Content uploaded by Maurizio Ferrarin
Author content
All content in this area was uploaded by Maurizio Ferrarin on Jan 03, 2019
Content may be subject to copyright.
TNSRE-2018-00370.R1
1
Abstract— Stair ascent is a challenging daily-life activity
highly related to independence. This task is usually assessed with
clinical scales suffering from partial subjectivity and limited
detail in evaluating different task’s aspects. In this study we
instrumented the assessment of stair ascent in people with
Multiple Sclerosis (MS), stroke (ST) and Parkinson’s disease
(PD) to analyze the validity of the proposed quantitative indexes
and characterize subjects’ performances. Participants climbed 10
steps wearing a magneto-inertial sensor (MIMU) at sternum
level. Gait pattern features (step frequency, symmetry,
regularity, harmonic ratios), and upper trunk sway were
computed from MIMU signals. Clinical mDGI (modified
Dynamic Gait Index) and mDGI-Item 8 (“Up stairs”) were
administered. Significant correlations with clinical scores were
found for gait pattern features (rs>=0.536) and trunk pitch sway
(rs<=-0.367) demonstrating their validity. Instrumental indexes
showed alterations in the three pathological groups compared to
healthy subjects, and significant differences, not clinically
detected, among MS, ST and PD. MS showed the worst
performance, with alterations of all gait pattern aspects and
larger trunk pitch sway. ST showed worsening in gait pattern
features, but not in trunk motion. PD showed fewer alterations
consisting in reduced step frequency and trunk yaw sway. These
results suggest that the use of a MIMU provided valid objective
indexes revealing between-group differences in stair ascent not
detected by clinical scales. Importantly, the indexes includes
upper trunk measures, usually not present in clinical tests, and
provides relevant hints for tailored rehabilitation.
Index Terms— Stair negotiation; Inertial sensors; Multiple
Sclerosis; Stroke; Parkinson’s disease.
I. INTRODUCTION
TAIR negotiation is a common daily life activity highly
related to independence and community participation [1].
Compared to level-ground walking, stairway walking is a
more demanding task, requiring larger moments and range of
motion at lower limb joints [2],[3], as well as additional
charge on balance control system [3]. This task is highly
Manuscript submitted September 04, 2018; revised October 29, 2018;
accepted November 09, 2018. This work was supported by Italian Ministry of
Health (Ricerca Corrente). (Corresponding author: Davide Cattaneo)
I. Carpinella, E. Gervasoni, D. Anastasi, T. Lencioni, D. Cattaneo, and M.
Ferrarin are with the IRCCS Fondazione Don Carlo Gnocchi, 20148 Milan,
Italy (e-mail: dcattaneo@dongnocchi.it).
dependent on muscle strength [2],[3], postural control [4],
sensory processing and integration [4], and psychological
factors including fear of falling [4]. Consequently, alterations
of stair walking have been reported in elderly subjects [5], and
people with Multiple Sclerosis (MS) [6], stroke (ST) [5] and
Parkinson’s disease (PD) [3], who are generally at high risk of
falls [7]. The aggravating factor is that stairway falls,
compared to falls while level-walking, involve a dramatically
higher risk of death or severe injuries [5]. In addition, as
discussed by Morone et al. [8] the ability to ascend stairs
highly affects daily-living independence, especially in patients
who usually have to manage stairs at home and/or at work.
Given the impact of stair negotiation on quality of life, this
task is increasingly included in both clinical assessment [9]
and rehabilitation [10],[11]. Regarding assessment, Van Iersel
et al. [9] found 43 clinical tests incorporating an item on stair
walking. In the following years, other scales evaluating stair
negotiation have been validated, such as the modified
Dynamic Gait Index (mDGI) [12]. Most of these tools require
the examiner to assign a score on an ordinal scale and/or to
measure the task duration through a stopwatch. Although
widely used, these tests suffer from partial subjectivity,
limited resolution, ceiling effect, and poor detail in analyzing
different components of the task [13], in particular upper trunk
movements that are often not assessed during the evaluation.
These limitations can be partly overcome by instrumental
methods, that can provide clinicians with additional
quantitative information to better characterize the task
performance, tailor the intervention, and objectively measure
its effects. In particular, cost-effective magneto-inertial
measurement units (MIMUs) allow clinicians to easily
perform objective evaluations of motor deficits during routine
exams, outside typical movement analysis laboratories.
MIMUs have been widely used to analyze level-ground
walking [14],[15], while only few studies exist about their use
during stair negotiation [16],[17]. In particular, these studies
investigated, in PD subjects, the anticipatory postural
adjustments preceding one-step ascent [16], and the trunk
rhythmicity during ascending and descending three steps [17].
The aim of the present study was to instrument the
assessment of stair ascent, as described by mDGI-Item 8,
using a single MIMU. Healthy subjects (HS), and people with
MS, ST and PD were tested to i) analyze the validity of the
Instrumental Assessment of Stair Ascent in People
with Multiple Sclerosis, Stroke and Parkinson’s
Disease: a Wearable-Sensor Based Approach
Ilaria Carpinella, Elisa Gervasoni, Denise Anastasi, Tiziana Lencioni, Davide Cattaneo, and Maurizio
Ferrarin Senior Member, IEEE
S
TNSRE-2018-00370.R1
2
MIMU-based indexes for evaluating stair ascent, and ii)
characterize and compare the performance of subjects with
different neurological diseases, providing potentially useful
information to complement clinical assessment.
II. METHODS
A. Participants
Twenty HS without any musculoskeletal or neurological
disorders [mean age (SD): 58.0 (14.5) years], and thirty
subjects with comparable age [59.9 (16.1) years] affected by
neurological diseases (NEU) were recruited. NEU group
consisted in ten subjects with MS, ten post-stroke (ST), and
ten with PD. Inclusion criteria were: ability to walk 10 meters,
ability to ascend 10 steps, and Mini Mental State Examination
> 24. Sample size calculation was based on step frequency
data published in three previous studies analyzing stair ascent
in MS [18], ST [19] and PD [17]. Step frequency was
considered since it is the only parameter present in all three
studies. Mean data reported for the 3 control groups (HS) and
for the 3 pathological samples (NEU) were averaged and the
standard deviation was set equal to the highest one shown in
the 3 studies. Analysis of the pooled data indicated that a
minimum of 36 subjects (18 HS and 18 NEU) was required to
detect a difference between HS and NEU (Cohen’s d = 1.48,
Power = 0.99, p = 0.05). All subjects gave written informed
consent to the protocol that was approved by the ethical
committee of Don Carlo Gnocchi Foundation (Milan, Italy).
