CoChat
Sign Up
Cited by 108Balance, Gait, and Falls PreventionLower Extremity Biomechanics and PathologiesGait Recognition and Analysis

The DOI matches its Crossref record, with a small difference in title or year; 42 of the 42 references it cites were found in a registry or an index; none of the 41 sources its claims cite is checked for support yet.

Human walking in the real world: Interactions between terrain type, gait parameters, and energy expenditure

Open access
Kowalsky, Daniel B., Rebula, John R., Ojeda, Lauro V., Adamczyk, Peter G., Kuo, Arthur D.·2021·PLOS One
DOI 10.1371/journal.pone.0228682PMC PMC7806134

Published version of a preprint

Open this paper in CoChat — free account

Abstract

Humans often traverse real-world environments with a variety of surface irregularities and inconsistencies, which can disrupt steady gait and require additional effort. Such effects have, however, scarcely been demonstrated quantitatively, because few laboratory biomechanical measures apply outdoors. Walking can nevertheless be quantified by other means. In particular, the foot’s trajectory in space can be reconstructed from foot-mounted inertial measurement units (IMUs), to yield measures of stride and associated variabilities. But it remains unknown whether such measures are related to metabolic energy expenditure. We therefore quantified the effect of five different outdoor terrains on foot motion (from IMUs) and net metabolic rate (from oxygen consumption) in healthy adults (N = 10; walking at 1.25 m/s). Energy expenditure increased significantly ( P < 0.05) in the order Sidewalk, Dirt, Gravel, Grass, and Woodchips, with Woodchips about 27% costlier than Sidewalk. Terrain type also affected measures, particularly stride variability and virtual foot clearance (swing foot’s lowest height above consecutive footfalls). In combination, such measures can also roughly predict metabolic cost (adjusted R 2 = 0.52, partial least squares regression), and even discriminate between terrain types (10% reclassification error). Body-worn sensors can characterize how uneven terrain affects gait, gait variability, and metabolic cost in the real world.

Overview

AI snapshot

Written from the paper’s full text.

TL;DR

Foot-mounted IMUs and portable respirometry were used to quantify how five outdoor terrains affect gait and metabolic cost in healthy adults walking at about 1.25 m/s. Energy expenditure increased from Sidewalk to Dirt, Gravel, Grass, and Woodchips, with Woodchips about 27% costlier than Sidewalk, and stride measures such as virtual clearance and variability were terrain-dependent and roughly predictive of metabolic cost.

Key findings

  • Net metabolic rate increased significantly across terrains (repeated-measures ANOVA, P = 7.11e-11), from Sidewalk 189.1 ± 29.00 W to Woodchips 240.8 ± 28.91 W; Woodchips was about 27% costlier than Sidewalk. Non-significant pairwise comparisons were Dirt vs. Sidewalk, Gravel vs. Dirt, and Grass vs. Gravel (P ≥ 0.05).
  • Terrain affected stride measures, especially virtual clearance: mean changed by up to 58% and RMS variability by up to 63% across terrains.
  • Multivariate stride measures predicted metabolic rate better than any single measure: partial least squares regression yielded adjusted R² = 0.52, principal components regression R² = 0.46, and the strongest single univariate regressor, mean virtual clearance, R² = 0.34 (individualized R² = 0.49).
  • Linear discriminants classified terrain from stride measures with 10% resubstitution error (5 errors out of 50 observations).

Methods

  • Ten healthy adults (N = 10; 5 male, 5 female; age 18–48; body mass 64.86 ± 10.10 kg; leg length 0.90 ± 0.07 m) walked outdoors in Nichols Arboretum on five surfaces: Sidewalk, Dirt, Gravel, Grass, and Woodchips, presented in random order for 8 minutes each.
  • Speed was loosely regulated at approximately 1.25 m/s via GPS while following an experimenter; metabolic data were collected with portable respirometry and only the last 3 minutes of each trial were used for net metabolic rate after subtracting quiet standing (97.29 ± 27.06 W).
  • Foot paths were reconstructed from foot-mounted IMUs; 90 strides per foot were analyzed from forward walking sections, with only left-foot data reported. Analyses included repeated-measures ANOVA, linear regressions, PCA, LDA, PCR, and PLSR, with stride distances normalized by leg length and metabolic power by Mg^1.5 L^0.5.

