Mechanical and metabolic consequences of sagittal trunk lean angle in walking – A dynamic walking perspective
Data files
Sep 01, 2025 version files 2.71 GB
-
Dryad_Submission_V2.zip
2.71 GB
-
README.md
10.81 KB
Abstract
Bipedal walking requires a balance of muscle work and energy losses, with models indicating that powering gait with ankle push-off is more energetically economical than powering with hip joint work. This study investigates how varying trunk lean angle affects joint mechanics and energy expenditure during walking. We hypothesized that leaning forward would increase hip work, reduce ankle push-off, and increase energy consumption, and leaning backward would have complementary effects. Healthy young adults walked at 1.3 m/s while adjusting their trunk angles from backward 15° to forward 60° using visual feedback from a chest-mounted inertial motion sensor. Center of mass mechanics and lower-body joint mechanics were estimated using motion capture and force treadmill measurements, alongside metabolic rate using respirometry. With forward trunk lean, center of mass (COM) work became more negative during collisions, increased in the middle of stance phase, and was reduced in push-off. At the joint level, forward trunk lean led to increasing stance-phase hip moment and hip work, while ankle work decreased for moderate trunk angles. The early vertical ground reaction force peak and loading rate also increased with forward trunk lean. Backward trunk lean led to reduced hip work, increased ankle work, and increased push-off work. Metabolic rate was minimized in the 0° condition and increased with trunk lean in either direction. Trunk lean significantly impacts lower-limb mechanics and energy consumption, with a trade-off between hip and ankle work, suggesting potential applications for improving walking in populations with diminished push-off, such as older adults.
This dataset contains data from n=10 healthy young adults. They walked with trunk lean angles of backwards 15 degrees, neutral, and forward 15, 30, 45, 60 degrees. Here we included the *.c3d motion capture files, an excel file explaining what all the conditions were named, the *.cmz visual 3D workspace for each of the 10 subjects, the Excel files of the metabolic data for all of the subjects, a summary Excel file of the metabolic rates for all the subjects, a MATLAB *.mat file containing the dimensionless and dimensioned results after processing (including both the averages for each condition and a matrix of the averages of each condition by subject), and MATLAB code used to process the results.
Dataset DOI: 10.5061/dryad.95x69p8w3
Description of the data and file structure
Human metabolic rate and lower-body biomechanics of walking with varying trunk lean angle.
This data set contains gait analysis data (motion capture, n=10) and energy expenditure data (n=12) for healthy young adults walking on an instrumented treadmill at a speed of 1.3 m/s with trunk lean angle targets of -15, 0, 15, 30, 45, and 60 degrees forward relative to their normal walking posture.
Participants were allocated numbers based on the overall dataset for all code and participants
All participants (1-12) were analyzed for metabolic rate.
Participant numbers analyzed for motion capture include: 1, 2, 3, 5, 6, 7, 8, 9, 10, and 12. Subjects 4 and 11 were excluded from the motion capture data set due to poor motion capture data quality.
There is also a data set of 10 separate subjects with metabolic rate only, from a Pilot Study preceding the main experiment. Conditions and test procedures were the same, but no motion capture data were recorded.
Files and variables
File: Dryad_Submission_V2.zip
Description:
Files in the main folder
This folder contains motion capture data and processing code in MATLAB format. These data are the exported signals resulting from analysis of raw motion/force data in Visual3D.
Trunk_Lean_Variables.mat: The resulting dimensioned and dimensionless variables referenced in the paper and in the table. Split out by condition and by subject: Rows = Different subjects; Columns = Trunk Condition (e.g. column 1 = -15 deg; column 2 = 0 deg (neutral trunk); column 3 = +15 deg; column 4 = +30 deg; etc). 10 subjects included.
Trunk Lean Variables Explanations.csv: Descriptions of variables in the “Trunk_Lean_Variables.mat” file.
Metabolic_Data folder
MetabolicsAllSubjects_FINAL.xlsx: “HeightWeight” tab = Height, weight, and leg length labeled by subject, “Subject#” tab = summary of data for that particular subject. 12 subjects included.
PilotStudy_metabolics.xlsx: Data from a pilot study (10 subjects), includes, height, weight, and energy expenditure data for all the subjects.