B. Experimental Protocol
Disease severity was defined for each pathological group
according to disease-specific scales: Expanded Disability
Status Scale (EDSS) [20] for MS, modified Rankin Scale
(mRS) [21] for ST, and Hoehn and Yahr stage (H&Y) [22] for
PD. All HS and NEU subjects were clinically assessed with
the mDGI [12], that consists of 8 items evaluating balance in
different walking conditions on a 8-point ordinal scale. The
mDGI maximum score is 64, indicating unaltered balance.
Ten-meter walk test (10MWT) [23] was also administered to
clinically assess gait velocity during level-ground walking.
Instrumental assessment was performed during the mDGI -
Item 8 (“Up stairs”). As described by Shumway-Cook et al.
[12], the subject stood upright at the bottom of a 10-step stair,
then he was required to walk up the stairs at self-selected
speed and stop with both feet on the 10th step. The use of
handrail was allowed. Participants climbed stairs wearing a
wireless MIMU (MTw, Xsens, NL) positioned on the sternum
with an elastic band [24], over the clothes [Fig. 1(a)]. The
MIMU consisted of a 3D accelerometer (±160 m/s2 range), a
3D gyroscope, (±1200deg/s range) and a 3D magnetometer
(±1.5 Gauss). MIMU’s orientation in space was estimated
from raw signals by a sensor fusion algorithm implemented on
a digital signal processor embedded in MIMU housing. MIMU
signals were sampled at 100 Hz. One trial was recorded from
each subject. PD subjects were tested while they were on-
phase during antiparkinsonian therapy, approximately 2 hours
after medication intake.
C. Data Processing
A set of instrumental parameters descriptive of stair gait
pattern and upper trunk movements were computed from
MIMU signals using MATLAB (MathWorks, Natick, MA).
Step frequency [step/s]: peak frequency of the power
spectrum of the vertical acceleration [25]. Step frequency
was used as an estimate of stair ascent velocity, in order to
avoid errors due to the integration of the acceleration. This
approximation was considered adequate considering that
stair ascent velocity can be computed as step distance X
step frequency, and that step distance is partly constrained
by the sizes of the staircase steps [i.e. step distance ~ √(step
depth2 + step height2)].
Step Symmetry and Stride Regularity [unitless]:
respectively, first and second peak of the normalized auto-
correlation function computed from the acceleration
modulus [26]. Increasing values, from 0 to 1, indicate
higher symmetry and regularity, respectively (see
Supplementary Figure S1).
Harmonic Ratio (HR) [unitless]: computed for each
component of the acceleration as described by Menz et al.
Fig. 1. (a) Placement of the magneto-inertial measurement unit. Orientation of sensing axes (X, Y, Z) is indicated. (b) Trunk pitch angle recorded during stair
ascent from a healthy subject (thick gray line), a subject with MS (dashed line), a subject with ST (thin black line), and a subject with PD (thick black line).
TNSRE-2018-00370.R1
3
[27]. Stride frequency is used as the fundamental frequency
of the periodic accelerations during stairway walking; the
fundamental period of such signals is a multiple of the
stride duration. The amplitudes of the first 10 even
harmonics (EvenHi, i: 1..10) and the first 10 odd harmonics
(OddHi, i: 1..10) are computed through a finite Fourier
series; then the HR is calculated following (1) by dividing
the sum of the amplitudes of the in phase harmonics by the
sum of the amplitudes of the out of phase harmonics.
(1)
Higher values of HR indicate more rhythmic movements
[17],[27]-[29]. HR is particularly important to be
investigated since previous studies on elderly and
parkinsonian populations [30],[31] found significantly
lower values (i.e. lower rhythmicity) in fallers compared to
non-fallers. Moreover, upper trunk HR demonstrated to be
a significant predictor of falls in elderly [30]
Trunk Sway [deg]: standard deviation of pitch, roll and yaw
angles recorded by the MIMU.
To avoid errors possibly introduced by the detection of step
initiation/termination, all indexed were computed considering
the entire portion of steady-state signals recorded during the
middle eight steps [32].
D. Statistical Analysis
Fisher exact test (FET) was used to compare sex and
number of handrail users among HS, MS, ST and PD groups.
The same test was applied to compare the number of mild
moderate and severe subjects in MS, ST and PD. Kruskal-
Wallis test with Bonferroni-Holm (BH) post-hoc procedure
was used to compare age and clinical features, since these data
were not normally distributed (Shapiro-Wilk’s test < 0.05).
Instrumental parameters were compared among groups using
ANCOVA with one between-group factor (Group: HS, MS,
ST, PD), and age and step frequency as covariates. The choice
of the covariates was due to previous data showing that both
age and cadence have an influence on MIMU-based
parameters (e.g. harmonic ratio [29]). In case of significant
differences (p<0.05) revealed by ANCOVA, separate post-hoc
comparisons were performed using Fisher’s test with
Bonferroni–Holm correction. Some variables did not meet the
assumptions of data normality and/or homogeneity of
variances (Shapiro-Wilk’s test and/or Levene test, p<0.05). In
these cases ANCOVA was applied on transformed data (Box-
Cox transformation). To check the possible effect of handrail
use, the same analysis was conducted excluding subjects using
handrail. Given the small sample size, in this case post-hoc
analysis was performed using Fisher’s test with no correction
for multiple comparisons. The concurrent validity of the
instrumental indexes was tested analyzing their correlation
with mDGI score and mDGI – Item 8 sub-score through
Spearman’s correlation coefficient (rs). Partial Spearman
coefficient (prs) was also calculated to correct for age, in the
case of step frequency, and for age and step frequency in the
case of the other parameters. Analyses were performed using
STATISTICA (Statsoft, Tulsa, OK).
III. RESULTS
A. Clinical Assessment
Results are reported in Table I. A statistically significant
difference was found in age, with PD patients being older than
HS and MS subjects (pBH=0.03). Time since diagnosis was
comparable between MS and PD (pBH=0.307), while it was
significantly lower in ST (pBH<=0.006). In particular, 4 ST
subjects were in the sub-acute stage (< 6 months post-stroke),
while 6 were in the chronic stage (>= 6 months post-stroke).
Six and four ST patients had left and right hemiparesis,
respectively. Seven and three ST subjects had ischemic and
hemorrhagic stroke, respectively. Regarding disease severity,
median (range) EDSS score for MS was 5 (2-6), with 1 subject
being in the mild stage (EDSS: 2), 6 in the moderate stage
(EDSS: 4.5-5.5), and 3 in the severe stage (EDSS: 6-6.5).