Limitations

  • The paper states that the analysis used a relatively small number of summary measures, that foot-path reconstruction errors exceed laboratory motion capture, and that the method relies on the foot being nearly stationary during stance, which may not occur on softer terrains such as Woodchips, adding uncertainty to stride height and variability.
  • The paper also states that reported variability includes terrain, inertial drift, and measurement noise; that kinematic measures cannot capture leg force or power; that inertial data cannot readily discern step width; and that more challenging terrain, speed and elevation variations, and carried loads were not examined.
  • Evident from the text: the sample was small (N = 10), only five terrains were studied, walking speed was only loosely controlled, and only left-foot data were used for reported measures.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1371/journal.pone.0228682

Cites

42 of 42 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

0 of 41 cited sources checked · 0 supported · 0 flagged

  • Findingown data

    PLSR yielded adjusted R2 = 0.52 to predict metabolic rate using only two regressor components.

    “This technique yielded adjusted R2 = 0.52 to predict metabolic rate using only two such regressor components.”
    Quote in the paperSubstantial claim
  • Findingreasoning

    The most notable sensitivities were for variability in stride height, stride width, virtual clearance, and lateral swing motion.

    “The most notable sensitivities were for variability in stride height, stride width, virtual clearance, and lateral swing motion.”
    Quote in the paperSubstantial claim
  • Findingown data

    The top four correlates of metabolic rate were mean virtual clearance, and RMS variabilities of virtual clearance, stride height, and stride width.

    “For goodness of fit, the top four correlates were mean virtual clearance, and RMS variabilities of virtual clearance, stride height, and stride width.”
    Quote in the paperSubstantial claim
  • Findingown data

    Lateral swing displacement differed across terrains to a lesser degree.

    “and to lesser degree, lateral swing displacement (mean and variability, summarized in Table 1)”
    Quote in the paperNot a substantial claim
  • Findingown data

    The only non-significant pairwise comparisons in net metabolic rate were Dirt vs. Sidewalk, Gravel vs. Dirt, and Grass vs. Gravel (P ≥ 0.05).

    “The only non-significant comparisons were Dirt vs. Sidewalk, Gravel vs. Dirt, and Grass vs. Gravel (P ≥ 0.05).”
    Quote in the paperNot a substantial claim
  • It may state two separate claims as one. 36
  • Probably not a substantial claim. 26

  • It may state two separate claims as one.

    Findingown data

    The top four correlates were strongly significant regressors (at most P = 3.1e-06) but had modest predictive ability, with adjusted R2 ranging 0.29-0.38.

    “These measures were all strongly significant regressors (at most P = 3.1e-06), although the actual predictive ability was modest, with adjusted R2 ranging 0.29–0.38).”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    The discriminated data show reasonably good discrimination between Gravel vs. Grass and Sidewalk vs. Dirt.

    “The discriminated data (Fig 6, bottom) show reasonably good discrimination between those same pairs.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Findingown data

    Virtual clearance of the swing foot increased on more challenging terrain and was highly correlated with energy expenditure.

    “Most notable was virtual clearance of the swing foot, which increased on more challenging terrain (Table 1), and was highly correlated with energy expenditure (Table 2).”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    Some individuals exhibited terrain-dependence in their stride length vs. speed relationship, but with no consistent statistical trend across subjects.

    “Some individuals exhibited terrain-dependence in their relationship (Fig 6, top), but with no consistent statistical trend across subjects.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Findingown data

    Net metabolic rate varied with terrain type groupwise (repeated measures ANOVA, P = 7.1e-11) and for most pair-wise comparisons (post hoc paired t-tests, P < 0.05).

    “Net metabolic rate E˙met varied with terrain type for groupwise (repeated measures ANOVA, P = 7.1e-11) and for most pair-wise comparisons (post hoc paired t-tests, P < 0.05)”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    Multiple stride parameters are terrain-dependent and correlated with energy cost.

