Development and validation of a new method to measure walking speed in free-living environments using the actibelt® platform

Real-Time Tracking Data and Machine Learning Approaches for Mapping Pedestrian Walking Behavior: A Case Study at the University of Moratuwa

The growing urban population and traffic congestion underline the importance of building pedestrian-friendly environments to encourage walking as a preferred mode of transportation. However, a major challenge remains, which is the absence of such pedestrian-friendly walking environments. Identifying locations and routes with high pedestrian concentration is critical for improving pedestrian-friendly walking environments. This paper presents a quantitative method to map pedestrian walking behavior by utilizing real-time data from mobile phone sensors, focusing on the University of Moratuwa, Sri Lanka, as a case study. This holistic method integrates new urban data, such as location-based service (LBS) positioning data, and data clustering with unsupervised machine learning techniques. This study focused on the following three criteria for quantifying walking behavior: walking speed, walking time, and walking direction inside the experimental research context. A novel signal processing method has been used to evaluate speed signals, resulting in the identification of 622 speed clusters using K-means clustering techniques during specific morning and evening hours. This project uses mobile GPS signals and machine learning algorithms to track and classify pedestrian walking activity in crucial sites and routes, potentially improving urban walking through mapping.

Changes in real-world walking speed following 60-day bed-rest

The aim of this work was to explore whether real-world walking speed (RWS) would change as a consequence of 60-day bed-rest. The main hypothesis was that daily RWS would decrease after the bed-rest, with a subsequent recovery during the first days of re-ambulation. Moreover, an exploratory analysis was done in order to understand whether there is an agreement between the loss in RWS after bed-rest and the loss in the maximum oxygen uptake capacity (VO2max), or the loss in maximal vertical jump power (JUMP) respectively. Twenty-four subjects were randomly assigned to one of three groups: a continuous artificial gravity group, an intermittent artificial gravity group, or a control group. The fitted linear mixed effects model showed a significant decrease (p < 0.001) of RWS after the 60-day bed-rest and a subsequent increase (p < 0.001) of RWS during the 14-day recovery period in the study facility. No or little agreement was found between the loss in RWS and the loss in VO2max capacity or the loss in maximal vertical jumping power (RWS vs. VO2max: p = 0.81, RWS vs. JUMP: p = 0.173). Decreased RWS after bed-rest, with a follow-up recovery was observed for all three groups, regardless of the training intervention. This suggests that RWS, also in these settings, was able to reflect a de-conditioning and follow-up recovery process.

Validity and reliability of an app-based medical device to empower individuals in evaluating their physical capacities

BACKGROUND

Cardiorespiratory fitness and muscle strength are valid markers of health and strong predictors of mortality and morbidity. The tests used to measure these variables require in-person visits with specialized equipment and trained personnel-leading to organizational constraints both for patients and hospitals, and making them difficult to implement at a large scale. In this context, technologies embedded in smartphones offer new opportunities to develop remote tests.

OBJECTIVES

This study aimed to test the validity and reliability of MediEval, a newly developed app-based medical device that allows individuals to perform the 6-minute walk test (6MWT) and the 30-second sit-to-stand (30s-STS) test on their own using GPS signal and camera detection with a skeleton extraction algorithm.

METHODS

A total of 53 healthy adults performed the two tests in three different sessions to determine the intra- and inter-day reproducibility. Test validity was assessed by comparing the results obtained from the app to gold standard measures. Pearson correlations and concordance correlation coefficients, the relative measurement error, intraclass correlation coefficients, the standard error of measure and the minimal detectable change were computed for each test.s.

RESULTS

The results revealed high to excellent validity of the app in comparison to gold standards (ρ = 0.84 for the 6MWT and ρ = 0.99 for the 30s-STS test) with low relative measurement error. The mean differences between the app and the gold standard measures were 8.96m for the 6MWT and 0.28 repetition for the 30s-STS test. Both tests had good test-retest reliability (ICCs = 0.77). The minimal detectable changes were respectively 97.56 meters for the 6MWT and 7.37 repetitions for the 30s-STS test.

CONCLUSION

The MediEval medical device proposes valid and reproducible measures of the 6MWT and the 30s-STS test. This device holds promise for monitoring the physical activity of large epidemiologic cohorts while refining patient experience and improving the scalability of the healthcare system. Considering minimal detectable change values, it may be important to ask participants to perform several tests and average them to improve accuracy. Future studies in clinical context are needed to evaluate the responsiveness and the smallest detectable changes of the device for specific populations with chronic diseases.

Association between daily gait speed patterns and cognitive impairment in community-dwelling older adults

Gait speed over a short distance is associated with cognitive impairment in older adults. Recently, daily gait speed has been assessed using accelerometers. However, because daily gait speed is only weakly correlation with gait speed over a short distance, its association with cognitive impairment needs to be investigated. The present study compared the daily gait speed patterns of normal cognition (NC), mild cognitive impairment (MCI), and general cognitive impairment (GCI) subjects measured every 3 h for two weeks using accelerometers. A total of 1959 participants were classified into the NC (N = 1519), MCI (N = 353), and GCI groups (N = 87). The results showed that the average daily gait speed of the GCI group was significantly lower than that of the NC group (p = 0.03). Furthermore, the average daily gait speeds of the MCI and NC groups were the same. However, the average daily gait speed of the MCI group during a specific time (12-15 o'clock) was significantly lower than that of the NC group (p < 0.01). These results suggest that changes in daily patterns may be detected by measuring daily gait speed, which depends on the degree of cognitive function.

Association between daily life walking speed and frailty measured by a smartphone application: a cross-sectional study

OBJECTIVES

To assess whether frailty can be assessed using a smartphone and whether daily walking speed (DWS) is associated with frailty.

DESIGN

Cross-sectional study.

SETTING

Three prefectures (Kanagawa, Saitama and Tokyo) in Japan.

PARTICIPANTS

The study enrolled 163 participants (65 in the robust group, 69 in the prefrailty group and 29 in the frailty group) by sending letters to house owners aged≥55 years.

PRIMARY AND SECONDARY OUTCOME MEASURES

The participants downloaded the DWS measurement application on their smartphones, which measured the daily walking (DW) parameters (DWS, step length and cadence) and the Kihon checklist for frailty assessment. The differences in the DW parameters between the robust, prefrailty and frailty groups were examined using one-way analysis of variance. We conducted logistic regression analysis for the Crude model (each DW parameter), model 1 (adjusted for the number of steps) and model 2 (model 1+age, sex and the number of chronic diseases).

