pith. sign in

arxiv: 2501.04721 · v1 · submitted 2025-01-02 · 📊 stat.AP · cs.LG· physics.med-ph

A Shape-Based Functional Index for Objective Assessment of Pediatric Motor Function

Shashwat Kumar , Arafat Rahman , Robert Gutierrez , Sarah Livermon , Allison N. McCrady , Silvia Blemker , Rebecca Scharf , Anuj Srivastava
show 1 more author
Laura E. Barnes
This is my paper

Pith reviewed 2026-05-23 06:23 UTC · model grok-4.3

classification 📊 stat.AP cs.LGphysics.med-ph
keywords motor function assessmentwearable sensorsshape-based PCADuchenne muscular dystrophyspinal muscular atrophypartial least squarespediatric kinematicsneuromuscular disorders
0
0 comments X

The pith

Shape-based PCA on wearable sensor trajectories from children with DMD and SMA extracts kinematic patterns whose projections, via partial least squares, correlate at r = 0.78 with muscle fat infiltration, Brooke scores, and age-related loss

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents a data-driven method to quantify motor function objectively during daily activities in 19 DMD, 9 SMA, and 13 control children using wearable sensors. Shape-based principal component analysis first aligns trajectories to remove limb-length and speed confounds, surfacing patterns such as asymmetry and speed variation. Projections onto these components are then linked through partial least squares to clinical variables, isolating one covariation mode that reaches canonical correlation r = 0.78 (95 % CI [0.34, 0.94]) with fat infiltration, Brooke score, and degenerative change. The resulting index is proposed for home deployment to track treatment response longitudinally. A sympathetic reader would value an objective, repeatable measure that can replace or supplement subjective clinical ratings in growing children.

Core claim

Shape-based principal component analysis of pediatric movement trajectories, after alignment for limb length and speed, yields distinct kinematic patterns; the projections of these patterns onto partial least squares components identify a single covariation mode with canonical correlation r = 0.78 to muscle fat infiltration, the Brooke motor-function score, and age-related degenerative changes, thereby defining a novel, deployable motor-function index.

What carries the argument

Shape-based principal component analysis (aligns trajectories and isolates kinematic modes such as asymmetry) followed by partial least squares regression to extract the canonical correlation with clinical variables.

If this is right

  • Both DMD and SMA groups contain individuals whose motor patterns are statistically indistinguishable from healthy controls.
  • SMA patients show stronger expression of the asymmetry pattern than DMD patients or controls.
  • The index can be computed from home-recorded daily movements without laboratory equipment.
  • Longitudinal use of the index could quantify treatment efficacy more finely than discrete Brooke scores.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same pipeline could be tested on other pediatric movement disorders to check whether the asymmetry mode generalizes.
  • If the index predicts future decline, it might serve as an early surrogate endpoint in clinical trials.
  • Combining the kinematic index with genetic or imaging biomarkers could improve individual prognosis.

Load-bearing premise

The aligned kinematic patterns reflect biologically meaningful motor impairment rather than sensor-placement artifacts or sampling noise in the small heterogeneous cohort.

What would settle it

A replication study on an independent cohort of at least 50 patients in which the reported canonical correlation falls below the lower bound of the given 95 % CI would falsify the index's claimed utility.

Figures

Figures reproduced from arXiv: 2501.04721 by Allison N. McCrady, Anuj Srivastava, Arafat Rahman, Laura E. Barnes, Rebecca Scharf, Robert Gutierrez, Sarah Livermon, Shashwat Kumar, Silvia Blemker.

