CoChat

Development and Validation of a Machine Learning Wrist-worn Step Detection Algorithm with Deployment in the UK Biobank

Open access

The DOI matches its Crossref record; 34 of the 35 references it cites were found in a registry or an index, and 1 are not checked yet; 5 of the 20 sources its claims cite were checked for support.

Abstract

Background: Step count is an intuitive measure of physical activity frequently quantified in a range of health-related studies; however, accurate quantification of step count can be difficult in the free-living environment, with step counting error routinely above 20% in both consumer and research-grade wrist-worn devices. This study aims to describe the development and validation of step count derived from a wrist-worn accelerometer and to assess its association with cardiovascular and all-cause mortality in a large prospective cohort study. Methods: We developed and externally validated a hybrid step detection model that involves self-supervised machine learning, trained on a new ground truth annotated, free-living step count dataset (OxWalk, n=39, aged 19-81) and tested against other open-source step counting algorithms. This model was applied to ascertain daily step counts from raw wrist-worn accelerometer data of 75,493 UK Biobank participants without a prior history of cardiovascular disease (CVD) or cancer. Cox regression was used to obtain hazard ratios and 95% confidence intervals for the association of daily step count with fatal CVD and all-cause mortality after adjustment for potential confounders. Findings: The novel step algorithm demonstrated a mean absolute percent error of 12.5% in free-living validation, detecting 98.7% of true steps and substantially outperforming other recent wrist-worn, open-source algorithms. Our data are indicative of an inverse dose-response association, where, for example, taking 6,596 to 8,474 steps per day was associated with a 39% [24-52%] and 27% [16-36%] lower risk of fatal CVD and all-cause mortality, respectively, compared to those taking fewer steps each day. Interpretation: An accurate measure of step count was ascertained using a machine learning pipeline that demonstrates state-of-the-art accuracy in internal and external validation. The expected associations with CVD and all-cause mortality indicate excellent face validity. This algorithm can be used widely for other studies that have utilised wrist-worn accelerometers and an open-source pipeline is provided to facilitate implementation. Funding Acknowledgements: This research has been conducted using the UK Biobank Resource under Application Number 59070. This research was funded in whole or in part by the Wellcome Trust [223100/Z/21/Z]. For the purpose of open access, the author has applied a CC-BY public copyright licence to any author accepted manuscript version arising from this submission. AD and SS are supported by the Wellcome Trust. AD and DM are supported by Swiss Re, while AS is an employee of Swiss Re. AD, SC, RW, SS, and SK are supported by HDR UK, an initiative funded by UK Research and Innovation, Department of Health and Social Care (England) and the devolved administrations. AD, DB, GM, and SC are supported by NovoNordisk. AD is supported by the BHF Centre of Research Excellence (grant number RE/18/3/34214). SS is supported by the University of Oxford Clarendon Fund. DB is further supported by the Medical Research Council (MRC) Population Health Research Unit. DC holds a personal academic fellowship from EPSRC. AA, AC and DC are supported by GlaxoSmithKline. SK is supported by Amgen and UCB BioPharma outside of the scope of this work. Computational aspects of this research were funded from the National Institute for Health Research (NIHR) Oxford Biomedical Research Centre (BRC) with additional support from Health Data Research (HDR) UK and the Wellcome Trust Core Award [grant number 203141/Z/16/Z]. The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health.

Verified 2026-11-19 · 1 reference not checked yet

Overview

CoChat Summary

TL;DR

The authors developed and validated a hybrid self-supervised machine-learning and peak-detection wrist-worn step-count algorithm, reporting 12.5% MAPE in free-living OxWalk validation and better accuracy than open-source comparators. Applied to 75,493 UK Biobank participants, higher daily step counts were associated with lower fatal CVD and all-cause mortality.