RESULTS

DWS was marginally significantly slower in the frailty group than in the prefrailty and robust group (robust 1.26 m/s vs prefrailty 1.25 m/s vs frailty 1.19 m/s, p=0.060). Step length was significantly smaller in the frailty group than in the robust group (robust 66.1 cm vs prefrailty 65.9 vs frailty 62.3 cm, p<0.01). Logistic regression analysis for the three models revealed that DWS was significantly associated with frailty.

CONCLUSIONS

DWS measured using the smartphone application was associated with frailty. This was probably due to the shorter step length and body height seen in frail individuals.

Predictivity of daily gait speed using tri-axial accelerometers for two-year incident disability among Japanese older adults

Gait speed is an important indicator of functional decline in older adults. Recently, daily gait speed has been assessed using accelerometers. However, it is unclear whether this parameter can predict the decline in functional abilities. This study investigates whether daily gait speed can be a predictor of incident disability risk as well as in-laboratory gait speed. A sample of 1860 older adults (Male: 728, Female: 1132; 70.1 ± 6.2 years) were instructed to wear accelerometers on the waist. The association between daily gait speed for two weeks and incident disability during a two-year period was analyzed by using the cut-off value for screening prefrailty in the previous study (106.3 cm/s). Furthermore, the associations with in-laboratory gait speed (cut-off value: 100 cm/s), number of steps (cut-off value: 6342.2 steps/day), and incident disability were also analyzed. Cox proportional hazards analysis showed a significant hazard ratio of low daily gait speed (HR, 2.97; p = 0.02) comparable to that of low in-laboratory gait speed (HR: 2.53; p = 0.01). Conversely, the number of steps had no significant association with incident disability (HR: 1.99; p = 0.12). These results suggest that daily gait speed can be a predictor of incident disability risk in older adults.

Deep Learning Methods for Speed Estimation of Bipedal Motion from Wearable IMU Sensors

The estimation of the speed of human motion from wearable IMU sensors is required in applications such as pedestrian dead reckoning. In this paper, we test deep learning methods for the prediction of the motion speed from raw readings of a low-cost IMU sensor. Each subject was observed using three sensors at the shoe, shin, and thigh. We show that existing general-purpose architectures outperform classical feature-based approaches and propose a novel architecture tailored for this task. The proposed architecture is based on a semi-supervised variational auto-encoder structure with innovated decoder in the form of a dense layer with a sinusoidal activation function. The proposed architecture achieved the lowest average error on the test data. Analysis of sensor placement reveals that the best location for the sensor is the shoe. Significant accuracy gain was observed when all three sensors were available. All data acquired in this experiment and the code of the estimation methods are available for download.

Estimate of gait speed by using persons' walk ratio or step-frequency in older adults

BACKGROUND AND AIMS

Gait speed estimation using wearable inertial sensors during daily activities suffers from high complexity and inaccuracies in distance estimation when integrating acceleration signals. The aim of the study was to investigate the agreement between the methods of gait speed estimation using the persons' walk ratio (step-length/step-frequency relation) or step-frequency (number of steps per minute) and a "gold standard".

METHODS

For this cross-sectional validation study, 20 healthy community-dwelling older persons (mean age 72.1 years; 70% women) walked at slow, normal, and fast speed over an instrumented walkway (reference measure). Gait speed was calculated using the person's pre-assessed walk ratio. Furthermore, the duration of walking and number of steps were used for calculation.

RESULTS

The agreement between gait speed calculation using the walk ratio or step-frequency (adjusted to body height) and reference was r = 0.98 and r = 0.93, respectively. Absolute and relative mean errors of calculated gait speed using pre-assessed walk ratio ranged between 0.03-0.07 m/s and 1.97-4.17%, respectively.

DISCUSSION AND CONCLUSIONS

After confirmation in larger cohorts of healthy community-dwelling older adults, the mean gait speed of single walking bouts during activity monitoring can be estimated using the person's pre-assessed walk ratio. Furthermore, the mean gait speed can be calculated using the step-frequency and body height and can be an additional parameter in stand-alone activity monitoring.

Mechanistic modelling of COVID-19 and the impact of lockdowns on a short-time scale

BACKGROUND

To mitigate the spread of the COVID-19 coronavirus, some countries have adopted more stringent non-pharmaceutical interventions in contrast to those widely used. In addition to standard practices such as enforcing curfews, social distancing, and closure of non-essential service industries, other non-conventional policies also have been implemented, such as the total lockdown of fragmented regions, which are composed of sparsely and highly populated areas.

METHODS

In this paper, we model the movement of a host population using a mechanistic approach based on random walks, which are either diffusive or super-diffusive. Infections are realised through a contact process, whereby a susceptible host is infected if in close spatial proximity of the infectious host with an assigned transmission probability. Our focus is on a short-time scale (∼ 3 days), which is the average time lag time before an infected individual becomes infectious.

RESULTS

We find that the level of infection remains approximately constant with an increase in population diffusion, and also in the case of faster population dispersal (super-diffusion). Moreover, we demonstrate how the efficacy of imposing a lockdown depends heavily on how susceptible and infectious individuals are distributed over space.

CONCLUSION

Our results indicate that on a short-time scale, the type of movement behaviour does not play an important role in rising infection levels. Also, lock-down restrictions are ineffective if the population distribution is homogeneous. However, in the case of a heterogeneous population, lockdowns are effective if a large proportion of infectious carriers are distributed in sparsely populated sub-regions.

Evaluation of the Intel RealSense T265 for tracking natural human head motion

Accurate and robust tracking of natural human head motion in natural environments is important for a number of applications including virtual and augmented reality, clinical diagnostics, as well as basic scientific research. IMU provide a versatile solution for recording inertial data including linear acceleration and angular velocity, but reconstructing head position is difficult or impossible. This problem can be solved by incorporating visual data using a technique known as visual-inertial simultaneous localization and mapping (VI-SLAM). A recently released commercial solution, the Intel RealSense T265, uses a proprietary VI-SLAM algorithm to estimate linear and angular position and velocity, but the performance of this device for tracking of natural human head motion in natural environments has not yet been comprehensively evaluated against gold-standard methods. In this study, we used a wide range of metrics to evaluate the performance of the T265 with different walking speeds in different environments, both indoor and outdoor, against two gold-standard methods, an optical tracking system and a so-called perambulator. Overall, we find that performance of the T265 relative to these gold-standard methods is most accurate for slow to normal walking speeds in small- to medium-sized environments. The suitability of this device for future scientific studies depends on the application; data presented here can be useful in making that determination.