Figure 1
Figure 1. Figure 1: Overview of the study and the proposed shape analysis pipeline. Wearable sensors capture IMU signals from participants performing activities of daily living. This data is combined with shape analysis and external assessments to develop a canonical index of motor function. The emergence of wearable-based motion assessments presents a promising solution to these challenges. By embedding sensors into everyday… view at source ↗
Figure 2
Figure 2. Figure 2: A simulated illustration of the alignment of arm curls. (a) An example of an arm curl. (b) Temporal rate or warping function of this arm curl. (c) An example of misaligned arm curls. (d) Functions after alignment. (1) µˆn: the overall mean shape of the given curves, (2) {γ ∗ i }: the phases that align individual curves to the mean shape, and (3) {β˜ i = βi ◦ γ ∗ i }: the set of aligned curves or amplitudes… view at source ↗
Figure 3
Figure 3. Figure 3: Results on performing curve registration and Fr´echet mean calculation with temporal matching. (a) Signals with only amplitude variability, (b) Warping functions, (c) Signals with amplitude and phase variability, (d) Signals after registration, (e) Reconstructed warping functions, (f) Euclidean and Shape mean. Note how the shape mean (blue) captures the symmetric shape better than the Euclidean mean (red).… view at source ↗
Figure 4
Figure 4. Figure 4: (a-d) Results on performing phase amplitude separation on healthy and (e-h) DMD/SMA cohorts. mean shape derived from healthy participants. This approach aims to highlight deviations from the healthy mean shape. Here, we observe a notable disparity between the peaks and valleys of the DMD/SMA cohort and the healthy mean. As depicted visually in Fig 4f, the DMD/SMA trajectories require substantial warping to… view at source ↗
Figure 5
Figure 5. Figure 5: (a-c) Vertical modes of variation obtained from Shape PCA on the curl data. (a) The first mode represents scaling, (b) the second asymmetry in motion while (c) the last represents noise. (d-f) Modes of variation obtained from knocking data. (d) The first mode represents scaling. (e) The second mode represents asymmetry in motion while (f) the last represents sensor noise [PITH_FULL_IMAGE:figures/full_fig_… view at source ↗
Figure 6
Figure 6. Figure 6: Interpretation of Vertical Principal Component 2 of arm curl (VPC2 Curl) in videos of 2 participants. The participants performed the upward motion of the arm curl more slowly than the downward motion, likely due to the resistance posed by gravity. Analyzing Cohort Differences In [PITH_FULL_IMAGE:figures/full_fig_p008_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Boxplots of some demographic variables along with important clinical measures and feature dimensions. (a) Age, (b) Brooke score, (c) Average Echogenicity (Avg Echo (gsv)), (d) Normalized Elbow Torque (NET (Nm/cm)), (e) VPC1 Curl (Speed), (f) VPC2 Curl (Asymmetry), (g) VPC1 Knock (Speed), and (h) VPC2 Knock (Asymmetry). asymmetry, is more pronounced in SMA compared to DMD and Healthy. This finding is intrig… view at source ↗
Figure 8
Figure 8. Figure 8: Pearson Cross-Correlation of different VPC modes with clinical measures for DMD (N=15), SMA (N=7), and Healthy (N=9). (a) Cross correlations for VPC1 Curl (Speed), and (b) VPC1 Knock (Speed) [PITH_FULL_IMAGE:figures/full_fig_p010_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Distribution of canonical correlations (first column) and coefficients. Our first canonical dimension has a median correlation of r = 0.78 (95% CI [0.34, 0.94]) with dimensions of muscle fat infiltration (Avg Echo), Brooke score, and Age-related degenerative changes. Speed of curl (VPC1 Curl) and knock (VPC1 Knock) have tighter spread in distribution than the asymmetry features (VPC2 Curl and VPC2 Knock). … view at source ↗
Figure 10
Figure 10. Figure 10: Comparison of different decomposition methods, (a-c) Shape PCA with alignment leads to much more interpretable modes of variation than (d-f) NMF, and (g-i) Functional PCA without alignment because of the phase variability [PITH_FULL_IMAGE:figures/full_fig_p011_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Relationship between Age and VPC1 Curl in DMD, SMA, and Healthy control groups. Here, colored lines represent the regression estimated conditional mean of each cohort, and points represent the VPC1 values of each participant. SMA (β = 2.530, corrected p = 0.002) cohorts, the positive slope coefficients indicate an age-related decline in the speed of curl, suggesting a loss of ability. Conversely, the Heal… view at source ↗
read the original abstract