    “We found that multiple stride parameters are indeed terrain-dependent and correlated with energy cost.”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    Stride lengths and speeds were highly correlated with each other, and overlapped for different terrains.

    “Stride lengths and speeds (filled symbols) were highly correlated with each other, and overlapped for different terrains.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Findingown data

    Virtual clearance mean changed by up to 58% and its variability by up to 63% across terrains.

    “particularly the measures of virtual clearance (mean changing by up to 58% and variability by up to 63%)”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    The second principal component accounted for an additional 21.7% of variability (both PCs 87.5%) and was dominated by increased stride length, increased stride height, and increased stride width variability.

    “The second PC accounted for an additional 21.7% (and thus both PCs 87.5%), and was dominated by increased stride length, increased stride height (apparent uphill slope), and increased stride width variability.”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    The first principal component accounted for 65.8% of all terrain-specific variability in stride measures and was dominated by increased stride length, increased walking speed, and negative stride height.

    “The first PC accounted for 65.8% of all terrain-specific variability in the stride measures, and was dominated by increased stride length, increased walking speed, and negative stride height (apparent downhill slope).”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    Both PLSR and PCR exceed the fit for the strongest single univariate regression (virtual clearance, R2 = 0.34).

    “Both of these exceed the fit for the strongest single univariate regression (virtual clearance, with R2 = 0.34).”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    Co-variation of speed and stride length dominated the first principal component of stride measures and predicted energy expenditure from the principal components regression.

    “Indeed, co-variation of speed and stride length dominated the first principal component of stride measures (Fig 5), and predicted energy expenditure from the principal components regression (PCR, Fig 7).”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    Nearly every stride measure exhibited significant correlation with energy expenditure, most strongly the RMS variabilities of stride height, virtual clearance, and stride width.

    “Nearly every stride measure exhibited significant correlation with energy expenditure, most strongly the RMS variabilities of stride height, virtual clearance, and stride width (Table 2).”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    All stride variability measures were individually correlated with energy cost, although they contributed relatively little to the first two principal components.

    “In addition, all stride variability measures were individually correlated with energy cost (Table 2), although they contributed relatively little to the first two principal components.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductcitation

    There was some variation in walking speed within each trial despite attempted approximate control of average walking speed.

    “Although we attempted to approximately control the average walking speed, there was some variation within each trial.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Interpretationreasoning

    Virtual clearance is merely an indicator of possible adaptations to true ground clearance.

    “Of course, the details of actual surface variations were unknown, and so virtual clearance is merely an indicator of possible adaptations to true ground clearance.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The details of actual surface variations were unknown.

  • Probably not a substantial claim.

    Study design and conductnone given

    Stride parameters and energy measures were normalized to account for differences in subject body size and height.

    “Stride parameters and energy measures were normalized to account for differences in subject body size and height.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. It may state two separate claims as one.

    Study design and conductcitation

    Walking speed was controlled at approximately 1.25 m/s by following the experimenter, with gentle speed corrections to avoid costs for artificial speed fluctuations.

    “Approximate speed of 1.25 m/s was controlled by following the experimenter, who walked according to GPS speed and attempted to make only gentle speed corrections, to avoid costs for artificial speed fluctuations [24].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [24] N Seethapathi, 2015, The metabolic cost of changing walking speeds is significant, implies lower optimal speeds for shorter distances, and increases daily energy estimates, Biol Lett source

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Premisecitation

    Overall energy expenditure is determined by carried load, movement speed, and grade or ground slope, each with readily identifiable effects.

    “Overall energy expenditure is also determined by other variables such as carried load, movement speed, and grade or ground slope [3–5], each with readily identifiable effects.”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites [3] KB Pandolf, 1976, Metabolic energy expenditure and terrain coefficients for walking on snow, Ergonomics source
    • cites [4] KB Pandolf, 1977, Predicting energy expenditure with loads while standing or walking very slowly, J Appl Physiol source
    • cites [5] RG Soule, 1978, Energy expenditure of heavy load carriage, Ergonomics source

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Premisecitation

    Walking speed normally fluctuates slightly, with correlated fluctuations in stride length, consistent with the preferred stride length relationship.

