Skip to main content

Table 5 List of included studies and data extracted from each article

From: Utilization of wearable technology to assess gait and mobility post-stroke: a systematic review

Article

Wearable technology (Brand)

Location of wearable on body

Gait variables or parameters

Main Findings for Primary and Secondary Outcomes

Reliability/validity process and metrics

Dorsch et al. [51]

Triaxial accelerometers (Gulf Coast Data Concepts)

Bilateral ankles

Average daily walking time (min)

Walking speed (m/s)

Feedback on 10-m walk speed plus a review of sensor-derived walking activity did not improve walking outcomes more than walking speed feedback alone [Primary: average daily walking time (p = 0.54) and 15-m walk speed (p = 0.96); Secondary: FAC scores (p = 0.39) and 3-min walking distance (p = 0.98)]

Primary: No difference between groups in the rate of change in time spent walking (p = 0.32)

Process: To determine the correlation between sensor-derived walking speeds and clinical measures of walking speed

Sensor-derived average daily walking speeds were highly correlated with 10-m walk speeds (r = 0.977, p < .001)

Sensor-derived maximum daily speed was moderately correlated with 15-m walk speeds (r = 0.647, p < 0.001)

Mansfield et al. [36]

Triaxial accelerometers (Model X6-2mini, Gulf Coast Data Concepts)

Bilateral limbs

Walking time (min)

Number of steps

Average cadence (steps/min)

Primary: There was no greater increase in daily walking activity (i.e. total walking duration, number of steps) for individuals whose physiotherapists provided accelerometer-based feedback compared to those who received no feedback (p > 0.20)

Secondary: Average cadence of daily walking did improve with feedback (p = 0.013)—interpreted to mean that daily walking was faster (i.e. more intense) when feedback was provided

NR

English et al. [52]

Triaxial accelerometers (activPAL3 and Actigraph GT3+)

Non-paretic hip

Stepping time (min/d)

Moderate-to-vigorous physical activity (MVPA) (min/d)

Primary: This intervention was both safe and feasible

Secondary: Daily siting time and prolonged sit times were reduced on average for both groups, and time spent standing and stepping increased on average; no within- or between-group effects were statistically significant

Average MVPA remained very low for all participants at baseline and post-intervention

NR

Givon et al. [53]

Accelerometer (Acticial Minimitter Co.)

Hip

Steps/day

Primary: Video game intervention is feasible in a community group setting

Gait speed significantly improved in each group (p = 0.04)

Secondary: There was no significant change in daily steps walked as assessed by accelerometers in either group

NR

Danks et al. [54]

StepWatch Activity Monitor (Orthocare Innovations)

Non-paretic ankle

Steps/day

Total walking time (h)

Self-selected walking speed (m/s)

Max walking speed (m/s)

Primary: A significant effect of time was observed in both groups for steps per day, total time walking, self-selected and maximal walking speed, and 6MWT distance (all p < 0.05). Subjects in the FAST + SAM group exhibited a larger increase in 6MWT distance compared to the FAST only group (p = 0.018)

Results suggest that subjects with low baseline levels of walking and long-distance walking will show greater benefit when a step activity monitoring program is used in conjunction with an intervention designed to increase walking capacity

NR

Kanai et al. [55]

Fitbit One three-dimensional accelerometer (Fitbit Inc.)

Wrist

Steps/day

Duration of activity (min/day)

Primary: Number of steps/day in the intervention group (i.e. use of accelerometer-based feedback plus supervised rehab) at follow-up were higher compared to the control group (supervised rehab only) (p < 0.001)

Secondary: Exercise energy expenditure and duration of activity were also higher in the intervention group at follow-up compared to the control group (p ≤ 0.001)

Results indicate that accelerometer-based feedback may increase physical activity, exercise energy expenditure and the duration of activity time in hospitalized patients with ischemic stroke

NR

Prajapati et al. [56]

The ABLE system (accelerometer for bilateral lower extremities): comprised 2 commercial triaxial accelerometers (Sparkfun Electronics)

Waist and bilateral ankles

Steps

Cadence

Number/mean of walking bouts

Total walking time

Total structured walking time

Swing symmetry Temporal gait symmetry

Primary: On average, patients exhibited 47.5 (± 26.6) minutes of total walking time and walking duration bouts of 54.4 (± 21.5) secs during an inpatient day

Secondary: A significant association was observed between the number of walking bouts and 1) total walking time (r = 0.76; p < 0.006) and 2) lab gait speed (r = 0.51; p < 0.045); and 2), as well as between slower lab gait speed and lower BBS score (r = 0.60; p < 0.013)

Patients were highly variable with respect to their frequency and duration of walking activity