Toward a remote assessment of walking bout and speed: application in patients with multiple sclerosis

Gait speed as a powerful biomarker of mobility is mostly assessed in controlled environments, e.g. in the clinic. With wearable inertial sensors, gait speed can be estimated in an objective manner. However, most of the previous works have validated the gait speed estimation algorithms in clinical settings which can be different than the home assessments in which the patients demonstrate their actual performance. Moreover, to provide comfort for the users, devising an algorithm based on a single sensor setup is essential. To this end, the goal of this study was to develop and validate a new gait speed estimation method based on a machine learning approach to predict gait speed in both clinical and home assessments by a sensor on the lower back. Moreover, two methods were introduced to detect walking bouts during daily activities at home. We have validated the algorithms in 35 patients with multiple sclerosis as it often presents with mobility difficulties. Therefore, the robustness of the algorithm can be shown in an impaired or slow gait. Against silver standard multi-sensor references, we achieved a bias close to zero and a precision of 0.15 m/s for gait speed estimation. Furthermore, the proposed machine learning-based locomotion detection method had a median of 96.8% specificity, 93.0% sensitivity, 96.4% accuracy, and 78.6% F1-score in detecting walking bouts at home. The high performance of the proposed algorithm showed the feasibility of the unsupervised mobility assessment introduced in this study.

Outdoor Walking Speeds of Apparently Healthy Adults: A Systematic Review and Meta-analysis

BACKGROUND

Walking outdoors can be used by many individuals to meet public health guidelines for moderate-to-vigorous-intensity physical activity. The speed at which adults walk may be a proxy for intensity. Traditional estimates of indoor walking speed are unlikely to reflect self-selected usual or other instructed paces of outdoor walking speed.

OBJECTIVE

To inform estimates of pace-based walking speed of apparently healthy adults in outdoor settings.

METHODS

We searched four electronic databases for articles published in English between January 1970 and March 2019. Studies that reported walking speed (m/s), cadence (steps/min), or intensity (mL/kg/min) of ambulatory, apparently healthy, and community-dwelling adults (> 18 years) were included. Walking speed categories were defined according to the description provided in each study. Meta-analysis was used to synthesise speed, cadence, and intensity data by slow, usual, medium, fast, and maximal pace (where reported).

RESULTS

Thirty-five studies, representing 14,015 participants (6808 women, 5135 men, and 2072 sex not specified), were identified. The mean (95% CI) walking speed for slow, usual, medium, fast, and maximal pace was 0.82 (0.77-0.86), 1.31 (1.27-1.35), 1.47 (1.44-1.49), 1.72 (1.64-1.81), and 1.62 (1.45-1.79) m/s, respectively. Mean cadence (95% CI) for usual and fast paces were 116.65 (114.95-118.35) and 126.75 (121.87-131.63) steps/min, respectively. The mean oxygen consumption (95% CI) for the usual and medium paces was 11.97 (11.69-12.25) and 13.34 (12.94-13.73) mL/kg/min, respectively.

CONCLUSION

These findings provide greater clarity with regard to how various indicators of enacted walking pace, speed, and intensity overlap and how each can be best communicated in the real-world setting to optimise health-related outcomes. Pace-based instructions can be used to support walking in outdoor settings within public health guidelines.

Robust Step Detection from Different Waist-Worn Sensor Positions: Implications for Clinical Studies

Analyzing human gait with inertial sensors provides valuable insights into a wide range of health impairments, including many musculoskeletal and neurological diseases. A representative and reliable assessment of gait requires continuous monitoring over long periods and ideally takes place in the subjects' habitual environment (real-world). An inconsistent sensor wearing position can affect gait characterization and influence clinical study results, thus clinical study protocols are typically highly proscriptive, instructing all participants to wear the sensor in a uniform manner. This restrictive approach improves data quality but reduces overall adherence. In this work, we analyze the impact of altering the sensor wearing position around the waist on sensor signal and step detection. We demonstrate that an asymmetrically worn sensor leads to additional odd-harmonic frequency components in the frequency spectrum. We propose a robust solution for step detection based on autocorrelation to overcome sensor position variation (sensitivity = 0.99, precision = 0.99). The proposed solution reduces the impact of inconsistent sensor positioning on gait characterization in clinical studies, thus providing more flexibility to protocol implementation and more freedom to participants to wear the sensor in the position most comfortable to them. This work is a first step towards truly position-agnostic gait assessment in clinical settings.

Estimating Walking Speed in the Wild

An individual's physical activity substantially impacts the potential for prevention and recovery from diverse health issues, including cardiovascular diseases. Precise quantification of a patient's level of day-to-day physical activity, which can be characterized by the type, intensity, and duration of movement, is crucial for clinicians. Walking is a primary and fundamental physical activity for most individuals. Walking speed has been shown to correlate with various heart pathologies and overall function. As such, it is often used as a metric to assess health performance. A range of clinical walking tests exist to evaluate gait and inform clinical decision-making. However, these assessments are often short, provide qualitative movement assessments, and are performed in a clinical setting that is not representative of the real-world. Technological advancements in wearable sensing and associated algorithms enable new opportunities to complement in-clinic evaluations of movement during free-living. However, the use of wearable devices to inform clinical decisions presents several challenges, including lack of subject compliance and limited sensor battery life. To bridge the gap between free-living and clinical environments, we propose an approach in which we utilize different wearable sensors at different temporal scales and resolutions. Here, we present a method to accurately estimate gait speed in the free-living environment from a low-power, lightweight accelerometer-based bio-logging tag secured on the thigh. We use high-resolution measurements of gait kinematics to build subject-specific data-driven models to accurately map stride frequencies extracted from the bio-logging system to stride speeds. The model-based estimates of stride speed were evaluated using a long outdoor walk and compared to stride parameters calculated from a foot-worn inertial measurement unit using the zero-velocity update algorithm. The proposed method presents an average concordance correlation coefficient of 0.80 for all subjects, and 97% of the error is within ±0.2m· s -1. The approach presented here provides promising results that can enable clinicians to complement their existing assessments of activity level and fitness with measurements of movement duration and intensity (walking speed) extracted at a week time scale and in the patients' free-living environment.

