CMI was not significantly different between fallers and non-fallers or people with and without CoF; however, our taxonomy revealed a large variety of cognitive conditions and a higher number of studies using mental tracking tasks, which make it impossible to draw firm conclusions. Future studies should use a more standardised and ecologically valid approach when evaluating the validity of DT gait performance in the prediction of falls, CoF or other age-related conditions.
The aim of this study was to examine the effects of concern about falling on gait speed, adjusted for physiological fall risk and cognitive function. Gait speed, especially under dual-task conditions, was affected by concern about falling. Concern about falling was the strongest predictor of gait speed under all 4 conditions and should be included in routine geriatric assessments.
Standardized tests of gait speed are regarded as being of clinical value, but they are typically performed under optimal conditions, and may not reflect daily-life gait behavior. The aim of this study was to compare 4-m gait speed to the distribution of daily-life gait speed. The 4-m gait speed is only weakly related to daily-life gait speed. Clinicians and researchers should consider that 4-m gait speed and daily-life gait speed represent two different constructs.
The ReacStick is a reliable test of reaction time and inhibitory EF in older people and could have value for fall-risk assessment.
These data provide a valuable reference and may call for more age- and gender-specific activity interventions.
These data provide a valuable reference and may call for more age- and gender-specific activity interventions.
eHealth solutions are increasingly being applied to deliver interventions for promoting an active lifestyle in the general population but also in older people. Objective assessment of daily physical activity (PA) is essential to accurately and reliably evaluate the effectiveness of such interventions. This review presents an overview of eHealth interventions that focus on promoting PA in community-dwelling older people, and discusses the methods used to objectively assess PA, and the effectiveness of the eHealth interventions in increasing PA. The twelve eHealth intervention studies that met our inclusion criteria used a variety of digital solutions, ranging from solely the use of an accelerometer or text messages, to interactive websites with access to (animated) coaches and peer support. Besides evaluating the effectiveness of an intervention on objectively assessed PA, all interventions also included continuous self-monitoring of PA as part of the intervention. Procedures for the collection and analysis of PA data varied across studies; five studies used pedometers to objectively assess PA and seven used tri-axial accelerometers. Main reported outcomes were daily step counts and minutes spent on PA. The current evidence seems to point to a positive short-term effect of increased PA (i.e. right after administering the intervention), but evidence for long-term effects is lacking. Many studies were underpowered to detect any intervention effects, and therefore larger studies with longer follow-up are needed to provide evidence on sustaining the PA increases that follow eHealth interventions in older people.
Early detection of high fall risk is an essential component of fall prevention in older adults. Wearable sensors can provide valuable insight into daily-life activities; biomechanical features extracted from such inertial data have been shown to be of added value for the assessment of fall risk. Body-worn sensors such as accelerometers can provide valuable insight into fall risk. Currently, biomechanical features derived from accelerometer data are used for the assessment of fall risk. Here, we studied whether deep learning methods from machine learning are suited to automatically derive features from raw accelerometer data that assess fall risk. We used an existing dataset of 296 older adults. We compared the performance of three deep learning model architectures (convolutional neural network (CNN), long short-term memory (LSTM) and a combination of these two (ConvLSTM)) to each other and to a baseline model with biomechanical features on the same dataset. The results show that the deep learning models in a single-task learning mode are strong in recognition of identity of the subject, but that these models only slightly outperform the baseline method on fall risk assessment. When using multi-task learning, with gender and age as auxiliary tasks, deep learning models perform better. We also found that preprocessing of the data resulted in the best performance (AUC = 0.75). We conclude that deep learning models, and in particular multi-task learning, effectively assess fall risk on the basis of wearable sensor data.