Median (range) mRS score for ST was 3 (2-3). Five ST
subjects showed mild disability (mRS:2) and 5 moderate
disability (mRS:3). Median (range) H&Y score for PD was 3
(2-4), with 4 subjects being in the mild stage of the disease
(H&Y: 2-2.5), 4 in the moderate stage (H&Y: 3), and 2 in the
severe stage (H&Y: 4). No significant difference between
groups was found in the number of mild, moderate and severe
subjects (pFET=0.208). Gait and balance clinical scores (Table
I) indicated that the three pathological groups showed lower
walking speed (10MWT, pBH<=0.018), impaired dynamic
balance (mDGI, pBH<0.001), and abnormal stair ascent
10
1
10
1
)(
ii
ii
APVT OddH
EvenH
HR
10
1
10
1
ii
ii
ML EvenH
OddH
HR
TABLE I
DEMOGRAPHIC AND CLINICAL CHARACTERISTICS OF HEALTHY SUBJECTS (HS, a) AND SUBJECTS WITH MULTIPLE SCLEROSIS (MS, b), STROKE (ST, c) AND
PARKINSON’S DISEASE (PD, d)
HS (a)
(N = 20)
MS (b)
(N = 10)
ST (c)
(N = 10)
PD (d)
(N = 10)
p-value
Sex [men/women]
10/10
4/6
4/6
2/8
0.474
Age [years]
57 (51-75) d
51 (35-66) d
59 (47-70)
73 (61-77) a,b
0.024
Time since diagnosis [years]
-
8.5 (7-17) c
0.6 (0.3-1.3) b,d
7.5 (3-13) c
0.001
10MWT - Walking speed [m/s]
1.22 (1.12-1.46) b,c,d
0.84 (0.60-1.12) a
0.85 (0.72-1.02) a
0.96 (0.63-1.20) a
0.001
mDGI Total Score [0-64]
64 (64-64) b,c,d
49 (34-55) a
47 (42-53) a
51 (39-55) a
<0.001
mDGI - Item 8 “Up stairs” score [0-8]
8 (8-8) b,c,d
6 (4-7) a
6 (5-7) a
7 (5-7) a
<0.001
Values are median (1st -3rd quartile) or number. 10MWT: 10-meter Walk Test; mDGI: modified Dynamic Gait Index.
p-value: results of Fisher Exact test (FET) for sex, and Kruskal-Wallis test for all the other variables. The superscript letters a, b, c, d indicate a statistically
significant difference (p<0.05) with respect to the corresponding group (Bonferroni-Holm post hoc test).
TNSRE-2018-00370.R1
4
performances (mDGI-Item 8, pBH<=0.002), compared to HS.
By contrast, clinical scores were comparable among MS, ST
and PD (pBH>=0.558). A similar number of subjects in the
three groups climbed stairs using handrail (MS: 6, ST: 6, PD:
5, pFET =0.814).
B. Concurrent Validity
Table II shows the correlations between instrumental
parameters describing stair ascent, and mDGI Total score and
mDGI-Item 8 sub-score on the whole sample. Statistically
significant correlations (rs) were found for all stair gait pattern
features (step frequency, step symmetry, stride regularity and
harmonic ratios), and for trunk pitch sway. Similar results
were found also after correcting for age and step frequency
(prs) with the exception of trunk pitch sway versus mDGI-Item
8 sub-score. An ancillary correlation analysis, performed
separately on MS, ST and PD, showed that step frequency and
step symmetry significantly correlated with mDGI-Item8 sub-
score in all groups (step frequency: 0.92<=rs<=0.95; Step
Symmetry: 0.58<=rs<=0.76, p<0.05). Different behavior was
noticed regarding the other parameters. In MS significant
correlations were found in stride regularity (rs=0.77), antero-
posterior HR (rs=0.64), trunk pitch (rs=-0.72) and roll sway
(rs=-0.66). In ST a significant correlation was found in trunk
yaw sway (rs=0.80), while in PD a trend toward significant
correlations (p=0.06) was found in stride regularity (rs = 0.59),
medio-lateral HR (0.61) and vertical HR (0.61).
C. Instrumental Assessment
As shown in Table III, statistically significant differences
were found in all instrumental parameters excluding stride
regularity and trunk roll sway. Post-hoc analysis revealed a
significant reduction of step frequency in all pathological
groups compared to HS (pBH<0.001). While stride regularity
was comparable among groups, step symmetry was
significantly higher in HS compared to MS and ST groups
(pBH<0.001). MS and ST subjects were less symmetrical than
PD, although the statistical significance was met only for MS
versus PD (pBH=0.006) (see Supplementary Figure S1)
Harmonic Ratios (HRs) showed a statistically significant
reduction in medio-lateral direction in all pathological groups
(pBH<=0.034), while antero-posterior and vertical HRs were
reduced only in MS (pBH<0.001) and ST (pBH<=0.013).
Antero-posterior HR was significantly higher in PD compared
to MS (pBH =0.004) and ST (pBH=0.012). A similar trend was
noticed in vertical HR, although the statistical significance
was found only for PD versus MS (pBH=0.032).
Trunk pitch sway was significantly larger in MS compared
to HS (pBH <0.001), ST (pBH=0.021) and PD (pBH=0.001) [see
Fig.1(b)]. Trunk roll sway was comparable among groups,
while trunk yaw sway was significantly reduced in PD,
compared to HS (pBH=0.018) and MS (pBH=0.016).
The analysis of subjects not using handrail (Table IV)
confirmed these results, with the exception of stride regularity,
which was significantly reduced in MS and ST compared to
HS and PD, and medio-lateral rhythmicity which was reduced
in MS and ST, but not in PD.
IV. DISCUSSION
In this study we instrumented the assessment of stair ascent
using a single MIMU applied on the upper trunk of healthy
subjects and subjects with MS, ST and PD to measure a
clinically relevant task. The use of a MIMU allowed to obtain
quantitative characterization of subjects performance directly
in clinical setting, where common stair flights are usually
present, with minimal preparation time (less than 1 minute for
MIMU placement), and without the need of expensive
equipment (e.g. instrumented staircase and/or optoelectronic
systems [3],[19]) or specialized personnel. The instrumented
assessment provided valid and objective parameters which
disclosed between-group differences not detected by clinical
scales. Importantly, the MIMU-based indexes include
measures of upper trunk movements, usually not considered
by clinical scales. Despite the high impact of stair walking on
independence and quality of life [1], to our knowledge this is
the first study addressing the use of wearable sensors to
characterize and compare 10-step stair ascent in subjects with
different neurological diseases but comparable number of
mild, moderate and severe subjects and comparable clinical
score for balance (mDGI) and level-walking speed (10MWT).
