EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Identifying characteristics and behavioral phenotypes of musculoskeletal disorders using wearables in All of Us

April Y Yan iD, Traci J Speed iD, Casey O Taylor iD

JAMIA Open · 06.11.2026 · 9 (6)

DOI: 10.1093/jamiaopen/ooag221 ↗

Kurz gefasst

Fachartikel in der Zeitschrift JAMIA Open, online erschienen am 17.11.2026. Verfasst von April Y Yan und 2 weiteren Autor:innen. Beteiligte Einrichtung: Department of Biomedical Engineering, Johns Hopkins School of Medicine, Baltimore, MD 21287, United States. Die Arbeit nennt 32 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Human Genome Research Institute.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Objective

Wearable devices provide continuous measures of daily activity, sleep, and heart-rate (HR)-derived characteristics, creating new opportunities for digital phenotyping of musculoskeletal disorders (MSD). However, longitudinal behavioral patterns and heterogeneity among individuals with MSD remain poorly understood. This study aimed to identify characteristics associated with MSD and characterize behavioral phenotypes using longitudinal wearable data. Materials and Methods We analyzed 2344 participants (MSD = 1036; non-MSD = 1308) from the All of Us Research Program. First, supervised machine-learning models were trained using sociodemographic, physical measurement, and wearable-derived features to distinguish individuals with and without MSD and identify factors associated with MSD status. Model interpretation was performed using SHapley Additive exPlanations values. Second, temporal profiling and unsupervised clustering were applied to 90-day wearable trajectories of participants with MSD to identify behavioral phenotypes.

Results

XGBoost achieved the best classification performance (accuracy = 0.929, area under the receiver operating characteristic curve = 0.930, area under the precision-recall curve = 0.870, F1 = 0.919). Important factors associated with MSD included lower activity, longer sleep, and female sex. Behavioral phenotyping identified 5 distinct phenotypes differing in activity patterns, sleep characteristics, HR-derived measures, pain burden, mental health conditions, and employment status. Notably, high-pain phenotypes exhibited greater prevalence of anxiety and depressive disorders, lower pulse rate variability, and distinct sleep profiles compared with more active phenotypes. These phenotypes demonstrated moderate reproducibility in bootstrap-based stability analysis (mean adjusted rand index = 0.62).

Discussion

Longitudinal wearable data revealed behavioral heterogeneity among individuals with MSD that was not fully explained by traditional diagnostic categories. These findings demonstrate the value of wearable-based digital phenotyping for characterizing differences in daily functioning and symptom burden among individuals with MSD.

Conclusion

Wearable-derived behavioral phenotyping provides a complementary approach for understanding heterogeneity in MSD and may support future efforts in patient monitoring, stratification, and personalized management.

Schlagwörter

Stichwörter der Autor:innen

cluster analysisdigital healthmachine learningmusculoskeletal disorderswearables

Autor:innen und Institutionen

  • Department of Biomedical Engineering, Johns Hopkins School of Medicine, Baltimore, MD 21287, United States.
  • Institute for Computational Medicine, Johns Hopkins Whiting School of Engineering, Baltimore, MD 21218, United States.
  • Department of Psychiatry and Behavioral Sciences, Johns Hopkins University School of Medicine, Baltimore, MD 21287, United States.
  • Division of General Internal Medicine, Department of Medicine, Johns Hopkins University School of Medicine, Baltimore, MD 21287, United States.

ORCID-Profile: April Y Yan, Traci J Speed, Casey O Taylor

Förderung

  • National Human Genome Research Institute

Zitieren

APA

Yan, A. Y., Speed, T. J., & Taylor, C. O. (2026). Identifying characteristics and behavioral phenotypes of musculoskeletal disorders using wearables in All of Us. JAMIA Open, 9(6). https://doi.org/10.1093/jamiaopen/ooag221

Vancouver

Yan AY, Speed TJ, Taylor CO. Identifying characteristics and behavioral phenotypes of musculoskeletal disorders using wearables in All of Us. JAMIA Open. 2026;9(6). doi:10.1093/jamiaopen/ooag221

BibTeX
@article{Yan2026,
  author = {Yan, April Y and Speed, Traci J and Taylor, Casey O},
  title = {Identifying characteristics and behavioral phenotypes of musculoskeletal disorders using wearables in All of Us},
  journal = {JAMIA Open},
  year = {2026},
  volume = {9},
  number = {6},
  doi = {10.1093/jamiaopen/ooag221}
}
Kostenlos starten →