Age and age-related diseases have been suggested to decrease entropy of human gait kinematics, which is thought to make older adults more susceptible to falls. In this study we introduce a new entropy measure, called phase-dependent generalized multiscale entropy (PGME), and test whether this measure improves fall-risk prediction in community-dwelling older adults. PGME can assess phase-dependent changes in the stability of gait dynamics that result from kinematic changes in events such as heel strike and toe-off. PGME was assessed for trunk acceleration of 30 s walking epochs in a re-analysis of 1 week of daily-life activity data from the FARAO study, originally described by van Schooten et al. (2016). The re-analyzed data set contained inertial sensor data from 52 single- and 46 multiple-time prospective fallers in a 6 months follow-up period, and an equal number of non-falling controls matched by age, weight, height, gender, and the use of walking aids. The predictive ability of PGME for falls was assessed using a partial least squares regression. PGME had a superior predictive ability of falls among single-time prospective fallers when compared to the other gait features. The single-time fallers had a higher PGME ( < 0.0001) of their trunk acceleration at 60% of their step cycle when compared with non-fallers. No significant differences were found between PGME of multiple-time fallers and non-fallers, but PGME was found to improve the prediction model of multiple-time fallers when combined with other gait features. These findings suggest that taking into account phase-dependent changes in the stability of the gait dynamics has additional value for predicting falls in older people, especially for single-time prospective fallers.
When compared to less frequent fallers, more frequent fallers had a lower risk for injury per fall. This appeared to be explained by differences in the circumstances of falls, and not by protective responses. Injury prevention strategies in long-term care should target both frequent and infrequent fallers, as the latter are more mobile and apt to sustain injury.
Falls are a major health concern for older adults. Understanding sex differences in fall circumstances may guide the design of fall management plans specifically to men and women. In this study, analyzed real-life falls captured on video to compare scenarios leading to falls between men and women in 2 long-term care (LTC) facilities. Our results elucidate differences between older men and women in the scenarios that lead to falls, to inform sex-specific fall prevention strategies in the LTC setting.
Identification of the factors that influence sedentary behaviour in older adults is important for the design of appropriate intervention strategies. In this study, we determined the prevalence of sedentary behaviour and its association with physical, cognitive, and psychosocial status among older adults residing in Assisted Living (AL). We found that sedentary behaviour among older adults in AL associated with TUG scores and falls-related self-efficacy, which are modifiable targets for interventions to decrease sedentary behaviour in this population.
Over the last decades, various measures have been introduced to assess stability during walking. All of these measures assume that gait stability may be equated with exponential stability, where dynamic stability is quantified by a Floquet multiplier or Lyapunov exponent. These specific constructs of dynamic stability assume that the gait dynamics are time independent and without phase transitions. In this case the temporal change in distance, d(t), between neighboring trajectories in state space is assumed to be an exponential function of time. However, results from walking models and empirical studies show that the assumptions of exponential stability break down in the vicinity of phase transitions that are present in each step cycle. Here we apply a general non-exponential construct of gait stability, called fractional stability, which can define dynamic stability in the presence of phase transitions. Fractional stability employs the fractional indices, α and β, of differential operator which allow modeling of singularities in d(t) that cannot be captured by exponential stability. The fractional stability provided an improved fit of d(t) compared to exponential stability when applied to trunk accelerations during daily-life walking in community-dwelling older adults. Moreover, using multivariate empirical mode decomposition surrogates, we found that the singularities in d(t), which were well modeled by fractional stability, are created by phase-dependent modulation of gait. The new construct of fractional stability may represent a physiologically more valid concept of stability in vicinity of phase transitions and may thus pave the way for a more unified concept of gait stability.
Self-reported physical activity has shown to affect muscle-related parameters. As self-report is likely biased, this study aimed to assess the association between instrumented assessment of physical activity (I-PA) and muscle-related parameters in a general population.
The interactions between gait characteristics and health status indicate that gait characteristics are differently associated with fall history between stroke survivors and older adults. Thus specific models are needed to predict fall risk in stroke survivors.