Process: To compare laboratory-based gait symmetry measures with wireless accelerometer-based measures of symmetry

A significant difference was found between wireless accelerometer-based swing symmetry measures and lab-based measures (p = 0.006); 12 of 16 patients were more asymmetrical during the course of the day (i.e. as measured by wireless sensors

Taraldsen et al. [42]

ActivPAL single-axis accelerometer (PAL Technologies)

Sternum and bilateral thighs

Gait speed (m/s)

Number of steps

Primary: Results indicate that the ActivPAL algorithms can accurately classify postures and transitions, but are not effective at detecting slow stepping. The step count algorithm is not acceptable for slow walking speeds (≤ 0.47 m/s) and needs to be improved before the ActivPAL system can be recommended for use in people who are frail

Secondary: Placement of the sensor on the nonaffected leg led to less underestimation of step counts than placement on the affected leg

Process: To evaluate the concurrent validity of the ActivPAL sensor system against video observations (main objective of study)

Tramon-tano et al. [57]

Triaxial wireless accelerometer

Lumbar spine

Walking speed (m/s)

Trunk acceleration

Primary: Persons with stroke demonstrated slower walking speeds than healthy adults when asked to dual task while walking (p = 0.005). There were no significant differences between groups in terms of trunk acceleration (p > 0.05); however, when controlling for walking speed, trunk acceleration was significantly different (p < 0.05), with persons with stroke exhibiting higher trunk accelerations

Differences in walking speeds between the two groups was attributed to persons with stroke walking slower in hopes of trying to control abnormal trunk accelerations

Secondary: A quadratic relationship between BBS score and changes in trunk acceleration RMS along the cranio-caudal axis was observed (p = 0.044)

NR

Wang et al. [58]

Textile capacitive pressure sensing insole (TCPSI) (Ajin Electronics)

Insole of shoes

Percentage of plantar pressure difference (PPD), step count, stride time, coefficient of variation, and phase coordination index (PCI)

Primary: Textile capacitive pressure sensing insoles were successfully used to analyze hemiparetic gait patterns and distinguish them from normal gait characteristics

During a 40-m walk, patients with stroke had 3 × higher plantar pressure difference, lower mean plantar pressure on the affected side, a higher step count, longer stride time on the affected side, and 3 × higher PCI (indicating less balance between feet) compared to healthy controls

NR

Seo et al. [59]

Smart insole sensor

Insole of shoes

TUG, walking speed, stride length, walking time, single support time, double support time, and differences in swing and stance duration

Primary: Smart insole sensor data were similar to those calculated manually during the TUG assessment

Significant differences in walking speed, stride length, TUG time, walking time, single support time, double support time, and differences in swing and stance duration were found between patients with stroke and healthy controls (p ≤ 0.005)

Secondary: FMA score was significantly correlated with smart insole data (p ≤ 0.02)

NR

Paul et al. [60]

ActivPAL activity monitor

(PAL Technologies)

Phone accelerometer (Samsung Galaxy SIII)

ActivPAL on non-paretic leg

Steps/day

Walking time (h)

The STARFISH app includes the behavior change techniques of goal setting, planning, monitoring, and feedback as well as rewards and social facilitation

Primary: Using the STARFISH app for six weeks led to a significant increase in physical activity (i.e. mean number of steps/day and walking time) compared to a usual care control group (p ≤ 0.005)

Secondary: Post-stroke fatigue reduced in the intervention group and increased in the control group (p = 0.003)

Process: To determine the correlation between phone accelerometer-based step counts and ActivPAL step counts

A moderate correlation was found between step count data from the phone accelerometer and the ActivPAL (r = 0.67); however, at slower walking speeds the reliability of accelerometers in detecting steps is reduced

Shin et al. [61]

IMU motion sensors (XSens)

Pelvis and bilateral thighs, shanks, and feet

Amount of motion (AoM)

Gait speed (m/s) Step number

Primary: Longitudinally recording joint kinematics during early gait rehabilitation post-stroke is feasible

Total AoM (i.e. sum of all individual joint displacements measured), step number, number of different tasks performed during therapy, treatment intensity (i.e. change in HR), and time post-stroke were all significantly correlated with gait speed (p < 0.01, except HR p < 0.05), with total AoM revealing the greatest explained variance (R2 = 32.1%)

NR

  1. NR not reported, MVPA moderate-vigorous physical activity, 6MWT 6 min Walk Test, FAST fast walking training, SAM StepWatch activity monitor, BBS Berg Balance Scale, RMS root mean square, TUG Timed Up & Go, FMA Fugl-Meyer assessment, IMU inertial measurement unit, HR heart rate