Smartphone Accelerometry: A Smart and Reliable Measurement of Real-Life Physical Activity in Multiple Sclerosis and Healthy Individuals

Background: Mobility impairment is common in persons with multiple sclerosis (pwMS) and can be assessed with clinical tests and surveys that have restricted ecological validity. Commercial research-based accelerometers are considered to be more valuable as they measure real-life mobility. Smartphone accelerometry might be an easily accessible alternative.

Objective: To explore smartphone accelerometry in comparison to clinical tests, surveys, and a wrist-worn ActiGraph in pwMS and controls.

Methods: Sixty-seven pwMS and 70 matched controls underwent mobility tests and surveys. Real-life data were collected with a smartphone and an ActiGraph over 7 days. We explored different smartphone metrics in a technical validation course and computed afterward correlation between ActiGraph (steps per minute), smartphone accelerometry (variance of vector magnitude), clinical tests, and surveys. We also determined the ability to separate between patients and controls as well as between different disability groups.

Results: Based on the technical validation, we found the variance of the vector magnitude as a reliable estimate to discriminate wear time and no wear-time of the smartphone. Due to a further association with different activity levels, it was selected for real-life analyses. In the cross-sectional study, ActiGraph correlated moderately (r = 0.43, p < 0.05) with the smartphone but less with clinical tests (rho between |0.211| and |0.337|). Smartphone data showed stronger correlations with age (rho = −0.487) and clinical tests (rho between |0.565| and |0.605|). ActiGraph only differed between pwMS and controls (p < 0.001) but not between disability groups. At the same time, the smartphone showed differences between pwMS and controls, between RRMS and PP-/SPMS, and between participants with/without ambulatory impairment (all p < 0.001).

Conclusions: Smartphone accelerometry provides better estimates of mobility and disability than a wrist-worn standard accelerometer in a free-living context for both controls and pwMS. Given the fact that no additional device is needed, smartphone accelerometry might be a convenient outcome of real-life ambulation in healthy individuals and chronic diseases such as MS.

Association between Daily Living Walking Speed and Walking Speed in Laboratory Settings in Healthy Older Adults

Although there is evidence on the predictors of adverse health outcomes in older individuals, walking speed has typically been measured in laboratory settings (LWS); LWS may be distinct from individuals' actual walking speed in their daily lives (DWS). We examined whether DWS differs from LWS among older adults, and its association with physical frailty. Participants were 90 community-dwelling older adults. A five-meter normal (LWS_nor) and maximum (LWS_max) walking speed was measured with a stopwatch. DWS was measured using a global positioning system-related smartphone application for one month during their daily lives. DWS_avr, DWS_max, and DWS_sd were defined as the average, maximum, and standard deviation of walking speed for one month. Participants' mean DWS_avr and DWS_max were 1.28 m/s and 2.14 m/s, respectively, significantly slower than the mean LWS_nor (1.42 m/s) and LWS_max (2.24 m/s); the intraclass correlation coefficient between DWS and LWS were 0.188 to 0.341. DWS was significantly correlated with grip strength, one-legged stance, and LWS. The area under the receiver operating characteristic curve of DWS_sd concerning pre-frailty was largest among DWSs, at 0.615, while that of LWS_nor was 0.643. The findings suggest that DWS differs from LWS and is associated with physical function and pre-frailty.

The Mobile-Based 6-Minute Walk Test: Usability Study and Algorithm Development and Validation

Background

The 6-min walk test (6MWT) is a convenient method for assessing functional capacity in patients with cardiopulmonary conditions. It is usually performed in the context of a hospital clinic and thus requires the involvement of hospital staff and facilities, with their associated costs.

Objective

This study aimed to develop a mobile phone–based system that allows patients to perform the 6MWT in the community.

Methods

We developed 2 algorithms to compute the distance walked during a 6MWT using sensors embedded in a mobile phone. One algorithm makes use of the global positioning system to track the location of the phone when outdoors and hence computes the distance travelled. The other algorithm is meant to be used indoors and exploits the inertial sensors built into the phone to detect U-turns when patients walk back and forth along a corridor of fixed length. We included these algorithms in a mobile phone app, integrated with wireless pulse oximeters and a back-end server. We performed Bland-Altman analysis of the difference between the distances estimated by the phone and by a reference trundle wheel on 49 indoor tests and 30 outdoor tests, with 11 different mobile phones (both Apple iOS and Google Android operating systems). We also assessed usability aspects related to the app in a discussion group with patients and clinicians using a technology acceptance model to guide discussion.

Results

The mean difference between the mobile phone-estimated distances and the reference values was −2.013 m (SD 7.84 m) for the indoor algorithm and −0.80 m (SD 18.56 m) for the outdoor algorithm. The absolute maximum difference was, in both cases, below the clinically significant threshold. A total of 2 pulmonary hypertension patients, 1 cardiologist, 2 physiologists, and 1 nurse took part in the discussion group, where issues arising from the use of the 6MWT in hospital were identified. The app was demonstrated to be usable, and the 2 patients were keen to use it in the long term.

Conclusions

The system described in this paper allows patients to perform the 6MWT at a place of their convenience. In addition, the use of pulse oximetry allows more information to be generated about the patient’s health status and, possibly, be more relevant to the real-life impact of their condition. Preliminary assessment has shown that the developed 6MWT app is highly accurate and well accepted by its users. Further tests are needed to assess its clinical value.