Falls can have devastating consequences for older people. We determined the relationship between the likelihood of fall incidents and daily-life behavior. We used wearable sensors to assess habitual physical activity and daily-life gait quality (in terms of e.g. stability, variability, smoothness and symmetry), and determined their predictive ability for time-to-first-and-second-falls. 319 older people wore a trunk accelerometer (Dynaport MoveMonitor, McRoberts) during one week. Participants further completed questionnaires and performed grip strength and trail making tests to identify risk factors for falls. Their prospective fall incidence was followed up for six to twelve months. We determined interrelations between commonly used gait characteristics to gain insight in their interpretation and determined their association with time-to-falls. For all data -including questionnaires and tests- we determined the corresponding principal components and studied their predictive ability for falls. We showed that gait characteristics of walking speed, stride length, stride frequency, intensity, variability, smoothness, symmetry and complexity were often moderately to highly correlated (r > 0.4). We further showed that these characteristics were predictive of falls. Principal components dominated by history of falls, alcohol consumption, gait quality and muscle strength proved predictive for time-to-fall. The cross-validated prediction models had adequate to high accuracy (time dependent AUC of 0.66-0.72 for time-to-first-fall and 0.69-0.76 for -second-fall). Daily-life gait quality obtained from a single accelerometer on the trunk is predictive for falls. These findings confirm that ambulant measurements of daily behavior contribute substantially to the identification of elderly at (high) risk of falling.
Gait characteristics revealed less stable, less symmetric, and more variable gait during daily life than on a treadmill, yet about half of the characteristics were significantly correlated between conditions. These results suggest that daily-life gait analysis is sensitive to static personal factors (i.e., physical and cognitive capacity) as well as dynamic situational factors (i.e., behavior and environment), which may both represent determinants of fall risk.
Daily-life accelerometry contributes substantially to the identification of individuals at risk of falls, and can predict falls in 6 months with good accuracy.
The objective of this study was to improve fall-risk prediction models by examining whether the use of extreme values strengthens the associations with falls. Associations were stronger for extreme values, indicating "high gait quality" situations (ie, 10th and 90th percentiles in case of positive and negative associations, respectively) and not for "low gait quality" situations. This suggests that gait characteristics during optimal performance gait provide more information about the risk of falling than high-risk situations. However, their added value over medians in prediction is limited.
Estimates of gait characteristics may suffer from errors due to discrepancies in accelerometer location. This is particularly problematic for gait measurements in daily life settings, where consistent sensor positioning is difficult to achieve. To address this problem, we equipped 21 healthy adults with tri-axial accelerometers (DynaPort MiniMod, McRoberts) at the mid and lower lumbar spine and anterior superior iliac spine (L2, L5 and ASIS) while continuously walking outdoors back and forth (20 times) over a distance of 20 m, including turns. We compared 35 gait characteristics between sensor locations by absolute agreement intra-class correlations (2, 1; ICC). We repeated these analyses after applying a new method for off-line sensor realignment providing a unique definition of the vertical and, by symmetry optimization, the two horizontal axes. Agreement between L2 and L5 after realignment was excellent (ICC>0.9) for stride time and frequency, speed and their corresponding variability and good (ICC>0.7) for stride regularity, movement intensity, gait symmetry and smoothness and for local dynamic stability. ICC values benefited from sensor realignment. Agreement between ASIS and the lumbar locations was less strong, in particular for gait characteristics like symmetry, smoothness, and local dynamic stability (ICC generally<0.7). Unfortunately, this lumbar-ASIS agreement did not benefit consistently from sensor realignment. Our findings show that gait characteristics are robust against limited repositioning error of sensors at the lumbar spine, in particular if our off-line realignment is applied. However, larger positioning differences (from lumbar positions to ASIS) yield less consistent estimates and should hence be avoided.