Clinical assessments for neuromuscular disorders, such as Spinal Muscular Atrophy (SMA) and Duchenne Muscular Dystrophy (DMD), continue to rely on subjective measures to monitor treatment response and disease progression. We introduce a novel method using wearable sensors to objectively assess motor function during daily activities in 19 patients with DMD, 9 with SMA, and 13 age-matched controls. Pediatric movement data is complex due to confounding factors such as limb length variations in growing children and variability in movement speed. Our approach uses Shape-based Principal Component Analysis to align movement trajectories and identify distinct kinematic patterns, including variations in motion speed and asymmetry. Both DMD and SMA cohorts have individuals with motor function on par with healthy controls. Notably, patients with SMA showed greater activation of the motion asymmetry pattern. We further combined projections on these principal components with partial least squares (PLS) to identify a covariation mode with a canonical correlation of r = 0.78 (95% CI: [0.34, 0.94]) with muscle fat infiltration, the Brooke score (a motor function score), and age-related degenerative changes, proposing a novel motor function index. This data-driven method can be deployed in home settings, enabling better longitudinal tracking of treatment efficacy for children with neuromuscular disorders.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript introduces shape-based principal component analysis (PCA) applied to wearable sensor trajectories from 19 DMD, 9 SMA, and 13 control pediatric subjects to align movements for limb-length and speed variation and extract kinematic patterns. Projections onto the retained components are then combined via partial least squares (PLS) to identify a single covariation mode that achieves a canonical correlation r = 0.78 (95% CI [0.34, 0.94]) with muscle fat infiltration, Brooke score, and age; the resulting index is proposed as an objective, home-deployable motor-function measure.

Significance. If the reported correlation can be shown to generalize, the work would supply a concrete, sensor-based alternative to subjective clinical scores for tracking treatment response in pediatric neuromuscular disease. The explicit handling of pediatric confounders (limb length, speed) via shape alignment is a methodological strength that could transfer to other growth-related movement studies. The absence of out-of-sample validation, however, currently limits the strength of the deployability claim.

major comments (2)
  1. [Results (PLS covariation mode) and Abstract] Results (PLS covariation mode) and Abstract: the headline canonical correlation r = 0.78 is obtained by fitting PLS directly on the same n = 41 observations used to derive the shape-based principal components, with no cross-validation, permutation test, or held-out cohort reported. Consequently the quoted value and CI reflect an in-sample canonical correlation rather than an out-of-sample predictive performance for the proposed motor-function index.
  2. [Methods (alignment and normalization)] Methods (alignment and normalization): no quantitative validation, sensitivity analysis, or error propagation is supplied for the limb-length normalization step that precedes shape-based PCA. Without such checks it remains possible that residual alignment artifacts or sensor-placement effects are absorbed into the retained components and subsequently into the PLS mode.
minor comments (1)
  1. [Abstract] Abstract: the total sample size (n = 41) is not stated explicitly even though the subgroup sizes are given; adding the total would improve immediate readability.

Simulated Author's Rebuttal

2 responses · 1 unresolved

We thank the referee for the constructive comments. We address each major point below and have revised the manuscript to improve clarity and add supporting analyses where feasible.

read point-by-point responses

  1. Referee: [Results (PLS covariation mode) and Abstract] Results (PLS covariation mode) and Abstract: the headline canonical correlation r = 0.78 is obtained by fitting PLS directly on the same n = 41 observations used to derive the shape-based principal components, with no cross-validation, permutation test, or held-out cohort reported. Consequently the quoted value and CI reflect an in-sample canonical correlation rather than an out-of-sample predictive performance for the proposed motor-function index.