Key findings

  • In free-living OxWalk validation (n=39), the novel algorithm achieved 12.5% MAPE, 1.3% underestimation, and r=0.98 against ground truth, detecting 98.7% of true steps; comparators performed worse: Ducharme 231.3% MAPE with 69.1% overestimation, and Verisense 63.5% MAPE with 7.2% underestimation and r=0.85.
  • In external Clemson validation (n=30), the novel algorithm had 16.5% MAPE and 16.6% underestimation overall, including 9.2% MAPE during regular walking; Verisense had 17.3% MAPE and 17.6% underestimation overall, while Ducharme had 47.5% MAPE and 46.9% overestimation.
  • In UK Biobank, among 75,493 participants without prevalent CVD or cancer, median follow-up was 6.9 years with 572 CVD mortality events and 1,844 all-cause mortality events. Taking 6,596–8,474 steps/day was associated with 39% [24–52%] lower fatal CVD and 27% [16–36%] lower all-cause mortality versus fewer steps; results also report 8,474–10,284 steps/day associated with 56% [43–66%] lower CVD mortality versus <6,596 steps/day, and 39% [30–47%] and 43% [34–51%] lower all-cause mortality in the middle and most active fifths.

Methods

  • OxWalk free-living ground truth: 39 adults aged 19–81 wore four triaxial accelerometers (wrist and hip) with video annotation; steps were doubly annotated. The model used a self-supervised ResNet-V2 pretrained on UK Biobank accelerometer data, trained on 10-second walk/non-walk epochs, smoothed with a Hidden Markov Model, and counted steps via peak detection on filtered acceleration; 10-fold cross-validation was used.
  • External validation used the Clemson dataset (n=30, mean 37 minutes across regular, semiregular, and irregular walking sessions) and compared the novel algorithm with the Ducharme threshold and Verisense algorithms using MAPE, bias, Spearman correlation, and Bland-Altman plots.
  • UK Biobank deployment processed raw wrist accelerometer data from 103,391 available participants, excluded insufficient wear or poor data, and analyzed 75,493 participants without prevalent CVD or cancer. Daily step count was the median across the 7-day measurement period; Cox proportional hazards regression with age as the timescale estimated associations with fatal CVD and all-cause mortality after confounder adjustment.

Limitations

  • The paper states that training data consisted of short 1-hour free-living windows, so some overestimation may occur in multiday protocols due to lack of extended sedentary inactivity, although class rebalancing was used; future work should assess robustness across populations and against 24-hour free-living ground truth.
  • Evident from the text: OxWalk validation was small (n=39) and short (mean wear 58 minutes), and Clemson external validation used scripted/laboratory activities (n=30, mean 37 minutes) rather than free-living conditions.
  • Evident from the text: the UK Biobank cohort was aged 40–69 at recruitment, so mortality associations may not generalize to younger or older populations.

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

Integrity signals

Registry

Crossref · 10.1101/2023.02.20.23285750

Cites

34 of 35 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Figures

Figure 1: Schematic of the process for generating step count from 30 seconds of raw triaxial accelerometer data using a hybrid self-supervised learning (SSL) and peak detection step counting model.
Figure 1: Schematic of the process for generating step count from 30 seconds of raw triaxial accelerometer data using a hybrid self-supervised learning (SSL) and peak detection step counting model.CC BY · PubMed Central
Figure 2: Bland-Altman plots with dotted 95% limits of agreement for the comparison of step counting models in the (Top) OxWalk free-living dataset of 39 adults and (Bottom) Clemson laboratory-based dataset of 30 young adults. Left: baseline acceleration threshold model8, Centre: Verisense algorithm25, and Right: the novel hybrid self-supervised learning model.
Figure 2: Bland-Altman plots with dotted 95% limits of agreement for the comparison of step counting models in the (Top) OxWalk free-living dataset of 39 adults and (Bottom) Clemson laboratory-based dataset of 30 young adults. Left: baseline acceleration threshold model8, Centre: Verisense algorithm25, and Right: the novel hybrid self-supervised learning model.CC BY · PubMed Central
Figure 3: Adjusted estimated marginal mean (95% confidence interval) daily step count according to self-reported overall health status, hospital data derived chronic disease status, and select diagnoses for 75,493 UK Biobank participants. Mean daily step counts are adjusted for age and sex.
Figure 3: Adjusted estimated marginal mean (95% confidence interval) daily step count according to self-reported overall health status, hospital data derived chronic disease status, and select diagnoses for 75,493 UK Biobank participants. Mean daily step counts are adjusted for age and sex.CC BY · PubMed Central
Figure 4: (Top) Forest plots for all-cause mortality and cardiovascular disease mortality associations with quintiles of daily step count, (Bottom) continuous daily step count for 75,493 UK Biobank participants. Hazard ratios (HR) and 95% confidence intervals were calculated using age as a timescale, adjusted for sex, ethnicity, education, alcohol intake, smoking status, Townsend deprivation index, processed meat intake, fresh fruit intake, oily fish intake, and added salt intake. HR is above and number of events is plotted below each data point. Spline plot of hazard ratio and 95% confidence interval of the association of continuously modelled median daily step count. Vertical bars along the step axis indicate distribution of participant daily step counts.
Figure 4: (Top) Forest plots for all-cause mortality and cardiovascular disease mortality associations with quintiles of daily step count, (Bottom) continuous daily step count for 75,493 UK Biobank participants. Hazard ratios (HR) and 95% confidence intervals were calculated using age as a timescale, adjusted for sex, ethnicity, education, alcohol intake, smoking status, Townsend deprivation index, processed meat intake, fresh fruit intake, oily fish intake, and added salt intake. HR is above and number of events is plotted below each data point. Spline plot of hazard ratio and 95% confidence interval of the association of continuously modelled median daily step count. Vertical bars along the step axis indicate distribution of participant daily step counts.CC BY · PubMed Central