Background. Gait characteristics extracted from trunk accelerations during daily life locomotion are complementary to questionnaire- or laboratory-based gait and balance assessments and may help to improve fall risk prediction. Objective. The aim of this study was to identify gait characteristics that are associated with self-reported fall history and that can be reliably assessed based on ambulatory data collected during a single week. Methods. We analyzed 2 weeks of trunk acceleration data (DynaPort MoveMonitor, McRoberts) collected among 113 older adults (age range, 65-97 years). During episodes of locomotion, various gait characteristics were determined, including local dynamic stability, interstride variability, and several spectral features. For each characteristic, we performed a negative binomial regression analysis with the participants' self-reported number of falls in the preceding year as outcome. Reliability of gait characteristics was assessed in terms of intraclass correlations between both measurement weeks. Results. The percentages of spectral power below 0.7 Hz along the vertical and anteroposterior axes and below 10 Hz along the mediolateral axis, as well as local dynamic stability, local dynamic stability per stride, gait smoothness, and the amplitude and slope of the dominant frequency along the vertical axis, were associated with the number of falls in the preceding year and could be reliably assessed (all P < .05, intraclass correlation > 0.75). Conclusions. Daily life gait characteristics are associated with fall history in older adults and can be reliably estimated from a week of ambulatory trunk acceleration measurements.
Stride-to-stride variability and local dynamic stability of gait kinematics are promising measures to identify individuals at increased risk of falling. This study aimed to explore the feasibility of using these metrics in clinical practice and ambulatory assessment, where only a small number of consecutive strides are available. The concurrent validity and reliability were assessed compared to more continuous walking. Twenty young adults walked continuously for 500 m, as well as 36 bouts of 20 m while wearing an accelerometer (DynaPort MiniMod) on the trunk. Within-day reliability was high for stride time variability, mediolateral trunk variability and local dynamic stability, while between-day reliability was low for both variability estimates and moderate for local dynamic stability. Stride time variability and mediolateral trunk variability were increased when walking short bouts and did not correlate well with the longer walking trials. Local dynamic stability did correlate highly between the long and short bouts trials, and 15 bouts of eight strides appeared to be sufficient for valid estimation. These results imply task-specific differences and low reliability of variability estimates rendering them unsuitable for application to short bouts of gait, while local dynamic stability can be readily employed.
We investigated the reliability of physical activity monitoring based on trunk accelerometry in older adults and assessed the number of measured days required to reliably assess physical activity. Seventy-nine older adults (mean age 79.1 ± 7.9) wore an accelerometer at the lower back during two nonconsecutive weeks. The duration of locomotion, lying, sitting, standing and shuffling, movement intensity, the number of locomotion bouts and transitions to standing, and the median and maximum duration of locomotion were determined per day. Using data of week 2 as reference, intraclass correlations and smallest detectable differences were calculated over an increasing number of consecutive days from week 1. Reliability was good to excellent when whole weeks were assessed. Our results indicate that a minimum of two days of observation are required to obtain an ICC ≥ 0.7 for most activities, except for lying and median duration of locomotion bouts, which required up to five days.
Characteristics of dynamical systems are often estimated to describe physiological processes. For instance, Lyapunov exponents have been determined to assess the stability of the cardio-vascular system, respiration, and, more recently, human gait and posture. However, the systematic evaluation of the accuracy and precision of these estimates is problematic because the proper values of the characteristics are typically unknown. We fill this void with a set of standardized time series with well-defined dynamical characteristics that serve as a benchmark. Estimates ought to match these characteristics, at least to good approximation. We outline a procedure to employ this generic benchmark test and illustrate its capacity by examining methods for estimating the maximum Lyapunov exponent. In particular, we discuss algorithms by Wolf and co-workers and by Rosenstein and co-workers and evaluate their performances as a function of signal length and signal-to-noise ratio. In all scenarios, the precision of Rosenstein's algorithm was found to be equal to or greater than Wolf's algorithm. The latter, however, appeared more accurate if reasonably large signal lengths are available and noise levels are sufficiently low. Due to its modularity, the presented benchmark test can be used to evaluate and tune any estimation method to perform optimally for arbitrary experimental data.