    Authors: We agree that the reported canonical correlation is an in-sample estimate. In the revision we have added a permutation test (10,000 permutations) to assess the statistical significance of the PLS mode and included the resulting p-value in the results. We have also revised the abstract, results, and discussion to explicitly state that the r=0.78 value and its CI are in-sample quantities, that the analysis is exploratory, and that independent validation in a larger cohort is required before claiming predictive performance or broad deployability. The wide CI is now highlighted as reflecting sample-size limitations. revision: yes

  2. Referee: [Methods (alignment and normalization)] Methods (alignment and normalization): no quantitative validation, sensitivity analysis, or error propagation is supplied for the limb-length normalization step that precedes shape-based PCA. Without such checks it remains possible that residual alignment artifacts or sensor-placement effects are absorbed into the retained components and subsequently into the PLS mode.

    Authors: We acknowledge that quantitative checks on the limb-length normalization were not originally provided. The revised methods section now includes a sensitivity analysis in which normalization scaling factors are varied by ±10% around the values derived from anthropometric data; the resulting changes in the first three shape-based principal components are quantified via Procrustes distance and shown to be small. We have also added a brief error-propagation estimate based on repeated sensor-placement trials performed on a subset of participants, confirming that placement variability does not materially alter the retained components or the subsequent PLS mode. revision: yes

standing simulated objections not resolved
  • Absence of an independent held-out cohort prevents out-of-sample validation of the motor-function index.

Circularity Check

0 steps flagged

No significant circularity detected in derivation chain

full rationale

The paper applies shape-based PCA to movement trajectories to extract kinematic patterns after alignment, then applies PLS to the resulting projections and clinical variables (Brooke score, fat infiltration, age) to identify a covariation mode and reports its canonical correlation r=0.78 as the direct output of that procedure. This is the standard result of the PLS algorithm on the given data and does not reduce any claimed prediction or first-principles result to its inputs by construction, nor does it rely on self-citations for load-bearing uniqueness or import ansatzes. The derivation chain remains self-contained as a sequence of standard statistical methods applied to the observed sample.

Axiom & Free-Parameter Ledger

2 free parameters · 2 axioms · 0 invented entities

The central claim rests on the untested premise that the shape alignment step fully removes confounding effects of growth and speed, plus the modeling choice to treat the PLS-derived mode as a generalizable index rather than a sample-specific fit. No free parameters are explicitly named, but the number of retained principal components and the PLS regularization implicitly function as such.

free parameters (2)
  • number of retained shape-based principal components
    The abstract does not state how many components were kept before feeding into PLS; this choice directly affects the input to the index.
  • PLS component selection criterion
    The number of PLS components or the cross-validation rule used to define the single covariation mode is not reported.
axioms (2)
  • domain assumption Shape-based PCA alignment successfully normalizes for inter-child differences in limb length and movement speed without distorting clinically relevant kinematic features.
    Invoked when the authors state that the method handles 'confounding factors such as limb length variations in growing children and variability in movement speed'.
  • domain assumption The extracted asymmetry and speed patterns are stable markers of motor impairment rather than session-specific or sensor-placement artifacts.
    Required for the claim that patients with SMA showed greater activation of the asymmetry pattern.

pith-pipeline@v0.9.0 · 5790 in / 1778 out tokens · 26370 ms · 2026-05-23T06:23:09.803717+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

40 extracted references · 40 canonical work pages

  1. [1]

    ISS-N1 makes the first FDA-approved drug for spinal muscular atrophy

    Ottesen EW. ISS-N1 makes the first FDA-approved drug for spinal muscular atrophy. Translational neuroscience. 2017;8(1):1–6

    work page 2017

  2. [2]

    Therapeutic strategies for spinal muscular atrophy: SMN and beyond

    Bowerman M, Becker CG, nez Mu˜ noz Y, J R, Ning K, Wood MJ, et al. Therapeutic strategies for spinal muscular atrophy: SMN and beyond. Disease models & mechanisms. 2017;10(8):943–954

    work page 2017

  3. [3]

    Gene and stem cell therapies

    Kaji EH, Leiden JM. Gene and stem cell therapies. Jama. 2001;285(5):545–550

    work page 2001

  4. [4]

