalities in the basal ganglia motor circuits

which in turn drives altered motor inputs in cortical

motor areas. Among PD motor symptoms, tremor is

one the most important and frequently observed in

PD patients. It typically appears in only a single arm

or leg, becoming bilateral later. Frequency of PD

tremor is typically between  4 and  6 hertz.

Tremor crucially worsens when PD patients are not

under dopaminergic therapy (OFF state), whereas it

improves when patients receive their dopaminergic

therapy (ON state). Hence, the long-term monitoring

of tremor amplitude may help the overall clinical

evaluation of PD patients and improve the

therapeutic strategies.

Gait disorders frequently occur in advanced PD

patients and consist of small shuffling steps, reduced

stride length and walking speed during free

ambulation while double support duration and

168

Lorenzi P., Rao R., Suppa A., Kita A., Parisi R., Romano G., Berardelli A. and Irrera F.

Wearable Wireless Inertial Sensors for Long-Time Monitoring of Speciﬁc Motor Symptoms in Parkinsonâ

Zs Disease.

DOI: 10.5220/0005279201680173

In Proceedings of the International Conference on Biomedical Electronics and Devices (BIOSTEC 2015), pages 168-173

ISBN: 978-989-758-071-0

cadence rate are increased. Freezing of gait (FOG) is

typically a transient episode, lasting less than a

minute, in which gait is halted and the patient

complains that feet are glued to the ground. FOG

can be experienced in narrow or tight quarters such

as a doorway or in the presence of an obstacle along

the path and in stressful situations such as when the

telephone or the doorbell rings or when the elevator

door opens. During FOG, PD patients undergo trunk

fluctuations back and forth, move hazardously the

body mass center and load in the forefoot regions,

sometimes resulting in stability loss. At an advanced

stage of the disease, FOG leads to falls in many

instances, in fact, about the 45% of falls of PD

patients occurs forward, due to trunk fluctuations

back and forth. Very recently, some authors have

proposed a IMU based-system which gives an alarm

feedback to assistants or relatives in the case of

patient’s fall (Cabestany, 2013). There are some

evidences that audio stimulations may help the

patient’s to reduce the tendency to undergo FOG.

Auditory stimulations are commonly rhythmic cues,

sometimes embedded in music, set at or slightly

above the patient’s usual cadence. An IMU based-

system has been recently proposed, which gives an

audio feedback to the patient in the case of FOG, to

help the subject to overcome the involuntary block

and prevent the risk of falls (Cabestany, 2013; Sama,

2013; Rodríguez-Martín, 2013). It is evident that the

correct identification of the FOG is crucial, since in

this case any misevaluation of the patient behavior

can be deleterious. For all these reasons, first of all it

is of most relevance to monitor FOG events,

unequivocally distinguishing them from any kind of

voluntary movement, quantifying the daily

frequency, identifying the environmental and the

individual conditions which lead that specific patient

to manifest FOG, finding correlations with the drug

administration and, finally, trying to prevent

catastrophic events such as falls. Optimized drug

therapy can be very effective, especially at an early

stage of the disease. However, drug therapy

optimization is difficult since the response of PD

patients to drugs may vary according to a number of

factors. 24 hours monitoring is the only way to

optimize the therapy and prevent worsening of

symptoms or catastrophic accidents (as falls) due to

incomplete clinical analysis of symptoms during the

day and a consequent not-optimized therapy. On the

other hand, hospitalization is really exceptional

today, due to finance cuts imposed by governments

to national health services. The sensing system

proposed here has the final topic of making possible

the long-time monitoring of specific motor

symptoms of PD while the patient is at home. The

system is composed by a network of several

biosensors disseminated on the patient body which

embed units for the direct non-invasive

measurement of the muscles activity (surface

electromyography, S-EMG). It is being used in the

real-time detection and analysis of PD motor

symptoms. The biosensors are wearable and not

invasive, easy to use and do not need any technical

skill from the patient side. Clinical advantage lies in

the optimization and customization of the drug

therapy for each individual patient. Social benefits

lye in a better quality of life of the patient and the

assisting family.

2 THE SYSTEM

The sensors network system presented in this paper

is designed for both collecting movement signals

and preliminarily analysing them in real-time. This

system is a flexible platform useful for collecting

data via a triaxial accelerometer, a gyroscope and a

magnetometer, with the possibility to incorporate

other information sources in real-time, as the S-

EMG which detects the muscle activity. The Flash

memory stores all inertial data and a Bluetooth

module sends information to other external devices.