Accelerometric Gait Analysis Devices in Children-Will They Accept Them? Results From the AVAPed Study

Aims: To assess children's acceptance to wear a 3D-accelerometer which is attached to the waist under real-world conditions, and also to compare gait speed during supervised testing with the non-supervised gait speed in every-day life. Methods: In a controlled observational, cross sectional study thirty subjects with cerebral palsy (CP), with level I&II of the Gross Motor Function Classification System (GMFCS) and 30 healthy control children (Ctrl), aged 3-12 years, were asked to perform a 1-min-walking test (1 mwt) under laboratory conditions, and to wear an accelerometric device for a 1-week wearing home measurement (1 WHM). Acceptance was measured via wearing time, and by a questionnaire in which subjects rated restrictions in their daily living and wearing comfort. In addition, validity of 3D-accelerometric gait speed was checked through gold standard assessment of gait speed with a mobile perambulator. Results: Wearing time amounted to 10.3 (SD 3.4) hours per day, which was comparable between groups (T = 1.10, P = 0.3). Mode for wearing comfort [CP 1, Range (1,4), Ctrl 1, Range (1,6)] and restriction of daily living [CP 1, Range (1,3), Ctrl 1, Range (1,4)] was comparable between groups. Under laboratory conditions, Ctrl walked faster in the 1 mwt than CP (Ctrl 1.72 ± 0.29 m/s, CP 1.48 ± 0.41 m/s, P = 0.018). Similarly, a statistically significant difference was found when comparing real-world walking speed and laboratory walking speed (CP: 1 mwt 1.48 ± 0.41 m/s, 1 WHM 0.89 ± 0.09 m/s, P = 0.012; Ctrl: 1mwt 1.72 ± 0.29, 1 WHM 0.97 ± 0.06, P < 0.001). Conclusion: 3D-accelerometry is well-enough accepted in a pediatric population of patients with CP and a Ctrl group to allow valid assessments. Assessment outside the laboratory environment yields information about real world activity that was not captured by routine clinical tests. This suggests that assessment of habitual activities by wearable devices reflects the functioning of children in their home environment. This novel information constitutes an important goal for rehabilitation medicine. The study is registered at the German Register of Clinical Trials with the title "Acceptance and Validity of 3D Accelerometric Gait Analysis in Pediatric Patients" (AVAPed; DRKS00011919).

Walking-speed estimation using a single inertial measurement unit for the older adults

Background

Although walking speed is associated with important clinical outcomes and designated as the sixth vital sign of the elderly, few walking-speed estimation algorithms using an inertial measurement unit (IMU) have been derived and tested in the older adults, especially in the elderly with slow speed. We aimed to develop a walking-speed estimation algorithm for older adults based on an IMU.

Methods

We used data from 659 of 785 elderly enrolled from the cohort study. We measured gait using an IMU attached on the lower back while participants walked around a 28 m long round walkway thrice at comfortable paces. Best-fit linear regression models were developed using selected demographic, anthropometric, and IMU features to estimate the walking speed. The accuracy of the algorithm was verified using mean absolute error (MAE) and root mean square error (RMSE) in an independent validation set. Additionally, we verified concurrent validity with GAITRite using intraclass correlation coefficients (ICCs).

Results

The proposed algorithm incorporates the age, sex, foot length, vertical displacement, cadence, and step-time variability obtained from an IMU sensor. It exhibited high estimation accuracy for the walking speed of the elderly and remarkable concurrent validity compared to the GAITRite (MAE = 4.70%, RMSE = 6.81 𝑐𝑚/𝑠, concurrent validity (ICC (3,1)) = 0.937). Moreover, it achieved high estimation accuracy even for slow walking by applying a slow-speed-specific regression model sequentially after estimation by a general regression model. The accuracy was higher than those obtained with models based on the human gait model with or without calibration to fit the population.

Conclusions

The developed inertial-sensor-based walking-speed estimation algorithm can accurately estimate the walking speed of older adults.

Long-term gait measurements in daily life: Results from the Berlin Aging Study II (BASE-II)

Background

Walking ability is an important prerequisite for activity, social participation and independent living. While in most healthy adults, this ability can be assumed as given, limitations in walking ability occur with increasing age. Furthermore, slow walking speed is linked to several chronic conditions and overall morbidity. Measurements of gait parameters can be used as a proxy to detect functional decline and onset of chronic conditions. Up to now, gait characteristics used for this purpose are measured in standardized laboratory settings. There is some evidence, however, that long-term measurements of gait parameters in the living environment have some advantages over short-term laboratory measurements.

Methods

We evaluated cross-sectional data from an accelerometric sensor worn in a subgroup of 554 participants of the Berlin Aging Study II (BASE-II). Data from the two BASE-II age groups (age between 22–36 years and 60–79 years) were used for the current analysis of accelerometric data for a minimum of two days and a maximum of ten days were available. Real world walking speed, number of steps, maximum coherent distance and total distance were derived as average data per day. Linear regression analyses were performed on the different gait parameters in order to identify significant determinants. Additionally, Mann-Whitney-U-tests were performed to detect sex-specific differences.

Results

Age showed to be significantly associated with real world walking speed and with the total distance covered per day, while BMI contributed negatively to the number of walking steps, maximum coherent distance and total distance walked. Additionally, sex was associated with walking steps. However, R²-values for all models were low. Overall, women had significantly more walking steps and a larger coherent distance per day when compared to men. When separated by age group, this difference was significant only in the older participants. Additionally, walking speed was significantly higher in women compared to men in the subgroup of older people.

Conclusions

Age- and sex-specific differences have to be considered when objective gait parameters are measured, e.g. in the context of clinical risk assessment. For this purpose normative data, differentiating for age and sex would have to be established to allow reliable classification of long-term measurements of gait.

Accurate Ambulatory Gait Analysis in Walking and Running Using Machine Learning Models

Wearable sensors have been proposed as alternatives to traditional laboratory equipment for low-cost and portable real-time gait analysis in unconstrained environments. However, the moderate accuracy of these systems currently limits their widespread use. In this paper, we show that support vector regression (SVR) models can be used to extract accurate estimates of fundamental gait parameters (i.e., stride length, velocity, and foot clearance), from custom-engineered instrumented insoles (SportSole) during walking and running tasks. Additionally, these learning-based models are robust to inter-subject variability, thereby making it unnecessary to collect subject-specific training data. Gait analysis was performed in N=14 healthy subjects during two separate sessions, each including 6-minute bouts of treadmill walking and running at different speeds (i.e., 85% and 115% of each subject's preferred speed). Gait metrics were simultaneously measured with the instrumented insoles and with reference laboratory equipment. SVR models yielded excellent intraclass correlation coefficients (ICC) in all the gait parameters analyzed. Percentage mean absolute errors (MAE%) in stride length, velocity, and foot clearance obtained with SVR models were 1.37%±0.49%, 1.23%±0.27%, and 2.08%±0.72% for walking, 2.59%±0.64%, 2.91%±0.85%, and 5.13%±1.52% for running, respectively. These findings provide evidence that machine learning regression is a promising new approach to improve the accuracy of wearable sensors for gait analysis.

