November 19, 2026, 8:46 am | Read time: 7 minutes
What does the way we move reveal about our health? Previous studies have shown that a slower walking pace in older adults is linked to an increased risk of dementia. A new study explores what other health indicators might be found in our movements–such as biological age, abdominal fat, or sleep quality. The study’s findings are remarkable.
The way people walk is highly individual. Everyone has their own walking rhythm, speed, and posture. In a new study published in the journal “Nature Medicine,” Israeli researchers from the Weizmann Institute of Science analyzed the movement records of 3,328 adults to discover the connections between their gait and their health.1 They developed a model based on artificial intelligence for this purpose.
What exactly did the researchers want to find out?
The Israeli research team wanted to know if the movement of the entire body contains more health information than traditional gait metrics like step frequency or stride length. Instead of calculating individual metrics, the scientists examined how joints move together over time. They then assessed how well existing health data could be estimated from these movement patterns. This included grip strength, body composition, lab values, and sleep characteristics. The study examined 3,210 features from 18 health areas, including:
- Body Composition: BMI, fat around internal organs
- Bone Density: Bone mineral density
- Physical Frailty: Hand grip strength
- Cardiovascular Function: Heart rate, ECG features
- Liver: Liver stiffness
- Kidney Function: Kidney features
- Blood Count and Hematopoiesis: Hemoglobin, hematocrit
- Blood Lipids: Cholesterol, triglycerides
- Metabolism: Metabolic products in blood serum
- Sleep: Sleep duration, sleep apnea metrics
- Mental Health: Depressive symptoms
- Lifestyle: Walking and movement habits
How was the study structured?
The participant data comes from the “Human Phenotype Project.”2 This is an international, large-scale health database. For the current study, the data was evaluated as a snapshot (cross-sectional analysis). It includes body measurements, lab values, medical information, and lifestyle details. Under controlled conditions, 3,328 adult study participants completed the following tasks:
- Walking on a treadmill at 3 km/h
- Walking on a treadmill again, this time at a preferred pace
- Walking in place
- Standing up from a chair
- Standing still for balance control
All these tasks were recorded by a special depth camera, capturing the spatial position of 26 joint points. The study examined not only walking but also standing up and standing still, and how joint positions changed during these activities. In this cross-sectional analysis, the researchers linked movement patterns with already collected health characteristics. Thus, the study did not investigate whether future diseases could be predicted from the recordings.
How the AI model was developed
Initially, the researchers obscured parts of the 3D motion recordings. The model was tasked with reconstructing what was missing from the visible remainder. This allowed it to learn typical movement sequences without needing information on diseases, lab values, or sleep during this training phase.
Afterward, the trained model converted each movement sequence into a compact numerical profile, known as an “embedding.” This profile describes not only how fast someone walks but also how the movements of multiple joints coordinate over time.
With these numerical profiles, the team trained additional AI models to estimate individual health characteristics. They compared the results with both traditional gait metrics and a baseline model that only considered age, body mass index (BMI), height, and the amount of abdominal fat around internal organs. This allowed them to assess whether the movement data provided additional benefits beyond these known characteristics.
The AI model demonstrated impressive accuracy: Body weight (BMI) could be determined extremely accurately (correlation value r = 0.91), abdominal fat very reliably (r = 0.83), and biological age could also be well inferred from gait (r = 0.69).
How gait and health are connected
The data analysis enabled researchers to establish connections between participants’ gait and health. The AI model identified statistically significant correlations with 1,887 of the 3,210 health features examined.
Some of the correlations involved age and body fat percentage. However, others pointed to health features that initially seem unrelated to walking. The discovery is also drawing international attention: The science magazine “ScienceAlert” highlights that the unexpected insights into liver values and sleep health underscore the potential of AI gait analysis. Researchers found that movement patterns could even provide clues about organic values like liver stiffness. The AI model offers valuable reference points that can aid doctors in early detection.
Another surprise is the connection between gait and sleep. The movement of the torso, the area between the shoulders and hips, provided information linked to sleep and physical activity. Based on the movement patterns, the AI model could predict values associated with sleep apnea. Additionally, the model could assess participants’ insomnia and sleep duration from the previous night based on their movements.
With the movement data, predictions improved for men in all 18 and for women in 17 of the 18 health areas examined. For metabolic features and physical frailty, leg movements were particularly informative for the model. This was evident when the team deliberately omitted data from specific body regions and assessed how much the predictions changed.
Researchers Can Predict Risk for 1,000 Diseases Based on Sleep
Voice May Indicate How Quickly the Body Ages
Conclusion and limitations of the study
Can our health be directly read from our gait in the future? The connections identified are not yet that simple and clear. However, researchers hope that patient movement analysis will become a routine health examination. Gait assessment could be a quick and easy test to provide doctors with clues about potential health-threatening changes. If gait analysis provides indications, more in-depth examinations like a blood test could follow.
It is important to note that the study results do not allow conclusions about whether, for example, sleep problems and other diseases lead to specific movement patterns. Since it was a cross-sectional analysis where participants were examined at a single point in time, researchers cannot determine which changes occurred first. Further studies are needed for this. Another limitation of the study: The participants were healthy, adult individuals from Israel. Therefore, these results cannot be generalized to all population groups and ethnicities.
The depth camera and accompanying software used for movement analysis were provided by the company Newton Tech. The manufacturer was a co-author of the study. Therefore, independent tests and clinical studies are still needed to enable the medical use of gait analysis.
Also interesting: How long you should walk continuously
Another study provides initial insights into the connection between gait and mental health
There is currently a lot of research on “gait and health.” Researchers at the University of Miami investigated whether the way we walk also reveals something about our mental health in the “MindStep” project.3 A total of about 95 students participated in two sub-studies. Over six weeks, they visited the lab four times. There, they answered questions about their mental well-being and solved various cognitive tasks while walking. Cameras recorded their movements. About a quarter of the participants reported anxiety or depressive symptoms.
An initial evaluation from one of the sub-studies suggests that students with self-reported depressive symptoms had more variable stride lengths when solving more challenging memory tasks while walking. The team also observed gait differences in students who tended to suppress their emotions.
Ultimately, researchers hope that changes in gait could be another indicator that someone is experiencing fluctuations in mental health. Today, changes in sleep or appetite behavior are used for anxiety and depression. In the future, movement aspects might also play a role.