The system allows pattern reconstruction of the

kinetic components of movements, discriminates

between voluntary and involuntary movements,

selects only those associated to specific PD

symptoms, reconstructs and, finally, quantifies their

amplitude and frequency. The great challenge of this

work is the automatic association of electronic

signals to specific PD symptoms, filtering all the

signals deriving from voluntary movements. In this

work engineers and neurologists are involved

contemporarily. The engineers develop the

hardware/software system, while the doctors carry

on the clinical research. Patients are clinically

evaluated and movements classified with standard

protocols and compared with voluntary movements

of healthy subjects. This step allows identifying

specific patterns correlated to PD symptoms and is

fundamental for the system calibration. On their

side, engineers optimize the hardware (type of

inertial sensors, protocol of the wireless

communication, entity of the data storage, power

consumption, battery, integration of the EMG, other)

and develop algorithms for data acquisition and

processing. Electrical signals from biosensors are

compared with clinical observations, in order to

achieve the automatic recognition of the disordered

Wearable Wireless Inertial Sensors for Long-Time Monitoring of Speciﬁc Motor Symptoms in Parkinsonâ

Zs Disease

169

movements associated to the disease. The

experiments with patients are performed in the

hospital. A few patients have been studied up to

date, but many others will be studied in the next, to

make reliable the system. In fact, the final goal of

the work is using the system at the patient home, far

from the visual inspection of doctors, and enabling

doctors to access data remotely by a PC. The real

challenge of the research is in the automatic

distinction of voluntary and involuntary movements,

passing through noise filtering, artefacts filtering,

movement reconstruction and the identification of

kinematic components of the PD with quantification

of tremors (amplitude and frequency) and FOG

(duration, trunk fluctuations with associated fall

risk). The complete hard-system, represented in

Fig.1, is composed by a network of 5 body sensors.

Two of them are positioned on the thighs, two on the

shins and the last one on the chest. All the sensors

are controlled by a PC trough the LabView software.

Signals coming from the sensors are acquired

through a PC and then they are post-processed and

displayed with MATLAB.

Figure 1: Sketch of the sensing system.

3 EXPERIMENTS

PD patients were recruited at the Movement

Disorders outpatient clinic of the Department of

Neurology and Psychiatry, Sapienza University of

Rome, Italy. All the experiments involving patients

have been performed at the Laboratory of Human

Motor Control at the same Department. Experiments

with patients are still at an early stage, in the sense

that only a few patients have been studied to date,

and only limb tremors and freezing of gate have

been investigated. Nevertheless, although results are

preliminary, the system demonstrated its potentiality

and versatility in recognizing and quantifying

specific disorders associated to the disease.

3.1 Tremor

In this paragraph traces related to tremor are studied.

Patient A had an asymmetric symptomatology, in

fact while the left hand exhibited a tremor, the right

hand did not. A sensor was positioned on the right

forearm while another was position on the left one.

The patient was asked to walk for 5” and then

turn, for three times. Fig.2a and Fig.2b report the

magnitude of resultant accelerations detected while

walking. As one can see, the two traces are very

different. In fact, in the right sensor it is present only

the signal related to arm oscillations during the steps

and the turnings, while in the left sensor the

oscillation during the steps is reduced with respect to

the right one since the left arm was stuck along the

body and only the tremor is present.

Figure 2: Signals from the y-axis accelerometer while

walking: a) Left Arm (tremor), b) Right Arm (no tremor).

Figure 3: Tremor signals from the accelerometer and S-

EMG while sitting a) Tremor; b) FFT; c) sEMG; d) FFT.

BIODEVICES 2015 - International Conference on Biomedical Electronics and Devices

170

Then, the patient was asked to sit with the arms

resting on the leg. In this way, the tremor amplitude

was detected without any contribution from gait.

Results are shown in Fig.3a. The fast Fourier

transform (FFT) of the acceleration trace gave its

power spectrum, highlighting the expected tremor

frequency component around 3.8 Hz (Fig.3b).

The muscle activity during tremor was also

detected positioning the S-EMG electrodes on the

left arm. As one can see in Fig.3c and 3d the EMG

signal and its FFT are of course compatible with the

accelerometer traces (the low frequency component

in the FFT of the S-EMG trace is not meaningful).

3.2 Freezing of Gait (FOG)

Patients have been asked to execute simple

exercises, as the Timed Up and Go test (TUG).

TUG is a simple test used to assess a person's

mobility and requires both static and dynamic

balance. During the test, the person is expected to

wear their regular footwear and use any mobility

aids that they would normally require. Patient A was

asked to execute the TUG. The exercise included the

following movements (see Fig.4): 1) standing-up; 2)

motion-less; 3) walking a few meters; 4) turning; 5)

walking; 6) sitting down. An obstacle was

positioned on the floor, along the walking trajectory.

Figure 4: Sketch of the axis orientation and TUG.

Referring to Fig. 4, the patient walking trajectory

is along the z-axis, the y-axis being along gravity

and the x axis perpendicular to the walking

direction, on the same plane.