Reliability of the 6-minute walking test smartphone application

OBJECTIVE

Objective functional measures such as the 6-minute walking test (6WT) are increasingly applied to evaluate patients with degenerative diseases of the lumbar spine before and after (surgical) treatment. However, the traditional 6WT is cumbersome to apply, as it requires specialized in-hospital infrastructure and personnel. The authors set out to compare 6-minute walking distance (6WD) measurements obtained with a newly developed smartphone application (app) and those obtained with the gold-standard distance wheel (DW).

METHODS

The authors developed a free iOS- and Android-based smartphone app that allows patients to measure the 6WD in their home environment using global positioning system (GPS) coordinates. In a laboratory setting, the authors obtained 6WD measurements over a range of smartphone models, testing environments, and walking patterns and speeds. The main outcome was the relative measurement error (rME; in percent of 6WD), with |rME| < 7.5% defined as reliable. The intraclass correlation coefficient (ICC) for agreement between app- and DW-based 6WD was calculated.

RESULTS

Measurements (n = 406) were reliable with all smartphone types in neighborhood, nature, and city environments (without high buildings), as well as with unspecified, straight, continuous, and stop-and-go walking patterns (ICC = 0.97, 95% CI 0.97-0.98, p < 0.001). Measurements were unreliable indoors, in city areas with high buildings, and for predominantly rectangular walking courses. Walking speed had an influence on the ME, with worse accuracy (2% higher rME) for every kilometer per hour slower walking pace (95% CI 1.4%-2.5%, p < 0.001). Mathematical adjustment of the app-based 6WD for velocity-dependent error mitigated the rME (p < 0.011), attenuated velocity dependence (p = 0.362), and had a positive effect on accuracy (ICC = 0.98, 95% CI 0.98-0.99, p < 0.001).

CONCLUSIONS

The new, free, spine-specific 6WT smartphone app measures the 6WD conveniently by using GPS coordinates, empowering patients to independently determine their functional status before and after (surgical) treatment. Measurements of 6WD obtained for the target population under the recommended circumstances are highly reliable.

Validity of accelerometry in step detection and gait speed measurement in orthogeriatric patients

BACKGROUND

Mobile accelerometry is a powerful and promising option to capture long-term changes in gait in both clinical and real-world scenarios. Increasingly, gait parameters have demonstrated their value as clinical outcome parameters, but validation of these parameters in elderly patients is still limited.

OBJECTIVE

The aim of this study was to implement a validation framework appropriate for elderly patients and representative of real-world settings, and to use this framework to test and improve algorithms for mobile accelerometry data in an orthogeriatric population.

METHODS

Twenty elderly subjects wearing a 3D-accelerometer completed a parcours imitating a real-world scenario. High-definition video and mobile reference speed capture served to validate different algorithms.

RESULTS

Particularly at slow gait speeds, relevant improvements in accuracy have been achieved. Compared to the reference the deviation was less than 1% in step detection and less than 0.05 m/s in gait speed measurements, even for slow walking subjects (< 0.8 m/s).

CONCLUSION

With the described setup, algorithms for step and gait speed detection have successfully been validated in an elderly population and demonstrated to have improved performance versus previously published algorithms. These results are promising that long-term and/or real-world measurements are possible with an acceptable accuracy even in elderly frail patients with slow gait speeds.

Relationship between Daily and In-laboratory Gait Speed among Healthy Community-dwelling Older Adults

Gait speed in laboratory settings (in-laboratory gait speed) is one of the important indicators associated with the decline in functional abilities in older adulthood. Recently, it has become possible to measure gait speed during daily living (daily gait speed) using accelerometers. However, the relationship between these two gait speed parameters is unclear. This study aimed to compare in-laboratory gait speed, measured by a sheet-type pressure sensor, and daily gait speed, measured by an accelerometer, in healthy community-dwelling older adults. Participants were aged ≥60 years, residing in Takahama city, Aichi, Japan. To calculate daily gait speed, participants were instructed to wear a tri-axial accelerometer on their waist. A total of 1965 participants were included in the final analysis. The results showed a weak association (r = 0.333, p < 0.001) between the two gait speed parameters. Furthermore, average daily gait speed was significantly lower than average in-laboratory gait speed. However, both gait speed parameters declined significantly with age. These results suggest that, in addition to in-laboratory gait speed, daily gait speed may be a helpful parameter for predicting decline in functional abilities.

Feeling older, walking slower-but only if someone's watching. Subjective age is associated with walking speed in the laboratory, but not in real life

The huge inter-individual differences in how people age have prompted researchers to examine whether people's own perception of how old they are-their subjective age-could be a better predictor of relevant outcomes than their actual chronological age. Indeed, how old people feel does predict mortality hazards, and health-related measures such as walking speed may account for this association. In the present study, we extended this line of work by investigating whether subjective age also predicts walking speed and running speed in daily life or whether the predictive effects of subjective age for behavior manifest only within a controlled performance situation. We used data from 80 older participants (age range 62-82 years; M = 69.50, SD = 4.47) from the Berlin Aging Study II (BASE-II). Subjective age was assessed by self-report. Walking speed in the laboratory was measured with the Timed Up and Go test, and walking speed and running speed in real life were measured with an accelerometer. Results showed that compared to participants who felt older, those who felt younger than they actually were indeed walked faster in the laboratory, but they did not walk or run faster in real life. These patterns of results held when age, gender, education, BMI, comorbidity, depression, physical activity, and cognition were covaried. We discuss the role of stereotype threat in accounting for these results.

Remote Monitoring in the Home Validates Clinical Gait Measures for Multiple Sclerosis

Background: The timed 25-foot walk (T25FW) is widely used as a clinic performance measure, but has yet to be directly validated against gait speed in the home environment. Objectives: To develop an accurate method for remote assessment of walking speed and to test how predictive the clinic T25FW is for real-life walking. Methods: An AX3-Axivity tri-axial accelerometer was positioned on 32 MS patients (Expanded Disability Status Scale [EDSS] 0-6) in the clinic, who subsequently wore it at home for up to 7 days. Gait speed was calculated from these data using both a model developed with healthy volunteers and individually personalized models generated from a machine learning algorithm. Results: The healthy volunteer model predicted gait speed poorly for more disabled people with MS. However, the accuracy of individually personalized models was high regardless of disability (R-value = 0.98, p-value = 1.85 × 10-22). With the latter, we confirmed that the clinic T25FW is strongly predictive of the maximum sustained gait speed in the home environment (R-value = 0.89, p-value = 4.34 × 10-8). Conclusion: Remote gait monitoring with individually personalized models is accurate for patients with MS. Using these models, we have directly validated the clinical meaningfulness (i.e., predictiveness) of the clinic T25FW for the first time.