    “Walking speed normally fluctuates slightly [29], with correlated fluctuations in stride length [16] consistent with the preferred stride length relationship [30].”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites [29] SH Collins, 2013, Two independent contributions to step variability during over-ground human walking, PloS One source
    • cites [16] LV Ojeda, 2015, Influence of contextual task constraints on preferred stride parameters and their variabilities during human walking, Med Eng Phys source
    • cites [30] DW Grieve, 1968, Gait patterns and the speed of walking, Biomed Eng

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductnone given

    Subjects provided written informed consent before the experiment.

    “Subjects provided written informed consent before the experiment.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim.

    Premisecitation

    Body-worn accelerometer signals can be correlated with energy expenditure.

    “whose signals can be correlated with energy expenditure [e.g., 8–12]”
    Quote in the paperSubstantial claim0 of 5 sources checked
    • cites [8] AG Brooks, 2005, Predicting walking METs and energy expenditure from speed or accelerometry, Med Sci Sports Exerc source
    • cites [9] SE Crouter, 2006, A novel method for using accelerometer data to predict energy expenditure, J Appl Physiol Bethesda Md 1985 source
    • cites [10] EE Kuffel, 2011, Validity of estimating minute-by-minute energy expenditure of continuous walking bouts by accelerometry, Int J Behav Nutr Phys Act source
    • cites [11] HJ Montoye, 1983, Estimation of energy expenditure by a portable accelerometer, Med Sci Sports Exerc source
    • cites [12] C Tudor-Locke, 2002, Utility of pedometers for assessing physical activity: convergent validity, Sports Med Auckl NZ source

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductnone given

    The study was approved by the University of Michigan Health Sciences Institutional Review Board (HUM00020554).

    “The study was approved by the University of Michigan Health Sciences Institutional Review Board (HUM00020554).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Study design and conductnone given

    Subjects walked on each surface in random order for 8 minutes.

    “Subjects walked on each surface, presented in random order, for 8 minutes.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim.

    Study design and conductreasoning

    Only the last 3 minutes of data from each surface were used for metabolic energy expenditure to allow time to reach steady-state.

    “To allow time to reach steady-state, only the last 3 minutes of data from each surface were used for metabolic energy expenditure.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Interpretationreasoning

    There were small changes in stride length and speed with terrain, which may be attributable in part to imperfectly controlled walking speed rather than the terrain itself.

    “There were also small changes in stride length and speed with terrain, which may be attributable in part to imperfectly controlled walking speed rather than the terrain itself.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim.

    Study design and conductnone given

    All surfaces were selected to have very little elevation change, in terms of visible undulations, total change (maximum net grade of 0.96% on Gravel), and cross-slope.

    “All of the surfaces were selected to have very little elevation change, in terms of visible undulations, total change (maximum net grade of 0.96% on Gravel), and cross-slope.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Premisecitation

    Typical terrain coefficient values for relative gross metabolic cost compared to treadmill walking are 1.0 for blacktop, 1.2 for light brush, 1.5 for heavy brush, and 2.1 for loose sand.

    “Typical values are 1.0 for blacktop surface, 1.2 for light brush, 1.5 for heavy brush, and 2.1 for loose sand [2].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [2] RG Soule, 1972, Terrain coefficients for energy cost prediction, J Appl Physiol source

  • Probably not a substantial claim. It may state two separate claims as one.

    Premisecitation

    Shoe-mounted inertial measurement units can reconstruct the foot's path in space and placement on ground.

    “Yet another possibility is to use shoe-mounted inertial measurement units (combining accelerometers and gyroscopes) to reconstruct the foot’s path in space and placement on ground [14,15].”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [14] L Ojeda, 2007, Non-GPS navigation for security personnel and first responders, J Navig source
    • cites [15] JR Rebula, 2013, Measurement of foot placement and its variability with inertial sensors, Gait Posture source

  • It may state two separate claims as one.

    Interpretationown data

    More challenging terrain caused increases in virtual ground clearance and in the variability of most measures, for example of lateral swing motion.