Rawfiles/MRTL##.xlsx: “Energy consumption Rate” tab = Energy consumption rate for the last 2 minutes of every trunk lean condition for subject ##, reported in 10-second increments.
“Data” tab = Results of the COSMED K5 Output for each subject from the time the mask was put on at the beginning of the collection to the time the mask was taken off after all the trials were finished.
- t (seconds): time from the start of the collection in 10-second increments
- VO2 (mL/min): volume rate of oxygen consumed by the body
- VCO2 (mL/min) volume rate of carbon dioxide produced by the body
- EEh (kcal/h): Energy expenditure per hour calculated from VO2 and VCO2
- “MRTL” refers to “Metabolic Rate Trunk Lean”, an internal name for the study.
Visual3D_rawfiles folder
MRTL_JEB_summary.xlsx: A description of the subjects used for motion capture analysis (n=10) with their IDs and corresponding functional joint center trials, along with subject height, leg length, and mass.
MRTL## folder: All the trunk lean conditions (.c3d), static trials, and functional joint center trials (.c3d) for subject ##.
How to Open .c3d Files:
The .c3d file format (C3D: Coordinate 3D) is a standard used for storing motion capture data. Opening these files requires specialized software capable of reading biomechanical or motion capture formats.
If you do not have access to Visual3D, a commercial software for analyzing C3D files, several open-source alternatives are available:
OpenSim – A free and widely used platform for modeling and simulation of human movement.
EZC3D – An open-source C++/Python/Matlab library for reading and writing C3D files. It can be accessed here: https://www.mathworks.com/matlabcentral/fileexchange/69372-ezc3d
Additional tools and resources for working with C3D files can be found on the C3D website: https://www.c3d.org/tools.html
MRTL_workspaces: Assembled workspace (.cmz) for each subject with functional joint center trials and trunk lean conditions labeled (CMZ files require Visual3D software to access).
Visual 3D pipelines/HAT_COG.v3s: A Visual 3D pipeline that estimates the hip midpoint and the center of mass of the head, arms, and trunk (HAT) segments
Visual 3D pipelines/TRUNKLEAN_PIPELINE_V3.v3s: This Visual 3D pipeline estimates lower-body biomechanics, including joint angles, joint moments, joint powers, intersegmental ankle power, and additional relevant kinematic and kinetic metrics. The pipeline is used to process motion capture data and extract key variables for biomechanical analysis.
Visual3D_MatfileOutput folder
MRTL##_recalc.mat: A MATLAB file of the kinetic results from Visual 3D for subject ##.
MRTL##_HAT_COG.mat: A MATLAB file of the center of gravity of the head, arms, and trunk, as well as the midpoint of the hip, for subject ##. Used to find the moment arm length of the upper body.
Marker set (Additional R designates right side, Additional L designates left side)
- 1st metatarsal head (MT_2)
- 5th metatarsal head (MT)
- Heel (HEE)
- Medial malleoli (MAN)
- Lateral malleoli (LAN)
- Medial femoral epicondyle (MKN)
- Lateral femoral epicondyle (LKN)
- Anterior superior iliac spine (ASI)
- Posterior superior iliac spine (PSI)
- Greater trochanter (GT)
- Acromion process (SH)
- Sternum (STER)
- C7 vertebrae (C7)
- Medial elbow epicondyle (MEL)
- Lateral elbow epicondyle (LEL)
- Styloid process of the radius (LW, MW)
- The head of the 3rd metacarpal (HAN)
- Two on the anterior aspect of the head (FH)
- Two on the posterior head (BH)
- 1 on the top of the head (MHE)
- Cluster of 4 markers on rigid plates were secured to the:
Thigh (TH)
Shank (SHI) - Clusters of 3 markers on each foot (FT)
(The medial metatarsal, medial malleolus, medial femoral epicondyle, and greater trochanter were for static poses only and were removed for motion trials)
Code/software
List of the code used
Files in the main folder
This folder contains motion capture data and processing code in MATLAB format. These data are the exported signals resulting from analysis of raw motion/force data in Visual3D.
Trunklean_JEB_processing_SampleCode.m: Sample code of how to load in a subject, find gait event indices (right heel strike RHS, right toe-off RTO, left heel strike LHS, left toe-off LTO), find the first 7 good strides for analysis, and process/plot a few basic variables (ground reaction force GRF, Ankle Power, Ankle Work, True Trunk Lean Angle, center of mass (COM) Work Rate, COM velocity Hodograph, and Metabolic Rate).