Experimental evaluation of regression model-based walking speed estimation using lower body-mounted IMU

This study provides a concurrent comparison of regression model-based walking speed estimation accuracy using lower body mounted inertial sensors. The comparison is based on different sets of variables, features, mounting locations and regression methods. An experimental evaluation was performed on 15 healthy subjects during free walking trials. Our results show better accuracy of Gaussian process regression compared to least square regression using Lasso. Among the variables, external acceleration tends to provide improved accuracy. By using both time-domain and frequency-domain features, waist and ankle-mounted sensors result in similar accuracies: 4.5% for the waist and 4.9% for the ankle. When using only frequency-domain features, estimation accuracy based on a waist-mounted sensor suffers more compared to the one from ankle.

A machine learning approach for gait speed estimation using skin-mounted wearable sensors: From healthy controls to individuals with multiple sclerosis

Gait speed is a powerful clinical marker for mobility impairment in patients suffering from neurological disorders. However, assessment of gait speed in coordination with delivery of comprehensive care is usually constrained to clinical environments and is often limited due to mounting demands on the availability of trained clinical staff. These limitations in assessment design could give rise to poor ecological validity and limited ability to tailor interventions to individual patients. Recent advances in wearable sensor technologies have fostered the development of new methods for monitoring parameters that characterize mobility impairment, such as gait speed, outside the clinic, and therefore address many of the limitations associated with clinical assessments. However, these methods are often validated using normal gait patterns; and extending their utility to subjects with gait impairments continues to be a challenge. In this paper, we present a machine learning method for estimating gait speed using a configurable array of skin-mounted, conformal accelerometers. We establish the accuracy of this technique on treadmill walking data from subjects with normal gait patterns and subjects with multiple sclerosis-induced gait impairments. For subjects with normal gait, the best performing model systematically overestimates speed by only 0.01 m/s, detects changes in speed to within less than 1%, and achieves a root-mean-square-error of 0.12 m/s. Extending these models trained on normal gait to subjects with gait impairments yields only minor changes in model performance. For example, for subjects with gait impairments, the best performing model systematically overestimates speed by 0.01 m/s, quantifies changes in speed to within 1%, and achieves a root-mean-square-error of 0.14 m/s. Additional analyses demonstrate that there is no correlation between gait speed estimation error and impairment severity, and that the estimated speeds maintain the clinical significance of ground truth speed in this population. These results support the use of wearable accelerometer arrays for estimating walking speed in normal subjects and their extension to MS patient cohorts with gait impairment.

Regression Model-Based Walking Speed Estimation Using Wrist-Worn Inertial Sensor

Walking speed is widely used to study human health status. Wearable inertial measurement units (IMU) are promising tools for the ambulatory measurement of walking speed. Among wearable inertial sensors, the ones worn on the wrist, such as a watch or band, have relatively higher potential to be easily incorporated into daily lifestyle. Using the arm swing motion in walking, this paper proposes a regression model-based method for longitudinal walking speed estimation using a wrist-worn IMU. A novel kinematic variable is proposed, which finds the wrist acceleration in the principal axis (i.e. the direction of the arm swing). This variable (called pca-acc) is obtained by applying sensor fusion on IMU data to find the orientation followed by the use of principal component analysis. An experimental evaluation was performed on 15 healthy young subjects during free walking trials. The experimental results show that the use of the proposed pca-acc variable can significantly improve the walking speed estimation accuracy when compared to the use of raw acceleration information (p<0.01). When Gaussian process regression is used, the resulting walking speed estimation accuracy and precision is about 5.9% and 4.7%, respectively.

Fampridine and real-life walking in multiple sclerosis: Low predictive value of clinical test for habitual short-term changes

BACKGROUND

Fampridine improves walking speed in patients with multiple sclerosis (MS) in performance-based tests. The impact on habitual mobility and its correlation with clinical tests has not been analysed.

OBJECTIVE

To investigate the association between clinical response criteria and habitual mobility in MS patients starting a fampridine treatment.

METHODS

During a four-week baseline-to-treatment study, we assessed in 28 patients (median EDSS 4.75, range 4-6.5) walking tests as the Timed-25-Foot-Walk (T25FW) and mobility questionnaires at day 0, 14 (start of treatment) and 28. Habitual steps and distance per day, total activity and walking speed was measured by accelerometry over four weeks. Beside improvement in real-life mobility, we investigated if such measures differed between non-responders and responders defined by a 20% improvement in clinical tests.

RESULTS

All clinical test, patient reported outcomes and total activity improved significantly (p<0.05). 46% improved (any change >0) in three of four real-life measures. Change of the T25FW predicted only an increase of distance per day. Subjective rating of patients performed better by predicting distance and walking speed changes correctly.

CONCLUSION

Fampridine might improve walking in daily life of MS, but clinical tests are weak predictors. Accelerometry opens a new perspective on mobility measurment, but the current data do not show a consistent effect on non-performance based accelerometry outcomes.

Ecological validity of walking capacity tests in multiple sclerosis

Background

Ecological validity implicates in how far clinical assessments refer to real life. Short clinical gait tests up to ten meters and 2- or 6-Minutes Walking Tests (2MWT/6MWT) are used as performance-based outcomes in Multiple Sclerosis (MS) studies and considered as moderately associated with real life mobility.

Objective

To investigate the ecological validity of 10 Meter Walking Test (10mWT), 2MWT and 6MWT.

Methods

Persons with MS performed 10mWT, 6MWT including 2MWT and 7 recorded days by accelerometry. Ecological validity was assumed if walking tests represented a typical walking sequence in real-life and correlations with accelerometry parameters were strong.

Results

In this cohort (n=28, medians: age=45, EDSS=3.2, disease duration=9 years), uninterrupted walking of 2 or 6 minutes occurred not frequent in real life (2.61 and 0.35 sequences/day). 10mWT correlated only with slow walking speed quantiles in real life. 2MWT and 6MWT correlated moderately with most real life walking parameters.

Conclusion

Clinical gait tests over a few meters have a poor ecological validity while validity is moderate for 2MWT and 6MWT. Mobile accelerometry offers the opportunity to control and improve the ecological validity of MS mobility outcomes.