All stair gait pattern features and trunk pitch sway
significantly correlated with clinical scores on mDGI and
mDGI-Item8 related to the whole sample. The correlation with
TABLE II
SPEARMAN’S CORRELATION COEFFICIENT (rs) AND PARTIAL SPEARMAN’S CORRELATION COEFFICIENT (prs) BETWEEN INSTRUMENTAL PARAMETERS AND
CLINICAL mDGI TOTAL SCORE AND mDGI – ITEM 8 (“UP STAIRS”) SUB-SCORE.
mDGI – Total score
mDGI - Item 8 score
Instrumental Parameter
rs
prs
rs
prs
Step Frequency[step/s]
0.916***
0.928***
0.931***
0.932***
Step Symmetry [unitless]
0.630***
0.399**
0.641***
0.373**
Stride Regularity [unitless]
0.598***
0.445***
0.621***
0.477***
Harmonic Ratio antero-posterior [unitless]
0.642***
0.355**
0.601***
0.195†
Harmonic Ratio medio-lateral [unitless]
0.583***
0.315*
0.554***
0.271*
Harmonic Ratio vertical [unitless]
0.553***
0.399**
0.536***
0.330*
Trunk Pitch Sway [deg]
-0.397**
-0.242*
-0.367**
-0.059
Trunk Roll Sway [deg]
-0.266†
-0.133
-0.239†
0.076
Trunk Yaw Sway [deg]
0.212
-0.255
0.269†
-0.030
†p<0.1, *p<0.05, **p<0.01, ***p<0.001 (Bonferroni-Holm correction).
TNSRE-2018-00370.R1
5
mDGI total score was maintained even after correcting for age
and step frequency, thus demonstrating the concurrent validity
of these parameters to quantitatively assess stair ascent in
neurologically-impaired subjects spanning from mild to severe
involvement. An ancillary correlation analysis conducted on
each pathological group revealed that, while step frequency
and symmetry significantly correlated with mDGI-Item8 sub-
score in all groups, different results were found for the other
features. In particular, increasing ability to ascend stair (i.e.
higher mDGI-Item8 score) was associated: i) in MS with
higher antero-posterior rhythmicity, and lower trunk pitch and
roll sway, ii) in ST with larger trunk yaw motion, iii) in PD
with higher medio-lateral and vertical rhythmicity. Despite the
small number of subjects in the 3 groups, these results seem to
indicate that different indexes could be used as best descriptor
of stair ascent ability in MS, ST and PD.
In accordance with mDGI-Item8 sub-score, the instrumental
indexes showed significant alterations of stair ascent in MS,
ST and PD groups compared to HS. Interestingly, statistically
significant differences in sensor-derived parameters were
found also among the three pathological groups, enforcing the
results found by Cattaneo et al. [33] about disease-specific
deficits of static balance. Importantly, the statistical model
used in the present study included as covariates two factors
which could influence the results, i.e. age and step frequency
[3],[17],[29], the latter used as an estimate of stair walking
speed. This suggests that the differences found among MS, ST
and PD are not attributable to these factors. In addition, since
the clinical mDGI-Item8 score was comparable among the
three groups, the present results suggested the higher
sensitivity of the instrumental assessment of stair ascent,
which could therefore complement the clinical evaluation.
In accordance with a previous study on standing balance
[33], MS subjects showed the greatest impairment during stair
ascent, with alterations of 7 out of 9 instrumental parameters.
Statistically significant reduction of step frequency, step
symmetry, stride regularity, and harmonic ratios was found in
MS compared to HS and PD groups. Stride regularity was
found to be altered only in subjects not using handrail,
confirming the stabilization effect of this aid. No specific
studies exist about instrumental assessment of stair ascent in
MS, however, the present results confirm previous works
about level-ground walking, which found altered spatio-
temporal parameters [34], increased asymmetry [35],[36],
reduced rhythmicity [28] and regularity [37]. These alterations
are potentially caused by proprioceptive deficits associated to
MS [38]. Although these deficits are present also in PD and
ST subjects, it can be speculated that they could be more
pronounced in MS, that is the only disease here considered
directly affecting spinal cord, causing significant reduction of
spinal afferent conduction velocity [6],[38]. Another aspect
could be the difficulty of MS subjects to correctly interpret
and integrate vestibular inputs, possibly due to increased head
instability. In fact, previous published data showed larger head
displacements in MS subjects during treadmill walking [39].
Although in the present study head motion during stair ascent
was not directly measured, the presence of head instability in
MS could be a plausible hypothesis considering that trunk
showed lower than normal rhythmicity (i.e. decreased HRs)
and larger pitch sway. These anomalies, in turn, could have
reduced the capacity of trunk to perform its important role of
attenuating movement-related oscillations from lower-body
segments to stabilize the head during locomotor tasks [40].
Taken together, the above aspects would lead to difficulties in
foot placement [41], thus worsening stair ascent pattern in MS
compared to the other groups. Interestingly, MS subjects
TABLE III
INSTRUMENTAL PARAMETERS DESCRIBING STAIR ASCENT IN HEALTHY SUBJECTS (HS, a), SUBJECTS WITH MULTIPLE SCLEROSIS (MS, b), STROKE (ST, c) AND
PARKINSON’S DISEASE (PD, d).
