CoChat

Diagnosis of disease affecting gait with a body acceleration-based model using reflected marker data for training and a wearable accelerometer for implementation

Open access

Published version of a preprint

The DOI matches its Crossref record; 55 of the 55 references it cites were found in a registry or an index.

Abstract

This paper demonstrates the value of a framework for processing data on body acceleration as a uniquely valuable tool for diagnosing diseases that affect gait early. As a case study, we used this model to identify individuals with peripheral artery disease (PAD) and distinguish them from those without PAD. The framework uses acceleration data extracted from anatomical reflective markers placed in different body locations to train the diagnostic models and a wearable accelerometer carried at the waist for validation. Reflective marker data have been used for decades in studies evaluating and monitoring human gait. They are widely available for many body parts but are obtained in specialized laboratories. On the other hand, wearable accelerometers enable diagnostics outside lab conditions. Models trained by raw marker data at the sacrum achieve an accuracy of 92% in distinguishing PAD patients from non-PAD controls. This accuracy drops to 28% when data from a wearable accelerometer at the waist validate the model. This model was enhanced by using features extracted from the acceleration rather than the raw acceleration, with the marker model accuracy only dropping from 86 to 60% when validated by the wearable accelerometer data.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/s41598-023-50727-8

Cites

55 of 55 references verified0 citation errors · 0 not checkable

Statistics

Figures

Figure 1 A summary of the two data sources used in this paper. The column on the left displays the characteristics of the time-series data obtained from the ActiGraph GT9X wearable accelerometer worn on the waist of the participants and sampled at 100 Hz. Meanwhile, the column on the right presents the traits of the marker data from different anatomical landmarks, including the sacrum, captured by infrared cameras and sampled at 60 Hz.
Figure 1 A summary of the two data sources used in this paper. The column on the left displays the characteristics of the time-series data obtained from the ActiGraph GT9X wearable accelerometer worn on the waist of the participants and sampled at 100 Hz. Meanwhile, the column on the right presents the traits of the marker data from different anatomical landmarks, including the sacrum, captured by infrared cameras and sampled at 60 Hz.CC BY · PubMed Central
Figure 2 Flowchart showing our two approaches to establish the best diagnostics framework. The first deals with the raw acceleration data (Part 1), and the second deals with gait characteristics extracted from marker data (Part 2). A maximum of four (the primary) gait features were extracted from most of the data. Due to the high consistency among multiple gait cycles, the marker data at the sacrum position provided 16 gait features. The foot carton in Part 2 marks the extra gait features. Recurrent neural networks (RNNs) that used the acceleration time-series data as the LSTM model were used in part 1. Typical machine learning methods, such as the SVM that use the discrete extracted gait features, were used in part 2. In each part, different paths (labeled from 1 to 7) were used to find the best implementation of the framework. Each path has a training and testing module. Using the marker data in a test module or the raw acceleration in training, marked by red, represents a limitation.
Figure 2 Flowchart showing our two approaches to establish the best diagnostics framework. The first deals with the raw acceleration data (Part 1), and the second deals with gait characteristics extracted from marker data (Part 2). A maximum of four (the primary) gait features were extracted from most of the data. Due to the high consistency among multiple gait cycles, the marker data at the sacrum position provided 16 gait features. The foot carton in Part 2 marks the extra gait features. Recurrent neural networks (RNNs) that used the acceleration time-series data as the LSTM model were used in part 1. Typical machine learning methods, such as the SVM that use the discrete extracted gait features, were used in part 2. In each part, different paths (labeled from 1 to 7) were used to find the best implementation of the framework. Each path has a training and testing module. Using the marker data in a test module or the raw acceleration in training, marked by red, represents a limitation.CC BY · PubMed Central
Figure 3 Different LSTM model performances were trained by raw acceleration data using data from markers mounted at the sacrum and the ASIS position and wearable accelerometers mounted near the waist. The F1 measure represents the precision and recall values.
Figure 3 Different LSTM model performances were trained by raw acceleration data using data from markers mounted at the sacrum and the ASIS position and wearable accelerometers mounted near the waist. The F1 measure represents the precision and recall values.CC BY · PubMed Central
Figure 4 Different SVM model performances with different training and testing combinations of marker data mounted near the sacrum and wearable accelerometers near the waist. Up to 16 extracted gait-based features from the marker data are used in developing some of those models. But only 4 features were extracted from the wearable accelerometer due to consistency issues in the gait cycles.
Figure 4 Different SVM model performances with different training and testing combinations of marker data mounted near the sacrum and wearable accelerometers near the waist. Up to 16 extracted gait-based features from the marker data are used in developing some of those models. But only 4 features were extracted from the wearable accelerometer due to consistency issues in the gait cycles.CC BY · PubMed Central
Figure 5 Graphic demonstration of gait cycles derived from the acceleration in the vertical direction obtained from the sacral marker trajectories (Av). S1 was derived from Gaussian Continuous Wavelet Transform (CWT) at a scale of 10 of vertical acceleration, and S2 was obtained from S1 differentiation. Initial contact was S1 local minima, and Final contact was S2 local maxima. Since we had consecutive steps, we assumed the first gait characteristics (like a step) as left (or right), then alternated values.
Figure 5 Graphic demonstration of gait cycles derived from the acceleration in the vertical direction obtained from the sacral marker trajectories (Av). S1 was derived from Gaussian Continuous Wavelet Transform (CWT) at a scale of 10 of vertical acceleration, and S2 was obtained from S1 differentiation. Initial contact was S1 local minima, and Final contact was S2 local maxima. Since we had consecutive steps, we assumed the first gait characteristics (like a step) as left (or right), then alternated values.CC BY · PubMed Central
Figure 6 The left plot shows the correlation between the x, y, and z-axis of the motion-based acceleration data. A sample of three-dimensional acceleration can be seen in the right plot.
Figure 6 The left plot shows the correlation between the x, y, and z-axis of the motion-based acceleration data. A sample of three-dimensional acceleration can be seen in the right plot.CC BY · PubMed Central
Figure 7 Box plot distributions of the extracted gait characteristics from the vertical acceleration of the motion-based data. The graphs demonstrate that gait characteristics have some distinguishable differences between patients with PAD and healthy controls. According to the box plots of the gait characteristics, there is a high level of separation between PAD and healthy controls in some variability and asymmetric features like 'SwingTime SV', 'SwingTime MV', 'StrideTime SV', and 'StepTime SA'.
Figure 7 Box plot distributions of the extracted gait characteristics from the vertical acceleration of the motion-based data. The graphs demonstrate that gait characteristics have some distinguishable differences between patients with PAD and healthy controls. According to the box plots of the gait characteristics, there is a high level of separation between PAD and healthy controls in some variability and asymmetric features like 'SwingTime SV', 'SwingTime MV', 'StrideTime SV', and 'StepTime SA'.CC BY · PubMed Central
Figure 8 The SHAP summary plot for the Random Forest (RF) model was generated using all 16 features. The Y-axis represents the gait features, while the X-axis displays the mean absolute SHAP value. The plot shows the features sorted in descending order of importance, as determined by their mean absolute SHAP value. A SHAP value of 0 indicates that the feature does not impact the predicted output. A SHAP value of 0.2 indicates that the feature increases the predicted output by a moderate amount, while a SHAP value of 0.1 indicates that the feature increases the predicted output by a smaller amount.
Figure 8 The SHAP summary plot for the Random Forest (RF) model was generated using all 16 features. The Y-axis represents the gait features, while the X-axis displays the mean absolute SHAP value. The plot shows the features sorted in descending order of importance, as determined by their mean absolute SHAP value. A SHAP value of 0 indicates that the feature does not impact the predicted output. A SHAP value of 0.2 indicates that the feature increases the predicted output by a moderate amount, while a SHAP value of 0.1 indicates that the feature increases the predicted output by a smaller amount.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