    “Notably, more challenging terrain caused increases in virtual ground clearance and in the variability of most measures, for example of lateral swing motion.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. It may state two separate claims as one.

    Interpretationnone given

    IMU-based measures are unlikely to replicate the accuracy of a portable respirometry system, but IMUs are less obtrusive and easier to wear, and may still yield data informative of metabolic cost.

    “Of course, IMU-based measures are unlikely to replicate the accuracy of a (portable) respirometry system, but IMUs are less obtrusive and easier to wear, especially in real-world conditions, and may still yield data informative of metabolic cost.”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Study design and conductreasoning

    Stride length was not explicitly controlled and differed slightly with terrain.

    “Walking speed also generally determines stride length [16,29,36], which was not explicitly controlled and differed slightly with terrain.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Interpretationreasoning

    A non-zero stride height would generally be expected to indicate how much the body is lifted or lowered against gravity, and therefore drive energy expenditure.

    “Even though elevation changes were modest on the terrains studied here, a non-zero stride height would generally be expected to indicate how much the body is lifted or lowered against gravity, and therefore drive energy expenditure.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Interpretationreasoning

    Uneven ground appears to disrupt gait to a substantial degree, and would be expected to require substantial active stabilization.

    “Uneven ground appears to disrupt gait to a substantial degree, and would be expected to require substantial active stabilization.”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Interpretationown data

    These results illustrate the importance of interdependencies among stride parameters, and the complex relationship of cost to gait parameters.

    “These results illustrate the importance of interdependencies among stride parameters, and the complex relationship of cost to gait parameters.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Study design and conductnone given

    Foot path estimation relies on the foot being nearly stationary at some point during stance, which may not occur for every stride on softer terrains such as Woodchips.

    “Our foot path estimation relies on the foot being nearly stationary at some point during stance, which may not occur for every stride on softer terrains such as Woodchips.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it. The paper may give no reason why it holds. It may state two separate claims as one.

    Interpretationnone given

    Each location in the world, whatever its classification, may have unique effects on gait, that may nonetheless be quantifiable.

    “Each location in the world, whatever its classification, may have unique effects on gait, that may nonetheless be quantifiable.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Interpretationreasoning

    IMU-derived foot paths are neither absolute nor comprehensive measures.

    “Thus, IMU-derived foot paths are neither absolute nor comprehensive measures.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Interpretationreasoning

    The barely significant correlation of stride length with energy cost could be due in part to an actual effect and in part to imperfect experimental control of speed.

    “By itself, stride length was a barely significant correlate of energy cost (Table 2), which could be due in part to an actual effect, and in part to imperfect experimental control of speed.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Interpretationown data

    A few stride measures, such as stride and swing foot variability, can directly measure a terrain's effect on gait, and even discriminate among terrains.

    “In contrast, a few stride measures, such as stride and swing foot variability (Figs 5 and 6) can directly measure a terrain’s effect on gait, and even discriminate among terrains.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductown data

    The analysis was based on a relatively small number of summary measures.

    “We based our analysis on a relatively small number of summary measures, but a more intensive approach might be to instead use the actual foot path trajectories directly, including both translation and orientation data.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. It may be too vague to check against evidence.

    Interpretationreasoning

    Foot-mounted IMUs appropriately meet the trade-off between data quantity and convenience and practicality for real-world usage.

    “We find that foot-mounted IMUs appropriately meet the trade-off between data quantity and convenience and practicality for real-world usage.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Premisecitation

    Step width appears to change on uneven terrain and could contribute to energy cost.

    “In addition, inertial data cannot readily discern step width, which also appears to change on uneven terrain [7] and could contribute to energy cost [18].”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [7] AS Voloshina, 2013, Biomechanics and energetics of walking on uneven terrain, J Exp Biol source
    • cites [18] JM Donelan, 2001, Mechanical and metabolic determinants of the preferred step width in human walking, Proc R Soc Lond B source

  • Probably not a substantial claim. It may state two separate claims as one.

    Interpretationnone given

    Foot motion exhibits a continuous correlation with energy expenditure, which could potentially have predictive applications.