HS (a)
(N = 20)
MS (b)
(N = 10)
ST (c)
(N = 10)
PD (d)
(N = 10)
F
(p-value)
Step Frequency [step/s]
2.05 (1.87-2.24) b,c,d
1.07 (0.80-1.34) a
1.22 (0.96-1.48) a
1.45 (1.18-1.73) a
16.98
(<0.001)
Step Symmetry [unitless]
0.81 (0.75-0.87) b,c
0.63 (0.49-0.76) a,d
0.69 (0.59-0.79) a
0.80 (0.74-0.86) b
3.41
(0.026)
Stride Regularity [unitless]
0.81 (0.74-0.88)
0.67 (0.55-0.79)
0.72 (0.63-0.81)
0.78 (0.70-0.86)
1.52
(0.223)
Harmonic Ratio antero-posterior [unitless]
2.73 (2.36-3.09) b,c
1.44 (0.96-1.92) a,d
1.61 (1.17-2.04) a,d
2.52 (2.01-2.95) b,c
6.95
(<0.001)
Harmonic Ratio medio-lateral [unitless]
3.07 (2.61-3.53) b,c,d
1.75 (1.14-2.36) a
2.29 (1.74-2.84) a
2.36 (1.81-2.91) a
3.23
(0.031)
Harmonic Ratio vertical [unitless]
3.22 (2.65-3.79) b,c
1.75 (0.99-2.50) a,d
2.16 (1.48-2.85) a
3.22 (2.55-3.90) b
3.83
(0.016)
Trunk Pitch Sway[deg]
1.34 (1.07-1.60) b
2.18 (1.60-2.76) a,c,d
1.42 (1.08-1.76) b
1.26 (0.96-1.57) b
3.67
(0.019)
Trunk Roll Sway [deg]
2.08 (1.46-2.69)
2.05 (1.24-2.86)
2.08 (1.35-2.82)
1.57 (0.84-2.30)
0.51
(0.680)
Trunk Yaw Sway [deg]
3.74 (3.01-4.47) d
5.06 (4.11-6.02) d
4.33 (3.46-5.20)
2.84 (1.98-3.70) a,b
4.10
(0.012)
Values are mean (95% confidence interval), adjusted though ANCOVA for age, in the case of Step Frequency, and for age and Step Frequency for all the
other parameters. F (p-value): results of ANCOVA. Degrees of freedom are (3,45) for Step Frequency, and (3,44) for all the other parameters. The superscript
letters a, b, c, d indicate a statistically significant difference (p<0.05) with respect to the corresponding group (Fisher’s post -hoc test with Bonferroni-Holm
correction).
TNSRE-2018-00370.R1
6
showed higher trunk pitch sway compared to HS, ST and PD
groups, complementing previous results showing larger
center-of-pressure sway during standing in MS patients [33].
Again, spinal cord involvement could play a role in the
significantly larger trunk sway of MS subjects, who
compensated for their slowed spinal afferent conduction by
increasing the magnitude of trunk postural responses to
maintain balance during stair ascent [6],[38]. A second
hypothesis can be formulated, considering the study of Nadeau
et al. [2]. The authors demonstrated that knee extension
moment is two times greater during stair ascent compared to
level-ground walking, implying that knee extensor muscles
have a dominant role in this task [2]. Since MS patients have
been demonstrated to show impairment of quadriceps and
hamstrings force [42] and increased fatigue [35], it can be
speculated that these subjects increase their trunk antero-
posterior movement to compensate for the reduced strength of
knee extensors. This strategy, in turn, may help them to ascend
stairs by reducing the knee moment and, consequently, the
quadriceps demand, as documented in other pathologic
subjects presenting weakness of knee extensors [43].
Unfortunately, to our knowledge, no published data exist
about joint moments during stair ascent in MS. However, this
seems a plausible and interesting temptative hypothesis
deserving further future investigation.
Similarly to subjects with MS, ST subjects showed
consistent alterations of all gait pattern parameters, confirming
previous studies showing reduced cadence [19], and decreased
regularity and symmetry due to spasticity and weakness of the
paretic side [44]. Significant reduction of rhythmicity,
quantified by harmonic ratios, was also found in ST compared
to HS during level-ground walking [45]. Noteworthy, trunk
movements were comparable between ST and HS groups,
despite a trend towards increased roll and yaw sway for ST
subjects not using handrail. Moreover, trunk pitch sway was
significantly smaller compared to MS subjects. It can be
hypothesized that this result is attributable to a different
compensatory strategy used by ST subjects to ascend stair,
that implies a weight distribution mainly towards the less
affected side [19], rather than an increase of trunk
compensatory movements, as demonstrated by increased
asymmetry between legs. Indeed, Novak et al. [19] found that
knee extensor moment was reduced on the paretic side only,
with the less affected side presenting moments comparable to
healthy participants [19]. This would reflect an increase of
extensor support of non-paretic limb to improve stability and
to raise the CoM in order to facilitate the clearance of the
paretic limb’s foot during swing.
Subjects with PD showed fewer alterations of stair ascent
than MS and ST patients. PD subjects showed reduced step
frequency, confirming previous studies [3],[17] and
highlighting the impact of bradykinesia and muscle weakness
[3]. A significant decrease in medio-lateral harmonic ratio was
also found in PD compared to HS, enforcing previous results
[17]. However, this difference disappeared when subjects
using rails were excluded. A first hypothesis could be the
detrimental effect of upper limb bradykinesia and reduced
arm/leg coordination [46], that could be more pronounced
when subjects interact with the rail to ascend stairs. A second
hypothesis, could be the better dynamic balance and the lower
severity of PD subjects not using handrails, as noticed by the
higher mDGI score (+14 points) and by the larger number of
patients (3 versus 1) in the mild stage of the disease
(H&Y<=2.5). Importantly, given the ability of HR in
differentiating PD fallers and non-fallers [31], the reduced
medio-lateral HR shown by moderate-severe patients using
handrail suggests a higher risk of falls in this sub-sample. All
the other gait pattern features were comparable to those
TABLE IV
INSTRUMENTAL PARAMETERS DESCRIBING STAIR ASCENT IN HEALTHY SUBJECTS (HS, a), SUBJECTS WITH MULTIPLE SCLEROSIS (MS, b), STROKE (ST, c) AND
PARKINSON’S DISEASE (PD, d) WHO DID NOT USE HANDRAIL.