The main scope of this experiment was the

detection of the FOG and therefore results will be

discussed in the following showing clearly the

occurrence of such an event while walking.

Patient A was asked to execute the TUG. Fig.5a

displays traces relative to the linear acceleration

Figure 5: a) acceleration along the y-axis revealed by the

sensor on the right thigh; b) angular velocity around the x

axis revealed by the sensor on the chest.

along the y-axis revealed by the sensor positioned on

the right thigh. Fig.5b displays the signal from the

gyroscope of the sensor positioned on the chest and

refers to the angular velocity of the trunk around the

x axis. The patient starts his TUG from the sitting

position. In this condition, gravity is not entirely

projected on the y-axis of the thigh sensor, in fact in

Fig.5a the y-axis acceleration is lower than 9 m/s

Traces outlines that:

1. STANDING-UP: When the patient stands up,

gravity becomes entirely projected on the y-axis of

the thigh sensor. This can be seen looking at the

black line of Fig.5a, which raises to 9 m/s

after 3”

approximately from the beginning of the TUG.

While standing-up from the sitting position, also the

gyroscope positioned on the chest clearly detects the

body movement (Fig.5b) and changes its value.

2. MOTIONLESS: the patient is asked to rest

for a few seconds, to assess its postural stability

(which looks quite upright, corresponding to 0 deg/s

from the chest gyroscope and 9 m/s2 from the thigh

accelerometer)

3. WALKING: the postural stability keeps on

while walking, in fact the angular velocity from the

chest sensor is still around zero. The patient makes

just three steps starting with the right leg, as clearly

shown in Fig.6a, where the very first instants of the

green trace indicate that the leg is moved up

Wearable Wireless Inertial Sensors for Long-Time Monitoring of Speciﬁc Motor Symptoms in Parkinsonâ

Zs Disease

171

(acceleration become greater than gravity) and then

moved down for three times. The highest peaks are

inertial impacts when the foot falls on the ground.

4. FOG: At a certain point of his walk, the

patient encounters the obstacle and experiments a

freezing of gait lasting several seconds. In that time

interval the patient makes many attempts to walk,

oscillating the body essentially back and forth and

loading the fore-toes, but he does not make any step

as if feet got glued to the ground. This can be seen in

Fig.5a, where the characteristic peaks related to

steps are absent, but a dense succession of smaller

peaks add to the y-axis acceleration trace, whose

mean value keeps around 9 m/s

. At the same time,

the chest sensor detects many changes of the angular

velocity around the x-axis (perpendicular to the

ground, along the walking direction) related to trunk

fluctuations (see Fig.5b). These traces are typical of

a FoG event, which, in this experiment, was favored

by the presence of an obstacle along the walking

trajectory.

It is interesting to compare the traces relative to

“FOG”, to those relative to “motionless”, which is

voluntary. Signals from the accelerometer are

similar, whereas signals from the gyroscope are

quite different. Therefore, just looking at the traces

from the linear accelerometer it is not possible to

distinguish the involuntary FOG from the voluntary

resting position. On the contrary, the measurement

of angular velocity around the x-axis is a clear

indication of trunk fluctuations due to FOG.

We wish to recall that FOG is often the cause of

falls, and its definitive identification is absolutely

necessary. To the aim, traces from the S-EMG are

being analyzed and results will be presented at the

conference.

5. OUT OF FoG: Finally, after about eight

seconds, the patient makes a step and the signal from

the y-accelerometer on the thigh changes its value

and exhibits a shape related to a step. At the same

time, trunk fluctuations tend to stop.

6. TURNING: the patient turns leftward, with a

few small steps. Nothing relevant on the trace.

7. WALKING: the second walking is similar to

the first one.

8. SITTING DOWN: The traces relative to

sitting-down are complementary to those relative to

standing-up. Nothing relevant on the trace.

4 CONCLUSIONS

It was proposed an electronic system for the long-

time monitoring of specific motor symptoms of

patients affected by Parkinson’s Disease while being

at home and making their usual daily activity.

The system is made of a network of non-

invasive wireless inertial sensors fixed on the patient

body. The muscles activity is contemporarily

analysed through the integration of a circuit for the

surface electromyography. Post-processing

algorithms quantify movements in terms of

amplitude and power spectrum.

A few patients of the Movement Disorders

outpatient clinic of the Department of Neurology

and Psychiatry, Sapienza University of Rome were

studied up to date, and asymmetric tremor and

freezing of gait were analysed.

As a result, starting from signals from the

accelerometers signals and the surface-

electromyography tremor was unequivocally

distinguished respect to voluntary movements, and

its amplitude was quantified; the power spectrum

revealed a tremor frequency around 3.8 Hz. On the

contrary, the freezing of gait was distinguished only

thanks to the detection of a gyroscope positioned on

the patient chest, since the accelerometers were not

able to distinguish unequivocally between voluntary

resting in the upright position and the involuntary

gait block.