Verification

Claims the paper makes

4 Valid · 0 Not valid · 10 Not checked · 39 Own results and interpretations

23 more claims

Create a free account to see all 53 claims and to verify the 15 references still open.

See all claims — free account
  • It may state two separate claims as one. 17
  • Probably not a substantial claim. 10

Create a free account to see these claims.

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.

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

1 of 1

Citations

Cites

Show all 35 references

Research works printed but not linked: 2

  1. 10.1001/JAMAINTERNMED.2022.4000 · DOI · Paper page
  2. 2016, Regression Models and Life-Tables, J R Stat Soc Ser B, 15, 1

Cited by

See all 46

Metadata

Authors
Small, S. R., Chan, S., Walmsley, R., Fritsch, L. V., Acquah, A., Mertes, G., Feakins, B. G., Creagh, A., Strange, A., Matthews, C. E., Clifton, D. A., Price, A. J., Khalid, S., Bennett, D., & Doherty, A.
Affiliations
  • University of Oxford (GB)
  • Division of Cancer Epidemiology and Genetics (US)
Corresponding author
Doherty, Aiden
Venue
medRxiv
Publisher
openRxiv
Published
2023-02-22
OpenAlex
W4321480397
Cited by
22
References
34
Citations by year
2026: 3 · 2025: 7 · 2024: 7 · 2023: 5
Type
posted content
Language
en
Open access
Licence
cc-by
Locations
  • medRxiv (acceptedVersion, cc-by)
  • PubMed (publishedVersion)
  • PubMed Central (submittedVersion, cc-by)
Funders
  • Wellcome Trust (203141/Z/16/Z, 223100/Z/21/Z)
  • UK Research and Innovation
  • British Heart Foundation (RE/18/3/34214)
  • Department of Health and Social Care
  • University of Oxford
  • Swiss Re
  • Clarendon Fund
Topics
  • Health Sciences › Medicine › Public Health, Environmental and Occupational Health › Physical Activity and Health
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Context-Aware Activity Recognition Systems
  • Physical Sciences › Engineering › Biomedical Engineering › Non-Invasive Vital Sign Monitoring
Keywords
wrist-worn accelerometers, self-supervised learning, physical activity measurement, cardiovascular mortality, all-cause mortality, dose-response relationship, UK Biobank, wearable device validation, open-source software pipeline, machine learning deployment

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.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-05, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-05 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (qwen3.8-omni-flash).

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: Development and Validation of a Machine Learning Wrist-worn Step Detection Algorithm with Deployment in the UK Biobank. https://app.cochat.ai/papers/10.1101/2023.02.20.23285750
The paper
Small, S. R., Chan, S., Walmsley, R., Fritsch, L. V., Acquah, A., Mertes, G., Feakins, B. G., Creagh, A., Strange, A., Matthews, C. E., Clifton, D. A., Price, A. J., Khalid, S., Bennett, D., & Doherty, A. (2023). Development and Validation of a Machine Learning Wrist-worn Step Detection Algorithm with Deployment in the UK Biobank. medRxiv. https://doi.org/10.1101/2023.02.20.23285750

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.