HS (a)
(N = 20)
MS (b)
(N = 4)
ST (c)
(N = 4)
PD (d)
(N = 5)
F
(p-value)
Step Frequency [step/s]
2.06 (1.92-2.20) b,c,d
1.28 (0.94-1.62) a
1.63 (1.32-1.93) a
1.68 (1.40-1.97) a
8.65
(<0.001)
Step Symmetry [unitless]
0.86 (0.91-0.91) b,c
0.57 (0.29-0.84) a,d
0.66 (0.50-0.82) a,d
0.87 (0.79-0.96) b,c
5.14
(0.006)
Stride Regularity [unitless]
0.86 (0.81-0.91) b,c
0.45 (0.04-0.86) a,d
0.68 (0.54-0.82) a,d
0.87 (0.79-0.95) b,c
6.21
(0.002)
Harmonic Ratio antero-posterior [unitless]
2.90 (2.57-3.22) b,c
1.39 (0.56-2.21) a,d
1.54 (0.88-2.19) a,d
2.86 (2.26-3.46) b,c
6.45
(0.002)
Harmonic Ratio medio-lateral [unitless]
3.12 (2.71-3.53) b,c
1.32 (0.27-2.37) a,d
2.00 (1.17-2.83) a,d
3.05 (2.28-3.81) b,c
3.97
(0.018)
Harmonic Ratio vertical [unitless]
3.45 (2.89-4.01) b,c
1.37 (0.01-2.73) a,d
1.81 (0.68-2.93) a,d
3.59 (2.57-4.63) b,c
3.92
(0.019)
Trunk Pitch Sway[deg]
1.24 (1.05-1.43) b
2.24 (1.37-3.11) a,c,d
1.36 (0.94-1.78) b
1.11 (0.80-1.43) b
3.08
(0.044)
Trunk Roll Sway [deg]
1.86 (1.33-2.38)
1.94 (0.61-3.27)
2.28 (1.23-3.34)
1.68 (0.71-2.64)
0.28
(0.839)
Trunk Yaw Sway [deg]
4.05 (3.38-4.72) d
5.37 (3.67-7.08) d
5.38 (4.03-6.73) d
3.00 (1.77-4.25) a,b,c
3.03
(0.046)
Values are mean (95% confidence interval), adjusted though ANCOVA for age, in the case of Step Frequency, and for age and Step Frequency for all the
other parameters. F (p-value): results of ANCOVA. Degrees of freedom are (3,28) for Step Frequency, and (3,27) for all the other parameters. The superscript
letters a, b, c, d indicate a statistically significant difference (p<0.05) with respect to the corresponding group (Fisher’s post-hoc test)
TNSRE-2018-00370.R1
7
characterizing HS, and higher than those related to MS and ST
subjects. Since balance and gait clinical scores were
comparable between groups, this result suggested that stair
ascent is a less challenging task for the tested PD patients in
the ON medication state compared with the other two
pathological groups. The present findings are confirmed by
Conway et al. [17], who hypothesized a beneficial effect of
visual cues provided by the horizontal edges of the steps that
could have helped PD subjects to improve their gait patterns,
thus contributing to the relatively few differences with HS.
The generally good movement pattern seems supported also
by previous results [3] showing that PD subjects exerted a net
support joint moment comparable to HS, with the exception of
lower contribution of ankle plantarflexors largely
compensated by higher knee extensors moment. Regarding
trunk, present results showed significantly smaller yaw
oscillations in PD, compared to HS and MS. This finding is in
accordance with previous studies that showed significant
reduction of trunk rotation during straight-line walking [47]
and turning [48], suggesting an “en bloc” trunk motion in PD
potentially attributable to increased axial stiffness [49].
The use of the present method expands the possibility to
assess gait pattern and trunk disorders during stair ascent, that
requires specific and personalized rehabilitation training,
mainly to improve independence in daily life activities, as
highlighted by Morone et al. [8]. Stair gait pattern features
related to MS and ST groups suggests that resistance training
could be a good way to improve muscular performance, in
particular of quadriceps, during a physically demanding
functional task such as stair ascent, possibly increasing stride
regularity and rhythmicity. Resistance and task-oriented
training seems also effective to reduce asymmetry
characterizing MS and ST subjects. In fact, Seo et al. [11]
found that patients undergoing these treatments during stair
negotiation improved weight-bearing symmetry more than
subjects performing straight-line walking. Asymmetry could
also be reduced with specific training of body-weight shift
between legs [50], with the therapist’s assistance and/or the
provision of visual/acoustic feedback to the subject. In
particular the latter approach has been proved beneficial in PD
subjects during standing and walking [51], and could be easily
used also during stair ascent. A further option is represented
by novel robotic systems allowing stair ascent [10], with the
advantage of increasing training intensity, reducing
physiotherapists’ assistance, and tuning robot’s parameters for
each subject based on the specific kinematic aspect to be
improved. Importantly, the proposed parameters include
indexes quantifying upper trunk sway, that has been found to
be abnormally larger in MS and smaller in PD subjects. Since
adequate trunk movements are crucial to maintain head
stability and correct sensory integration [40], trunk
rehabilitation is a key factor to improve dynamic balance,
especially during challenging tasks such as stair ascent [52]. In
particular, stair training should include interventions aimed at
improving trunk muscle forces, reducing trunk pitch sway in
MS subjects, and increasing trunk yaw oscillations in PD
patients, for example through core stability exercises and
electromyographic or angular biofeedback systems.
This study has some limitations. The tested sample was
small and included also handrail users, since we recruited
subjects usually attending the rehabilitation unit, who were
asked to climb stairs as they usually do in daily living. In this
context, a further limitation is that the way of using handrail
(light touch versus “heavy” use to pull themselves up the
stairs) was not recorded [53]. This did not allow to analyze if
subjects used rails mainly to increase balance control or to
unload lower limbs [53]. Despite these limitations, the present
results suggest that the use of a single MIMU provided valid
and objective parameters which revealed between-group
differences in stair ascent not detected by clinical scales.
These parameters can provide suggestions for tailored
rehabilitation and can be used to quantify its effects. Future
studies on a larger sample, analyzing also fear of falling,
should be performed to corroborate present findings and
further investigate handrail’s effects. Also the inclusion of a
second MIMU on head could provide additional information
about head stability during stair ascent. Finally, a comparison
with level-ground walking in patients with minimal disease
severity should be performed to analyze if stairway walking is
a more sensitive task to reveal sub-clinical signs.
REFERENCES
[1] M.A. Alzahrani, C.M. Dean, and L. Ada, “Ability to negotiate stairs
predicts free-living physical activity in community-dwelling people with
stroke: an observational study,” Aust. J. Physiother., vol. 55, no. 4, pp.
277-281, 2009.
[2] S. Nadeau, B.J. McFadyen, and F. Malouin, “Frontal and sagittal plane
analyses of the stair climbing task in healthy adults aged over 40 years:
what are the challenges compared to level walking?,” Clin. Biomech.,
vol. 18, no. 10, pp. 950-959, Dec. 2003.
[3] Z.J. Conway, P.A. Silburn, T. Blackmore, M.H. Cole, “Evidence of
compensatory joint kinetics during stair ascent and descent in
Parkinson’s disease,” Gait Posture, vol. 52, pp. 33-39, Feb. 2017.
[4] A.C. Tiedemann, C. Sherrington, and S.R. Lord, “Physical and
psychological factors associated with stair negotiation performance in
older people,” J. Gerontol. A Biol. Sci. Med. Sci., vol. 62, no. 11, pp.
1259-1265, Nov. 2007.
[5] J.V. Jacobs, “A review of stairway falls and stair negotiation: Lessons
learned and future needs to reduce injury,” Gait Posture, vol. 49, pp.
