Walking Fingerprinting Using Wrist Accelerometry during Activities of Daily Living in NHANES

Lily Koffman

Department of Biostatistics, Johns Hopkins School of Public Health

Before we get started…

Join the Mobile and Wearable Data Science ASA Interest Group! wearableds.github.io/involvement

Introduction: accelerometry data

Introduction: accelerometry data

Problem setup

Problem setup

Problem setup

Big picture method: time series to scalar predictors

Model fitting

“Fingerprints” summarize predictors and are different across individuals

The method works really well

32 individuals, 6 minutes of walking each

100% rank-1 accuracy (Koffman et al. 2023)

The method works really well

32 individuals, 6 minutes of walking each

100% rank-1 accuracy (Koffman et al. 2023)

153 individuals, 3 minutes of walking each

93% rank-1 accuracy (Koffman, Crainiceanu, and Leroux 2024)

The method works really well

32 individuals, 6 minutes of walking each

100% rank-1 accuracy (Koffman et al. 2023)

153 individuals, 3 minutes of walking each

93% rank-1 accuracy (Koffman, Crainiceanu, and Leroux 2024)

But…we know when people are walking

Will the method work in a large, unlabeled data set?

NHANES data

Walking fingerprinting in NHANES

Process

  • Use algorithm to find walking
  • Partition data into train/test
  • Fit models

Walking fingerprinting in NHANES

Process

  • Use algorithm to find walking
  • Partition data into train/test
  • Fit models

Spoiler alert: the models perform worse (1-41% rank-1 accuracy)

Why?

Walking identification


ADaptive Empirical Pattern Transformation (ADEPT) (Karas et al. 2019)

library(adept)
adept::segmentWalking(
  xyz, # data frame of tri-axial accelerometry (3 cols)
  xyz.fs = 100, # sample rate 
  template = templates # list of templates for pattern matching
)



stepcount (Small et al. 2024)

devtools::install_github("jhuwit/stepcount")
library(stepcount)
stepcount::stepcount(file = sample_data, model_type = "ssl")

Walking identification

Walking identification results comparison

More specific algorithm is better

Partition data into train/test

Train/test partition results comparison

Temporal setting is harder

Sample size

Accuracy \(\downarrow\) with increasing sample size

Amount of training data

More data is better

Model choice

It depends… and computational time is not equal

Model improvements

Dataset Model Rank 1 Rank 1%
Random
(n=13,367) 		Logistic 9.7 68
Oversampled at 10% 41 68
Weighted 34 96
Two-stage 20 68
Rank 1 and rank 1% accuracies of different model types on the entire population

Fingerprints

Fingerprints

Summary


  • We can identify individuals from their walking patterns in large, unlabeled datasets using predictors derived from acceleration, lag acceleration
  • Performance depends on sample size, walking identification algorithm, train/test partition, length of training data, model choice
  • Preprint on arXiv (Koffman, Muschelli, and Crainiceanu 2025)
  • Next steps: deep learning, using fingerprint as outcome instead of predictor

Thank you!



References

Karas, Marta, Marcin Stra̧czkiewicz, William Fadel, Jaroslaw Harezlak, Ciprian M Crainiceanu, and Jacek K Urbanek. 2019. “Adaptive Empirical Pattern Transformation (ADEPT) with Application to Walking Stride Segmentation.” Biostatistics 22 (2): 331–47. https://doi.org/10.1093/biostatistics/kxz033.
Koffman, Lily, Ciprian Crainiceanu, and Andrew Leroux. 2024. “Walking Fingerprinting.” Journal of the Royal Statistical Society Series C: Applied Statistics 73 (5): 1221–41. https://doi.org/10.1093/jrsssc/qlae033.
Koffman, Lily, John Muschelli, and Ciprian Crainiceanu. 2025. “Walking Fingerprinting Using Wrist Accelerometry During Activities of Daily Living in NHANES.” https://arxiv.org/abs/2506.17160.
Koffman, Lily, Yan Zhang, Jaroslaw Harezlak, Ciprian Crainiceanu, and Andrew Leroux. 2023. “Fingerprinting Walking Using Wrist-Worn Accelerometers.” Gait & Posture 103 (June): 92–98. https://doi.org/10.1016/j.gaitpost.2023.05.001.
Small, Scott R, Shing Chan, Rosemary Walmsley, Lennart von Fritsch, Aidan Acquah, Gert Mertes, Benjamin G Feakins, et al. 2024. “Self-Supervised Machine Learning to Characterize Step Counts from Wrist-Worn Accelerometers in the UK Biobank.” Medicine and Science in Sports and Exercise 56 (10): 1945.

tinyurl.com/koff-enar25