NOTE: For all subjects, variables{1,1} = data for -15 deg condition, variables{2,1} = data for +15 deg condition, variables{3,1} = +30 deg, variables{4,1} = +45 deg, variables{5,1} : +60 deg.
For subjects 1 and 5, variables{8,1} = data for 0 deg condition (i.e., same trunk angle as normal walking). For all other subjects, variables{7,1} = data for zero deg condition.
Trunklean_JEB_processing_everything.m: Full code used to process variables.
PilotStudy_anyfit.m: Matlab code that plots the metabolic results from the pilot study. “anyfit” refers to a custom version of a linear mixed model regression, allowing a different offset in the dependent variable for each subject.
MATLAB functions
findEventIndicesForWalkingOnTreadmill.m: This function takes in the right and left foot ground reaction forces and outputs the right heel strike, left heel strike, right toe off, and left toe off
anyfit_MRTL.m: This function performs a multi-term mixed-effects linear regression of arbitrary form, returning coefficients of each term with statistical characteristics of the fit. It takes in an x-variable (independent or predictor variable), y-variable (dependent or outcome variable) and a MATLAB expression defining the form of the basis functions of a linear regression. It outputs mixed effect model coefficients as well as statistics.
calculateComWorkRateForSteadyStateGait.m: This function takes in the right and left foot ground reaction forces, the average walking speed, and time duration and outputs the right COM work rate, left COM work rate, and the velocity of the COM using standard Center of Mass mechanics procedures.
removeBadStridesFromWalkingOnTreadmill.m: Function that looks at strides and removes the bad ones (e.g., cross-over steps with contact on both treadmill force plates).
interpGaitCycle.m: Resamples gait cycle (or other variable) to stretch across N points.
findEventIndicesForWalkingOnTreadmill_Hodograph.m: Finds heel strike (HS) and toe off (TO) events from walking on a split-belt force treadmill. Input right and left ground reaction forces in Nx3 array and sampling rate (optional). Outputs indices for right HS, left HS, right TO and left TO.
List of the software used
Force-instrumented treadmill (Bertec Corp, Columbus, OH, USA; 1900 Hz)
Motion capture system (Motion Analysis Corporation, Rohnert Park, CA, USA; 190 Hz)
Cortex software for motion and force data (Motion Analysis Corporation, Rohnert Park, CA, USA).
Indirect calorimetry system (K5, COSMED, Pavona, RM, Italy)
COSMED OMNIA software (COSMED, Pavona, RM, Italy).
Inertial measurement unit (IMU) for trunk lean feedback (Opal, APDM Inc., Portland, OR, USA)
APDM Motion Studio control software for real-time streaming (APDM Inc., Portland, OR, USA)
Visual 3D v.6 for kinetics data analysis, joint angles, joint moments, joint powers, COM metrics, GRFs, and spatiotemporal variables (C-Motion Inc., Germantown, MD, USA)
MATLAB v.2019a - v. 2022a for real-time display of IMU angle on a laptop, data processing, data analysis, statistical analysis (The Mathworks, Natick, MA, USA)
Access information
Other publicly accessible locations of the data:
-
Selected results and visualizations derived from this dataset (e.g., tables and figures) are presented in the associated publication.
Roembke, R. A., Hernandez-Hernandez, S., Adamczyk, P. G., (2025). Mechanical and Metabolic Consequences of Sagittal Trunk Lean Angle in Walking – A Dynamic Walking Perspective. The Journal of Experimental Biology.
Human subjects data
We received explicit consent from your participants to publish the de-identified data in the public domain. We deidentified the data by assigning the subjects sequential integers. There is no further identifying information besides age, weight, and sex.
Human subjects (healthy young adults) walked on a treadmill at 1.3 m/s with their trunk inclined to different controlled sagittal angles (-15, 0, 15, 30, 45, 60 degrees forward of their customary walking posture). We measured oxygen consumption, motion capture, and ground reaction forces, and computed energy expenditure and biomechanical outcomes to determine the relationship between hip and ankle work across conditions and other accompanying biomechanical phenomena. For further details refer to the methods in the associated paper.