At the conference, results relative to the

automatic recognition of the freezing of gait will be

presented, obtained on a statistically meaningful

number of patients.

ACKNOWLEDGEMENTS

Authors wish to thank patients of the Movement

Disorders outpatient clinic of the Department of

Neurology and Psychiatry, Sapienza University of

Rome, Italy, who accepted to be involved in the

research.

REFERENCES

Patel, S.; et al.., "Monitoring Motor Fluctuations in

Patients With Parkinson's Disease Using Wearable

Sensors," Information Technology in Biomedicine,

IEEE Transactions on vol.13, no.6, pp.864,873

Nov.2009.

Pantelopoulos, A; Bourbakis, N.G., "A Survey on

Wearable Sensor-Based Systems for Health

Monitoring and Prognosis," Systems, Man, and

Cybernetics, Part C: Applications and Reviews, IEEE

Transactions on vol.40, no.1, pp.1,12, Jan.2010.

Zwartjes, D.G.M et al., "Ambulatory Monitoring of

Activities and Motor Symptoms in Parkinson's

BIODEVICES 2015 - International Conference on Biomedical Electronics and Devices

172

Disease," Biomedical Engineering, IEEE Transactions

on vol.57, no.11, pp.2778,2786,Nov.2010.

Patel, S.; et al., "Home monitoring of patients with

Parkinson's disease via wearable technology and a

web-based application," Engineering in Medicine and

Biology Society (EMBC), 2010 Annual International

Conference of the IEEE , vol., no., pp.4411,4414, Aug.

31 2010.

Bächlin M, et al., “A wearable system to assist walking of

Parkinsons disease patients”, Methods Inf Med. 2010;

49(1):88-95.

Schepers, H.; et al., Ambulatory human motion tracking

by fusion of inertial and magnetic sensing with

adaptive actuation Medical & Biological Engineering

& Computing, Springer-Verlag, 2010, 48, 27-37.

Gouwanda, D. & Senanayake, S. M. N. A. Periodical gait

asymmetry assessment using real-time wireless

gyroscopes gait monitoring system, Journal of Medical

Engineering & Technology, 2011, 35, 432-440.

Becq, G.; et al., P. Classification of epileptic motor

manifestations using inertial and magnetic sensors,

Computers in Biology and Medicine, Computers in

Biology and Medicine, Elsevier, 41,46-55, Jan.2011.

Taraldsen, K.; et al., J. L. Physical activity monitoring by

use of accelerometer-based body-worn sensors in older

adults: A systematic literature review of current

knowledge and applications, Maturitas, Maturitas,

Elsevier, 71, 13-19, 2011.

Niazmand, K.; et al., "Freezing of Gait detection in

Parkinson's disease using accelerometer based smart

clothes," Biomedical Circuits and Systems Conference

(BioCAS), 2011 IEEE, vol., no., pp.201,204, 10-12

Nov. 2011.

Sama, A. et al.; "Dyskinesia and motor state detection in

Parkinson's Disease patients with a single movement

sensor," Engineering in Medicine and Biology Society

(EMBC), pp.1194, 1197, Aug. 28 2012-Sept. 1 2012.

Caldara, M.; et al. “A novel body sensor network for

Parkinson’s disease patients rehabilitation assessment”

2014 11th Int. Conf. on Wearable and Implantable

Body Sensor Network, 81, 86.

Berardelli, A.; et al., “Pathophysiology of bradykinesia in

Parkinson’s disease”. Brain, 2001;124:2131-2146.

Cabestany, J.; et al., “One step towards an automatic aging

people fall detection service," Mixed Design of

Integrated Circuits and Systems (MIXDES), 2013

Proceedings of the 20th International Conference,

vol., no., pp.545,552, 20-22 June 2013.

Cabestany, J.; et al.: “REMPARK: When AI and

technology meet Parkinson Disease assessment,"

Mixed Design of Integrated Circuits and Systems

(MIXDES), 2013 Proceedings of the 20th

International Conference , vol., no., pp.562,567, 20-22

June 2013.

Sama, A.; et al., “A heterogeneous database for movement

knowledge extraction in parkinson’s disease”

European symposium on artificial neural networks,

computational intelligence and machine learning,

2013.

Rodríguez-Martín D, et al., “A wearable inertial

measurement unit for long-term monitoring in the

dependency care area” Sensors (Basel). 2013 Oct 18;

13(10):14079-104.

Wearable Wireless Inertial Sensors for Long-Time Monitoring of Speciﬁc Motor Symptoms in Parkinsonâ

Zs Disease

173