    Clinical trial in duchenne dystrophy

    Brooke MH, Griggs RC, Mendell JR, Fenichel GM, Shumate JB, Pellegrino RJ. Clinical trial in duchenne dystrophy. I. The design of the protocol. Muscle & Nerve. 1981;4(3):186–197. doi:https://doi.org/10.1002/mus.880040304

    work page doi:10.1002/mus.880040304 1981

  5. [5]

    Validation of the children’s Hospital of Philadelphia infant test of neuromuscular disorders (CHOP INTEND)

    Glanzman AM, McDermott MP, Montes J, Martens WB, Flickinger J, Riley S, et al. Validation of the children’s Hospital of Philadelphia infant test of neuromuscular disorders (CHOP INTEND). Pediatric Physical Therapy. 2011;23(4):322–326

    work page 2011

  6. [6]

    Wearable full-body motion tracking of activities of daily living predicts disease trajectory in Duchenne muscular dystrophy

    Ricotti V, Kadirvelu B, Selby V, Festenstein R, Mercuri E, Voit T, et al. Wearable full-body motion tracking of activities of daily living predicts disease trajectory in Duchenne muscular dystrophy. Nature Medicine. 2023;29(1):95–103

    work page 2023

  7. [7]

    At-home wearables and machine learning sensi- tively capture disease progression in amyotrophic lateral sclerosis

    Gupta AS, Patel S, Premasiri A, Vieira F. At-home wearables and machine learning sensi- tively capture disease progression in amyotrophic lateral sclerosis. Nature Communications. 2023;14(1):5080

    work page 2023

  8. [8]

    Upper limb movements as digital biomarkers in people with ALS

    Straczkiewicz M, Karas M, Johnson SA, Burke KM, Scheier Z, Royse TB, et al. Upper limb movements as digital biomarkers in people with ALS. Ebiomedicine. 2024;101

    work page 2024

  9. [9]

    Detection and Assessment of Point-to-Point Movements During Func- tional Activities Using Deep Learning and Kinematic Analyses of the Stroke-Affected Wrist

    Oubre B, Lee SI. Detection and Assessment of Point-to-Point Movements During Func- tional Activities Using Deep Learning and Kinematic Analyses of the Stroke-Affected Wrist. IEEE Journal of Biomedical and Health Informatics. 2024;28(2):1022–1030. doi:10.1109/JBHI.2023.3337156

    work page doi:10.1109/jbhi.2023.3337156 2024

  10. [10]

    How wearable sensors can support Parkinson’s disease diagnosis and treatment: a systematic review

    Rovini E, Maremmani C, Cavallo F. How wearable sensors can support Parkinson’s disease diagnosis and treatment: a systematic review. Frontiers in neuroscience. 2017;11:555

    work page 2017

  11. [11]

    Reliability and validity of the Roche PD Mobile Application for remote monitoring of early Parkinson’s disease

    Lipsmeier F, Taylor KI, Postuma RB, Volkova-Volkmar E, Kilchenmann T, Mollenhauer B, et al. Reliability and validity of the Roche PD Mobile Application for remote monitoring of early Parkinson’s disease. Scientific reports. 2022;12(1):12081

    work page 2022

  12. [12]

    Virtual exam for Parkinson’s disease enables frequent and reliable remote measurements of motor function

    Burq M, Rainaldi E, Ho KC, Chen C, Bloem BR, Evers LJ, et al. Virtual exam for Parkinson’s disease enables frequent and reliable remote measurements of motor function. NPJ digital medicine. 2022;5(1):65. 13/16

    work page 2022

  13. [13]

    Improved measurement of disease progression in people living with early Parkinson’s disease using digital health technologies

    Czech MD, Badley D, Yang L, Shen J, Crouthamel M, Kangarloo T, et al. Improved measurement of disease progression in people living with early Parkinson’s disease using digital health technologies. Communications Medicine. 2024;4(1):49

    work page 2024

  14. [14]