159-167, Sep. 2016.
[6] S.H. Corporaal, H. Gensicke, J. Kuhle, L. Kappos, J.H. Allum, and Ö
Yaldizli, “Balance control in multiple sclerosis: correlations of trunk
sway during stance and gait tests with disease severity,” Gait Posture,
vol. 37, no. 1, pp. 55-60, Jan. 2013.
[7] E. Beghi, et al., “Prediction of Falls in Subjects Suffering From
Parkinson Disease, Multiple Sclerosis, and Stroke,” Arch. Phys. Med.
Rehabil., vol. 99, no. 4, pp. 641-651, Apr. 2018.
[8] G. Morone, M. Matamala-Gomez, M.V. Sanchez-Vives, S. Paolucci,
and M. Iosa, “Watch your step! Who can recover stair climbing
independence after stroke?,” Eur. J. Phys. Rehabil. Med., ahead of print,
Aug. 2018, doi: 10.23736/S1973-9087.18.04809-8.
[9] M.B. van Iersel, M.G. Olde Rikkert, and G.P. Mulley, “Is stair
negotiation measured appropriately in functional assessment scales?,”
Clin. Rehabil., vol. 17, no. 3, pp. 325-333, May 2003.
[10] S. Hesse, C. Tomelleri, A. Bardeleben, C. Werner, and A. Waldner,
“Robot-assisted practice of gait and stair climbing in nonambulatory
stroke patients,” J. Rehabil. Res. Dev., vol. 49, no. 4, pp. 613-622, 2012.
[11] K. Seo, J. Kim, and G. Wi, “The effects of stair gait exercise on static
balance ability of stroke patients,” J. Phys. Ther. Sci. vol. 26, no. 11, pp.
1835-1838, Nov. 2014.
[12] A. Shumway-Cook, C.S. Taylor, P.N. Matsuda, M.T. Studer, and B.K.
Whetten, “Expanding the scoring system for the Dynamic Gait Index,”
Phys. Ther., vol. 93, no. 11, pp. 1493-1506, Nov. 2013.
TNSRE-2018-00370.R1
8
[13] F. Horak, L. King, and M. Mancini, “Role of body-worn movement
monitor technology for balance and gait rehabilitation,” Phys. Ther., vol.
95, no.3, pp. 461-470, Mar. 2015.
[14] P.B. Shull, W. Jirattigalachote, M.A. Hunt, M.R. Cutkosky, and S.L.
Delp, “Quantified self and human movement: a review on the clinical
impact of wearable sensing and feedback for gait analysis and
intervention,” Gait Posture, vol. 40, no. 1, pp. 11-19, 2014.
[15] M. Bertoli, et al., “Estimation of spatio-temporal parameters of gait from
magneto-inertial measurement units: multicenter validation among
Parkinson, mildly cognitively impaired and healthy older adults,”
Biomed. Eng. Online, vol. 17, no. 1, May 2018, Art. no. 58.
[16] C. Bonora, I. Carpinella, D. Cattaneo, L. Chiari, and M. Ferrarin, “A
new instrumented method for the evaluation of gait initiation and step
climbing based on inertial sensors: a pilot application in Parkinson’s
disease,” J. Neuroeng. Rehabil., vol. 12, May 2015, Art. no. 45.
[17] Z.J. Conway, T. Blackmore, P.A. Silburn, and M.H. Cole, “Dynamic
balance control during stair negotiation for older adults and people with
Parkinson disease,” Hum. Mov. Sci., vol. 59, pp. 30-36, Jun. 2018.
[18] L. Hale, K. Williams, C. Ashton, T. Connole, H. McDowell, and C.
Taylor, “Reliability of RT3 accelerometer for measuring mobility in
people with multiple sclerosis: pilot study,” J. Rehabil. Res. Dev., vol.
44, no. 4, pp. 619-627, 2007.
[19] A.C. Novak, and B. Brouwer, “Kinematic and kinetic evaluation of the
stance phase of stair ambulation in persons with stroke and healthy
adults: a pilot study,” J. Appl. Biomech., vol. 29, no 4, pp. 443-452, Aug.
2013.
[20] J.F. Kurtzke, “Rating neurologic impairment in multiple sclerosis: an
expanded disability status scale (EDSS),” Neurology, vol. 33, no. 11, pp.
1444-1452, Nov. 1983.
[21] J.C. van Swieten, P.J. Koudstaal, M.C. Visser, H.J. Schouten, and J. van
Gijn, “Interobserver agreement for the assessment of handicap in stroke
patients,” Stroke, vol. 19, no. 5, pp. 604-607, May 1988.
[22] M. Hoehn and M. D. Yahr, “Parkinsonism: Onset, progression and
mortality,” Neurology, vol. 17, no. 5, pp. 427–442, May 1967.
[23] J.E. Graham, G.V. Ostir, S.R. Fisher, and K.J. Ottenbacher, “Assessing
walking speed in clinical research: a systematic review,” J. Eval. Clin.
Pract., vol. 14, no. 4, pp. 552-562, Aug. 2008.
[24] A. Salarian, F.B. Horak, C. Zampieri, P. Carlson-Kuhta, J.G. Nutt, and
K. Aminian, “iTUG, a sensitive and reliable measure of mobility,” IEEE
Trans. Neural Syst. Rehabil. Eng., vol 18, no. 3, pp. 303-310, Jun. 2010.
[25] B. Auvinet, et al., “Reference data for normal subjects obtained with an
accelerometric device,” Gait Posture. vol. 16, no. 2, pp. 124-134, Oct.
2002.
[26] A. Tura, M. Raggi, L. Rocchi, A.G. Cutti, and L. Chiari, “Gait symmetry
and regularity in transfemoral amputees assessed by trunk
accelerations,” J. Neuroeng. Rehabil. vol. 7, Jan. 2010, Art. no. 4.
[27] H.B. Menz, S.R. Lord, and R.C. Fitzpatrick, “Acceleration patterns of
the head and pelvis when walking on level and irregular surfaces,” Gait
Posture. vol. 18, no. 1, pp. 35-46, Aug. 2003.
[28] M. Pau, et al., “Smoothness of gait detects early alterations of walking
in persons with multiple sclerosis without disability,” Gait Posture., vol.
58, pp. 307-309, Oct. 2017.
[29] J.S. Brach, et al., “Validation of a measure of smoothness of walking,” J.
Gerontol. A Biol. Sci. Med. Sci., vol. 66, no. 1, pp: 136-141, Jan. 2011.