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?

Self-check

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

The self-check needs the paper’s full text.

Citations

Cites

Show all 55 references

Research works printed but not linked: 1

  1. Brownlee, J. Long Short-Term Memory Networks With Python: Develop Sequence Prediction Models with Deep Learning (Machine Learning Mastery, 2017).

Cited by

See all 11

Metadata

Authors
Takallou, M. A., Fallahtafti, F., Hassan, M., Al-Ramini, A., Qolomany, B., Pipinos, I., Myers, S., & Alsaleem, F.
Affiliations
  • University of Nebraska–Lincoln (US)
  • University of Nebraska at Omaha (US)
  • University of Nebraska at Kearney (US)
  • Omaha VA Medical Center (US)
Corresponding author
Alsaleem, Fadi
Venue
Scientific Reports
Published
2024-01-11
Volume
14
Issue
1
Pages
1075-1075
ISSN
2045-2322 (ISSN-L)
OpenAlex
W4390751334
Cited by
11
References
55
Citations by year
2026: 4 · 2025: 5 · 2024: 2
FWCI
1.20
Type
journal article
Language
en
Open access
Licence
Locations
  • Scientific Reports (publishedVersion, cc-by)
  • Research Square (acceptedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • Lincoln (University of Nebraska) (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
NIH
Topics
  • Physical Sciences › Engineering › Biomedical Engineering › Gait Recognition and Analysis
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Context-Aware Activity Recognition Systems
  • Health Sciences › Health Professions › Physical Therapy, Sports Therapy and Rehabilitation › Balance, Gait, and Falls Prevention
Keywords
peripheral arterial disease, gait analysis, body acceleration, wearable accelerometers, early diagnosis, feature extraction
MeSH terms
Acceleration, Wearable Electronic Devices, Gait, Humans, Peripheral Arterial Disease, Accelerometry

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.

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

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Diagnosis of disease affecting gait with a body acceleration-based model using reflected marker data for training and a wearable accelerometer for implementation. https://app.cochat.ai/papers/10.1038/s41598-023-50727-8
The paper
Takallou, M. A., Fallahtafti, F., Hassan, M., Al-Ramini, A., Qolomany, B., Pipinos, I., Myers, S., & Alsaleem, F. (2024). Diagnosis of disease affecting gait with a body acceleration-based model using reflected marker data for training and a wearable accelerometer for implementation. Scientific Reports. https://doi.org/10.1038/s41598-023-50727-8

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.