    Free-living monitoring of Parkinson’s disease: Lessons from the field

    Del Din S, Godfrey A, Mazz` a C, Lord S, Rochester L. Free-living monitoring of Parkinson’s disease: Lessons from the field. Movement Disorders. 2016;31(9):1293–1313

    work page 2016

  15. [15]

    Temporal Alignment Improves Feature Quality: An Experiment on Activity Recognition with Accelerometer Data

    Choi H, Wang Q, Toledo M, Turaga P, Buman M, Srivastava A. Temporal Alignment Improves Feature Quality: An Experiment on Activity Recognition with Accelerometer Data. In: 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)

    work page 2018

  16. [16]

    Shape Analysis for Pediatric Upper Body Motor Function Assessment

    Kumar S, Gutierezz R, Datta D, Tolman S, Mccrady A, Blemker S, et al. Shape Analysis for Pediatric Upper Body Motor Function Assessment. In: Proceedings of the 2022 ACM International Symposium on Wearable Computers. ISWC ’22. New York, NY, USA: Association for Computing Machinery; 2022. p. 39–43

    work page 2022

  17. [17]

    A survey on deep learning in medicine: Why, how and when? Information Fusion

    Piccialli F, Di Somma V, Giampaolo F, Cuomo S, Fortino G. A survey on deep learning in medicine: Why, how and when? Information Fusion. 2021;66:111–137

    work page 2021

  18. [18]

    Detecting parkinsonian tremor from IMU data collected in-the-wild using deep multiple- instance learning

    Papadopoulos A, Kyritsis K, Klingelhoefer L, Bostanjopoulou S, Chaudhuri KR, Delopoulos A. Detecting parkinsonian tremor from IMU data collected in-the-wild using deep multiple- instance learning. IEEE Journal of Biomedical and Health Informatics. 2019;24(9):2559–2569

    work page 2019

  19. [19]

    Stroke walking and balance characteristics via principal component analysis

    Cho J, Ha S, Lee J, Kim M, Kim H. Stroke walking and balance characteristics via principal component analysis. Scientific Reports. 2024;14(1):10465

    work page 2024

  20. [20]

    Multimodal assessment of Parkinson’s disease: a deep learning approach

    V´ asquez-Correa JC, Arias-Vergara T, Orozco-Arroyave JR, Eskofier B, Klucken J, N¨ oth E. Multimodal assessment of Parkinson’s disease: a deep learning approach. IEEE journal of biomedical and health informatics. 2018;23(4):1618–1630

    work page 2018

  21. [21]

    Detection and classification of stroke gaits by deep neural networks employing inertial measurement units

    Wang FC, Chen SF, Lin CH, Shih CJ, Lin AC, Yuan W, et al. Detection and classification of stroke gaits by deep neural networks employing inertial measurement units. Sensors. 2021;21(5):1864

    work page 2021

  22. [22]

    Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead

    Rudin C. Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead. Nature machine intelligence. 2019;1(5):206–215

    work page 2019

  23. [23]

    A survey on the interpretability of deep learning in medical diagnosis

    Teng Q, Liu Z, Song Y, Han K, Lu Y. A survey on the interpretability of deep learning in medical diagnosis. Multimedia Systems. 2022;28(6):2335–2355

    work page 2022

  24. [24]

    Functional and shape data analysis

    Srivastava A, Klassen EP. Functional and shape data analysis. vol. 1. Springer; 2016

    work page 2016

  25. [25]

    Shape-based functional data analysis

    Wu Y, Huang C, Srivastava A. Shape-based functional data analysis. Test. 2024;33(1):1–47

    work page 2024

  26. [26]

    Shape analysis of elastic curves in euclidean spaces

    Srivastava A, Klassen E, Joshi SH, Jermyn IH. Shape analysis of elastic curves in euclidean spaces. IEEE transactions on pattern analysis and machine intelligence. 2010;33(7):1415–1428

    work page 2010

  27. [27]

    Generative models for functional data using phase and amplitude separation

    Tucker JD, Wu W, Srivastava A. Generative models for functional data using phase and amplitude separation. Computational Statistics & Data Analysis. 2013;61:50–66

    work page 2013

  28. [28]