[30] T. Doi, S. Hirata, R. Ono, K. Tsutsumimoto, S. Misu, and H. Ando,
“The harmonic ratio of trunk acceleration predicts falling among older
people: results of a 1-year prospective study,” J. Neuroeng. Rehabil.,
vol. 10, Jan. 2013, Art. no. 7.
[31] M.H. Cole, M. Sweeney, Z.J. Conway, T. Blackmore, and P.A. Silburn,
“Imposed Faster and Slower Walking Speeds Influence Gait Stability
Differently in Parkinson Fallers,” Arch. Phys. Med. Rehabil., vol. 94, no.
4, pp. 639-648, Apr. 2017.
[32] F. Riva, M.J. Toebes, M. Pijnappels, R. Stagni, and J.H. van Dieën,
“Estimating fall risk with inertial sensors using gait stability measures
that do not require step detection,” Gait Posture, vol. 38, no. 2, pp. 170-
174, Jun. 2013.
[33] D. Cattaneo, I. Carpinella, I. Aprile, L. Prosperini, A. Montesano, and J.
Jonsdottir, “Comparison of upright balance in stroke, Parkinson and
multiple sclerosis,” Acta Neurol. Scand., vol. 133, no. 5, pp. 346-354,
May 2016.
[34] U. Givon, G. Zeilig, and A. Achiron, “Gait analysis in multiple sclerosis:
characterization of temporal-spatial parameters using GAITRite
functional ambulation system,” Gait Posture, vol. 29, no. 1, pp. 138-
142, Jun. 2009.
[35] A. Kalron, “Association between perceived fatigue and gait parameters
measured by an instrumented treadmill in people with multiple sclerosis:
a cross-sectional study,” J. Neuroeng. Rehabil., vol. 12, Apr. 2015, Art.
no. 34.
[36] L.H. Chung, J.G. Remelius, R.E. Van Emmerik, and J.A. Kent-Braun,
“Leg power asymmetry and postural control in women with multiple
sclerosis,” Med. Sci. Sports Exerc., vol. 40, no. 10, pp. 1717-1724, Oct.
2008.
[37] M.J. Socie and J.J. Sosnoff, “Gait variability and multiple sclerosis,”
Mult. Scler. Int., vol. 2013, 2013, Art. no. 645197.
[38] M.H. Cameron, F.B. Horak, R.R. Herndon, and D. Bourdette,
“Imbalance in multiple sclerosis: a result of slowed spinal
somatosensory conduction,” Somatosens. Mot. Res., vol. 25, no. 2, pp.
113-122, 2008.
[39] D. Cattaneo, M. Ferrarin, W. Frasson, and A. Casiraghi, “Head control:
volitional aspects of rehabilitation training in patients with multiple
sclerosis compared with healthy subjects,” Arch. Phys. Med. Rehabil.,
vol. 86, no. 7, pp. 1381-1388, Jul. 2005.
[40] J. Kavanagh, R. Barrett, and S. Morrison, “The role of the neck and
trunk in facilitating head stability during walking,” Exp. Brain Res., vol.
172, no. 4, pp. 454-463, Jul. 2006.
[41] R. Cromwell and R. Wellmon, “Sagittal plane head stabilization during
level walking and ambulation on stairs,” Physiother. Res. Int., vol. 6, no.
3, pp. 179-192, 2001.
[42] S. Güner, S. Haghari, F. Inanici, S. Alsancak, and G. Aytekin, “Knee
muscle strength in multiple sclerosis: relationship with gait
characteristics,” J. Phys. Ther. Sci., vol. 27, no. 3, pp. 809-813, Mar.
2015.
[43] J.L. Asay, A. Mündermann, and T.P. Andriacchi, “Adaptive patterns of
movement during stair climbing in patients with knee osteoarthritis,” J.
Orthop. Res., vol. 27, no. 3, pp. 325-329, Mar. 2009.
[44] C.M. Kim, and J.J. Eng, “The relationship of lower-extremity muscle
torque to locomotor performance in people with stroke,” Phys. Ther.,
vol. 83, no. 1, pp. 49-57, Jan. 2003.
[45] M. Iosa, et al., “Stability and Harmony of Gait in Patients with Subacute
Stroke,” J. Med. Biol. Eng., vol. 36, no. 5, pp. 635-643, 2016.
[46] P. Crenna, et al., “Influence of basal ganglia on upper limb locomotor
synergies. Evidence from deep brain stimulation and L-DOPA treatment
in Parkinson’s disease,” Brain, vol. 131, pt. 12, pp. 3410-3420, Dec.
2008.
[47] M. Ferrarin, M. Rizzone, L. Lopiano, M. Recalcati, and A. Pedotti,
“Effects of subthalamic nucleus stimulation and L-dopa in trunk
kinematics of patients with Parkinson’s disease,” Gait Posture, vol. 19,
no. 2, pp. 164-171, Apr. 2004.
[48] I. Carpinella, et al., “Locomotor function in the early stage of
Parkinson’s disease,” IEEE Trans. Neural Syst. Rehabil. Eng., vol. 15,
no. 4, pp. 543-551, Dec. 2007.
[49] R.E. Van Emmerik, R.C. Wagenaar, A. Winogrodzka, and E.C. Wolters,
“Identification of axial rigidity during locomotion in Parkinson disease,”
Arch. Phys. Med. Rehabil., vol. 80, no. 2, pp. 186-191, Feb. 1999.
[50] I. Carpinella, et al. “Counteracting postural perturbations through body
weight shift: a pilot study using a robotic platform in subjects with
Parkinson’s disease,” IEEE Trans. Neural Syst. Rehabil. Eng., vol. 26,
no. 9, pp. 1794-1802, Sep. 2018.
[51] I. Carpinella, et al., “Wearable sensor-based biofeedback training for
balance and gait in Parkinson disease: a pilot randomized controlled
trial,” Arch. Phys. Med. Rehabil., vol. 98, no. 4, pp. 622-630.e3, Apr.
2017.
[52] R.P. Hubble, G.A. Naughton, P.A. Silburn, and M.H. Cole, “Trunk
muscle exercises as a means of improving postural stability in people
with Parkinson’s disease: a protocol for a randomised controlled trial,”
BMJ Open, vol. 4, no.12, Dec. 2014, Art. no. e006095.
[53] N.D. Reeves, M. Spanjaard, A.A. Mohagheghi, V. Baltzopoulos, and
C.N. Maganaris, “Influence of light handrail use on the biomechanics of
stair negotiation in old age,” Gait Posture, vol. 28, no. 2, pp. 327-336,
Aug. 2008.