Key Findings
The gait foundation model significantly improved the prediction of health phenotypes across 18 body systems. When compared to a baseline model using height, age, body mass index (BMI), and visceral adipose tissue, the gait embeddings improved predictions in all 18 systems for males and 17 of 18 for females. Additionally, the model enhanced the prediction of medication use and clinical diagnoses. Out of 3,210 tested phenotypes, the embeddings significantly predicted 1,887.
Technical Details
The model was developed using 3D skeletal motion data from 3,328 deeply phenotyped adults, recorded via a single-depth camera during five motor tasks. Unlike conventional gait analysis that relies on engineered features—such as stride length and cadence—this model translates movement into learned embeddings. These embeddings are numerical sets describing the synchronized movement of all body joints over time.
The learned embeddings demonstrated high predictive accuracy for several metrics:
- Body mass index: Pearsonʼs r = 0.91
- Visceral adipose tissue area: Pearsonʼs r = 0.83
- Age: Pearsonʼs r = 0.69
Anatomical ablation analyses further revealed that torso dynamics encode lifestyle and sleep phenotypes, while leg dynamics dominate predictions for frailty and metabolic health.
Background & Context
Gait has traditionally been viewed as a symptom of specific pathologies. However, this research treats it as a systemic biomarker. The study moves beyond simple measurements to capture the full pattern of human movement, aiming to predict diverse health indicators such as sleep quality and grip strength.
Significance & Outlook
These findings establish gait as an independent, multisystem biosignal. By demonstrating that passive 3D motion capture can provide deep insights into systemic health, the research motivates the integration of gait analysis into routine clinical assessments as a scalable vital sign.
Source: https://www.nature.com/articles/s41591-026-04724-9
Get our weekly technology intelligence — free
Receive an infographic that lets you judge at a glance whether each field’s analysis report is worth reading.
Subscribe Free — Weekly Tech Intelligence
By subscribing, you’ll receive Troy-Technical’s weekly technology intelligence newsletter.
- Your email and selected fields are used only to deliver the newsletter.
- We never share your information with third parties.
- You can unsubscribe anytime via the link in each email.
See our Privacy Policy for details.
Takes about a minute · Unsubscribe anytime