    “It also exhibits a continuous correlation with energy expenditure, which could potentially have predictive applications.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper may give no reason why it holds. It may state two separate claims as one.

    Interpretationnone given

    Foot motion is highly sensitive to the type of terrain, and has a discrete ability to categorize or discriminate terrains objectively.

    “Although foot motion hardly encompasses all of the gait adaptations for terrain, it is highly sensitive to the type of terrain, and has a discrete ability to categorize or discriminate terrains objectively.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Interpretationreasoning

    Gait measures are unlikely to discriminate better than visual observation, but they do offer continuous quantification of a terrain's effects.

    “Gait measures are unlikely to discriminate better than visual observation, but they do offer continuous quantification of a terrain’s effects.”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Interpretationreasoning

    The overall effect is that uneven ground leads to uneven foot motion and uneven steps.

    “Aggregating these various contributions, the overall effect is that uneven ground leads to uneven foot motion and uneven steps.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim.

    Study design and conductreasoning

    The study attempted to control walking speed at a fixed value across terrains, with e.g. 0.5% speed difference between Woodchips and Sidewalk.

    “Our interest here was in factors other than speed, which we therefore attempted to control at fixed value across terrains (e.g. 0.5% speed difference between Woodchips and Sidewalk).”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Premisecitation

    Variability in stride length and timing and fluctuations in speed have been reported to affect metabolic cost, perhaps due to the effort of varying gait.

    “Variability in stride length and timing [37] and fluctuations in speed [24] have been reported to affect metabolic cost, perhaps due to the effort of varying gait.”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [37] SM O’Connor, 2012, Energetic cost of walking with increased step variability, Gait Posture source
    • cites [24] N Seethapathi, 2015, The metabolic cost of changing walking speeds is significant, implies lower optimal speeds for shorter distances, and increases daily energy estimates, Biol Lett source
49 of 49

  • The paper relies on this without support.

    There is presently scant understanding of how terrain affects a person’s actual movements and actions, which are the ultimate determinants of energy expenditure.

    “But aside from this overall effect, there is presently scant understanding of how terrain affects a person’s actual movements and actions, which are the ultimate determinants of energy expenditure.”

  • The paper relies on this without support.

    It is challenging to determine the biomechanical adaptations for different terrains.

    “It is challenging to determine the biomechanical adaptations for different terrains.”

  • The paper relies on this without support.

    Traditional laboratory measures are difficult to obtain outdoors.

    “But such laboratory measures are difficult to obtain outdoors.”

  • The paper relies on this without support.

    The details of actual surface variations were unknown.

    “Of course, the details of actual surface variations were unknown”
4 of 4

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Walking outdoors on relatively simple natural surfaces (e.g. grass, gravel, woodchips) can increase walking costs by up to 27% owing to factors such as stride variability and height of swing foot clearance.

    “Walking outdoors on relatively simple natural surfaces (e.g. grass, gravel, woodchips) can increase walking costs by up to 27% owing to factors such as stride variability and height of swing foot clearance (Kowalsky, Rebula, Ojeda, Adamczyk, & Kuo, 2021)”

    From Forest terrains influence walking kinematics among indigenous Tsimane of the Bolivian Amazon (2022) · cites this paper as Kowalsky et al., 2021

    The Results (page unavailable) report that woodchips had a 27% higher net metabolic rate than sidewalk, and the Discussion links stride variability and virtual foot clearance with energy cost. But the paper reports correlations and possible gait adaptations, not that those factors caused the increase; the claim’s “owing to” wording is stronger than the evidence. (LLM, conf=0.86)

    Experimental — accuracy not yet measured

  • Only Matthis et al. (2018) and Kowalsky et al. (2021) have used IMUs to record walking on complex natural terrains outdoors, and neither report joint kinematics in their study.

    “To our knowledge, only Matthis et al. (2018) and Kowalsky et al. (2021) have used IMUs to record walking on complex natural terrains outdoors, and neither report joint kinematics in their study.”

    From Forest terrains influence walking kinematics among indigenous Tsimane of the Bolivian Amazon (2022) · cites this paper as Kowalsky et al., 2021

  • The difference in energy cost between forest walking and open field walking would probably be greater than the 24% increase in cost of transport measured by Kowalsky et al. (2021) for participants walking on woodchips vs. a dirt path.