Novel algorithm for a smartphone-based 6-minute walk test application: algorithm, application development, and evaluation

Background

The 6-minute walk test (6MWT: the maximum distance walked in 6 minutes) is used by rehabilitation professionals as a measure of exercise capacity. Today’s smartphones contain hardware that can be used for wearable sensor applications and mobile data analysis. A smartphone application can run the 6MWT and provide typically unavailable biomechanical information about how the person moves during the test.

Methods

A new algorithm for a calibration-free 6MWT smartphone application was developed that uses the test’s inherent conditions and smartphone accelerometer-gyroscope data to report the total distance walked, step timing, gait symmetry, and walking changes over time. This information is not available with a standard 6MWT and could help with clinical decision-making.

The 6MWT application was evaluated with 15 able-bodied participants. A BlackBerry Z10 smartphone was worn on a belt at the mid lower back. Audio from the phone instructed the person to start and stop walking. Digital video was independently recorded during the trial as a gold-standard comparator.

Results

The average difference between smartphone and gold standard foot strike timing was 0.014 ± 0.015 s. The total distance calculated by the application was within 1 m of the measured distance for all but one participant, which was more accurate than other smartphone-based studies.

Conclusions

These results demonstrated that clinically relevant 6MWT results can be achieved with typical smartphone hardware and a novel algorithm.

Electronic supplementary material

The online version of this article (doi:10.1186/s12984-015-0013-9) contains supplementary material, which is available to authorized users.

Progressive adaptation in physical activity and neuromuscular performance during 520d confinement

To understand whether prolonged confinement results in reductions in physical activity and adaptation in the musculoskeletal system, six subjects were measured during 520 d isolation in the Mars500 study. We tested the hypothesis that physical activity reduces in prolonged confinement and that this would be associated with decrements of neuromuscular performance. Physical activity, as measured by average acceleration of the body's center of mass ("activity temperature") using the actibelt® device, decreased progressively over the course of isolation (p<0.00001). Concurrently, countermovement jump power and single-leg hop force decreased during isolation (p<0.001) whilst grip force did not change (p≥0.14). Similar to other models of inactivity, greater decrements of neuromuscular performance occurred in the lower-limb than in the upper-limb. Subject motivational state increased non-significantly (p = 0.20) during isolation, suggesting reductions in lower-limb neuromuscular performance were unrelated to motivation. Overall, we conclude that prolonged confinement is a form of physical inactivity and is associated with adaptation in the neuromuscular system.

A new method to estimate the real upper limit of the false alarm rate in a 3 accelerometry-based fall detector for the elderly

Falls are a major concern for the elderly and their ability to remain healthy. Fall detection systems may notify emergency responders when no one apart from the injured is present. However, their real-world application is limited by a number of factors such as high false positive rates, low-compliance, poor-usability and short battery lifetime. In order to improve these aspects we have developed a miniaturized 3D accelerometer integrated in a belt buckle, the actibelt(®), and a fall detection algorithm. We have used a new evaluation method to assess the upper limit of the false alarm rate of our algorithm using a large set of long term standardized acceleration measurements recorded under real life conditions. Our algorithm has a false alarm rate of seventeen false alarms per month and has the potential to be reduced down to at most three false alarms per month when activities which require the sensor to be removed are eliminated. In laboratory settings, the algorithm has a sensitivity of 100%. The algorithm was sucessfully validated using data from a real-world fall.

Accuracy of the actibelt(®) accelerometer for measuring walking speed in a controlled environment among persons with multiple sclerosis

BACKGROUND

Advances in portable sensor technology have opened an era for objective, real-life monitoring of walking speed in persons with multiple sclerosis (MS).

PURPOSE

The present study examined the accuracy of the actibelt(®) accelerometer for measuring walking speed during a standard 6-min walk (6MW) and the possibility that disability status influenced the degree of accuracy among persons with MS.

METHODS

On a single testing session, 51 persons with MS and Expanded Disability Status Scale scores between 2.0 and 6.5 performed a 6MW while wearing an actibelt(®) in the body's sagittal symmetry plane and close to the body's centre of mass.

RESULTS

All 51 participants completed the 6MW without stopping, falling, or any adverse events, and the actibelt(®) provided walking speed data for each of the participants. The actibelt(®) significantly overestimated walking speed (actual minus actibelt(®)) by a mean±standard deviation of -0.12±0.17 m/s for the overall sample (p<0.0001). There was no significant overestimation in the sample with mild disability (-0.02±0.11 m/s), but there was in the samples with moderate (-0.10±0.16 m/s) and severe (-0.26±0.12 m/s) disability.

CONCLUSION

The actibelt(®) is ready for real-life monitoring of walking speed in persons with mild MS, but caution is necessary when interpreting the accuracy of the walking speed data for those with MS who have moderate and severe disability.

Association between walking speed and age in healthy, free-living individuals using mobile accelerometry--a cross-sectional study

CONTEXT

Walking speed is a fundamental parameter of human motion and is increasingly considered as an important indicator of individuals' health status.

OBJECTIVE

To evaluate the relationship of gait parameters, and demographic and physical characteristics in healthy men and women.

DESIGN, SETTING, AND PARTICIPANTS

Recruitment of a subsample (n = 358) of male and female blood donors taking part in the Cambridge CardioResource study. Collection of demographic data, measurement of physical characteristics (height, weight and blood pressure) and assessment of 7-day, free-living activity parameters using accelerometry and a novel algorithm to measure walking speed. Participants were a median (interquartile range[IQR]) age of 49 (16) years; 45% women; and had a median (IQR) BMI of 26 (5.4).

MAIN OUTCOME MEASURE

Walking speed.

RESULTS

In this study, the hypothesis that walking speed declines with age was generated using an initial 'open' dataset. This was subsequently validated in a separate 'closed' dataset that showed a decrease of walking speed of -0.0037 m/s per year. This is equivalent to a difference of 1.2 minutes, when walking a distance of 1 km aged 20 compared to 60 years. Associations between walking speed and other participant characteristics (i.e. gender, BMI and blood pressure) were non-significant. BMI was negatively correlated with the number of walking and running steps and longest non-stop distance.

CONCLUSION

This is the first study using accelerometry which shows an association between walking speed and age in free-living, healthy individuals. Absolute values of gait speed are comparable to published normal ranges in clinical settings. This study highlights the potential use of mobile accelerometry to assess gait parameters which may be indicative of future health outcomes in healthy individuals.