This study aimed to investigate kinematic changes experienced during running-induced fatigue. Further, the study examined relations between kinematic changes and core endurance. Novice runners displayed an overall increase in trunk inclination and increased ankle eversion peak angles when fatigued utilizing a running-induced fatigue protocol. As most pronounced changes were found for the trunk, trunk kinematics appear to be significantly affected during fatigued running and should not be overlooked. Core endurance measures displayed unexpected relations with running kinematics and require further investigation to determine the significance of these relations.
Estimating local dynamic stability is considered a powerful approach to identify persons with balance impairments. Its validity has been studied extensively, and provides evidence that short-term local dynamic stability is related to balance impairments and the risk of falling. Thus far, however, this relation has only been proven on group level. For clinical use, differences on the individual level should also be detectable, requiring reliability to be high. In the current study, reliability of short-term local dynamic stability was investigated within and between days. Participants walked 500 m back and forth on a straight outdoor footpath, on 2 non-consecutive days, and 3D linear accelerations were measured using an accelerometer (DynaPort MiniMod). The state space was reconstructed using 4 common approaches, all based on delay embedding. Within-session intra-class correlation coefficients were good (≥0.70), however between-session intra-class correlation coefficients were poor to moderate (≤0.63) and influenced by the reconstruction method. The same holds for the smallest detectable difference, which ranged from 17% to 46% depending on the state space reconstruction method. The best within- and between-session intra-class correlation coefficients and smallest detectable differences were achieved with a state space reconstruction with a fixed time delay and number of embedding dimensions. Overall, due to the influence of biological variation and measurement error, the short-term local dynamic stability can only be used to detect substantial differences on the individual level.
For targeted prevention of falls, it is necessary to identify individuals with balance impairments. To test the sensitivity of measures of variability, local stability and orbital stability of trunk kinematics to balance impairments during gait, we used galvanic vestibular stimulation (GVS) to impair balance in 12 young adults while walking on a treadmill at different speeds. Inertial sensors were used to measure trunk accelerations, from which variability in the medio-lateral direction and local and orbital stability were calculated. The short-term Lyapunov exponent and variability reflected the destabilizing effect of GVS, while the long-term Lyapunov exponent and Floquet multipliers suggested increased stability. Therefore, we concluded that only short-term Lyapunov exponents and variability can be used to asses stability of gait. In addition, to investigate the feasibility of using these measures in screening for fall risk, the presence or absence of GVS was predicted with variability and the short-term Lyapunov exponent. Predictions were good at all walking speeds, but best at preferred walking speed, with a correct classification in 83.3% of the cases.
Impaired balance control during gait can be detected by local dynamic stability measures. For clinical applications, the use of a treadmill may be limiting. Therefore, the aim of this study was to test sensitivity of these stability measures collected during short episodes of over-ground walking by comparing normal to impaired balance control. Galvanic vestibular stimulation (GVS) was used to impair balance control in 12 healthy adults, while walking up and down a 10 m hallway. Trunk kinematics, collected by an inertial sensor, were divided into episodes of one stroll along the hallway. Local dynamic stability was quantified using short-term Lyapunov exponents (λ(s)), and subjected to a bootstrap analysis to determine the effects of number of episodes analysed on precision and sensitivity of the measure. λ(s) increased from 0.50 ± 0.06 to 0.56 ± 0.08 (p = 0.0045) when walking with GVS. With increasing number of episodes, coefficients of variation decreased from 10 ± 1.3% to 5 ± 0.7% and the number of p values >0.05 from 42 to 3.5%, indicating that both precision of estimates of λ(s) and sensitivity to the effect of GVS increased. λ(s) calculated over multiple episodes of over-ground walking appears to be a suitable measure to calculate local dynamic stability on group level.