    “However, the difference would probably be greater than the 24% increase in cost of transport measured by Kowalsky et al. (2021) for participants walking on woodchips vs. a dirt path, because avoiding natural obstacles and maintaining balance in the forest understory would probably require more substantial adjustments to walking kinematics than does walking on woodchips.”

    From Forest terrains influence walking kinematics among indigenous Tsimane of the Bolivian Amazon (2022) · cites this paper as Kowalsky et al., 2021

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Stride-height variability is called terrain-sensitive despite a non-significant test

    Major

    The discussion names stride-height variability among the most notable terrain sensitivities, but Table 1 reports P = 0.284 for it and leaves its terrain-significance indicator blank.

  • Stride-width variability falls outside the stated R2 range

    Minor

    The stated top four correlates include stride-width variability and are said to have adjusted R2 values of 0.29-0.38. Table 2 instead gives stride-width variability an R2 of 0.14.

  • Stride-width variability is called terrain-sensitive despite a non-significant test

    Minor

    The discussion names stride-width variability among the most notable terrain sensitivities, but Table 1 reports P = 0.423 for it and leaves its terrain-significance indicator blank.

  • Lateral-swing variability is called non-significant despite P = 0.00937

    Minor

    The text says lateral swing variability was not significantly correlated with metabolic rate, but Table 2 marks its regression significant and reports P = 0.00937, below the stated 0.05 threshold.

4 of 4

Citations

Cites

Flagged first, then by citations

Research works printed but not linked: 3

  1. AR Wu, 2015, Energetic tradeoffs of foot-to-ground clearance during swing phase of walking
  2. DW Grieve, 1968, Gait patterns and the speed of walking, Biomed Eng, 3, 119
  3. JC Russ, 2013, Fractal surfaces

Cited by

Sorted by citations

Related

Metadata

Authors
Kowalsky, D. B., Rebula, J. R., Ojeda, L. V., Adamczyk, P. G., & Kuo, A. D.
Affiliations
University of Michigan (US)
ORCID
Kuo, Arthur D.
Venue
PLOS One
Publisher
Public Library of Science (PLoS)
Published
2021-01-13
Volume
16
Issue
1
Pages
e0228682
ISSN
1932-6203 (ISSN-L)
DOI
10.1371/journal.pone.0228682
PMC
PMC7806134
OpenAlex
W3118943433
Semantic Scholar
660c9f976920ea34525f9f010f69b7ee207ed3aa
Cited by
106
References
42
Citations by year
2026: 21 · 2025: 25 · 2024: 22 · 2023: 16 · 2022: 12 · 2021: 5
FWCI
17.77
Type
journal article
Language
en
Open access
Licence
Locations
  • PLoS ONE (publishedVersion, cc-by)
  • bioRxiv (Cold Spring Harbor Laboratory) (submittedVersion, cc-by-nc)
  • Europe PMC (PubMed Central) (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • RePEc: Research Papers in Economics (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
  • Congressionally Directed Medical Research Programs (W81XWH-09-2-0142)
  • National Institute on Aging (AG030815)
  • Office of Naval Research (ETOWL)
Topics
  • Health Sciences › Health Professions › Physical Therapy, Sports Therapy and Rehabilitation › Balance, Gait, and Falls Prevention
  • Physical Sciences › Engineering › Biomedical Engineering › Lower Extremity Biomechanics and Pathologies
  • Physical Sciences › Engineering › Biomedical Engineering › Gait Recognition and Analysis
Keywords
human walking, outdoor terrain, metabolic cost, gait variability, stride-to-stride variability, foot trajectory, inertial measurement unit, oxygen consumption, uneven terrain, wood chips, gravel, Poaceae, dirt, wearable sensors
MeSH terms
Adolescent, Adult, Biomechanical Phenomena, Energy Metabolism, Female, Foot, Gait, Humans, Male, Middle Aged, Oxygen Consumption, Walking, Young Adult

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

A language model drew the claims from the paper on 2026-11-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-11-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review