    Cross-sectional area and muscular strength: a brief review

    Jones EJ, Bishop PA, Woods AK, Green JM. Cross-sectional area and muscular strength: a brief review. Sports medicine. 2008;38:987–994

    work page 2008

  29. [29]

    Quantitative muscle ultrasound measures rapid declines over time in children with SMA type 1

    Ng KW, Connolly AM, Zaidman CM. Quantitative muscle ultrasound measures rapid declines over time in children with SMA type 1. Journal of the Neurological Sciences. 2015;358(1):178– 182. 14/16

    work page 2015

  30. [30]

    Quantitative muscle ultrasound is a promising longitudinal follow-up tool in Duchenne muscular dystrophy

    Jansen M, van Alfen N, van der Sanden MWN, van Dijk JP, Pillen S, de Groot IJ. Quantitative muscle ultrasound is a promising longitudinal follow-up tool in Duchenne muscular dystrophy. Neuromuscular disorders. 2012;22(4):306–317

    work page 2012

  31. [31]

    Upper Extremity Examination for Neuromuscular Diseases (U-EXTEND): Protocol for a Multimodal Feasibility Study

    Gutierrez R, McCrady A, Masterson C, Tolman S, Boukhechba M, Barnes L, et al. Upper Extremity Examination for Neuromuscular Diseases (U-EXTEND): Protocol for a Multimodal Feasibility Study. JMIR Res Protoc. 2022;11(10):e40856. doi:10.2196/40856

    work page doi:10.2196/40856 2022

  32. [32]

    MMR+ – METAMOTIONR+; 2024

    MBIENTLAB. MMR+ – METAMOTIONR+; 2024. https://mbientlab.com/store/ mmrp-metamotionrp/

    work page 2024

  33. [33]

    Dynamic programming and optimal control

    Bertsekas DP, et al. Dynamic programming and optimal control. Belmont, MA: Athena Scientific. 2011;1

    work page 2011

  34. [34]

    PLS-regression: a basic tool of chemometrics

    Wold S, Sj¨ ostr¨ om M, Eriksson L. PLS-regression: a basic tool of chemometrics. Chemometrics and intelligent laboratory systems. 2001;58(2):109–130

    work page 2001

  35. [35]

    Tests of statistical hypotheses concerning several parameters when the number of observations is large

    Wald A. Tests of statistical hypotheses concerning several parameters when the number of observations is large. Transactions of the American Mathematical society. 1943;54(3):426–482

    work page 1943

  36. [36]

    On the adaptive control of the false discovery rate in multiple testing with independent statistics

    Benjamini Y, Hochberg Y. On the adaptive control of the false discovery rate in multiple testing with independent statistics. Journal of educational and Behavioral Statistics. 2000;25(1):60–83

    work page 2000

  37. [37]

    Scikit-learn: Machine Learning in Python

    Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, et al. Scikit-learn: Machine Learning in Python. Journal of Machine Learning Research. 2011;12:2825–2830

    work page 2011

  38. [38]

    Statsmodels: Econometric and Statistical Modeling with Python

    Skipper Seabold, Josef Perktold. Statsmodels: Econometric and Statistical Modeling with Python. In: St´ efan van der Walt, Jarrod Millman, editors. Proceedings of the 9th Python in Science Conference; 2010. p. 92 – 96

    work page 2010

  39. [39]

    A survey of functional principal component analysis

    Shang HL. A survey of functional principal component analysis. AStA Advances in Statistical Analysis. 2014;98:121–142

    work page 2014

  40. [40]

    Algorithms for non-negative matrix factorization

    Lee D, Seung HS. Algorithms for non-negative matrix factorization. Advances in neural information processing systems. 2000;13. Data Availability A smaller version of the data to replicate the results is available at https://github.com/Arafat245/U- Extend Codes. The full dataset generated and/or analyzed during the current study are available from the corr...

    work page 2000