REHOVOT, Israel — A person’s gait may hold hidden indicators of overall health that extend far beyond their legs and feet, according to a study published in the journal Nature Medicine.
Computer scientists and health researchers at the Weizmann Institute of Science in Israel used an artificial intelligence system to evaluate full-body movements in 3,328 adults. The system detected subtle biomechanical patterns linked to internal health factors, including liver organ stiffness, sleep quality, and metabolic conditions such as diabetes.
Beyond Speed and Stride
Physicians have traditionally examined walking speed, posture, and balance as basic indicators during physical examinations. To evaluate whether technology could identify more nuanced signals, researchers asked study participants to complete five routine physical tasks while monitored by a depth-sensing camera system in a controlled laboratory setting.
The participants walked on a treadmill at a set speed of 3 kilometers per hour, walked at their preferred pace, walked in place, rose from a chair, and stood still for a balance assessment. The specialized camera tracked joint positions across three dimensions. Researchers then trained a deep learning computer model by removing portions of the video data and forcing the system to reconstruct missing body movements. Through this process, the model learned how distinct body regions move in tandem.
When the model converted body mechanics into numerical patterns, researchers cross-referenced them against 3,210 medical traits, laboratory results, and body measurements. The AI identified statistically significant links with 1,887 of those health characteristics.
Predicting Liver Stiffness and Sleep Issues
Among the study’s most notable findings were connections between movement and internal organ health. The system successfully estimated liver stiffness—a key marker doctors use to assess tissue scarring—with predictive value comparable to standard liver blood tests.
The system also uncovered unexpected correlations between torso motion and sleep indicators. Movement of the upper body between the shoulders and hips carried patterns associated with exercise habits, insomnia, and sleep duration from the previous night. The model also identified markers related to sleep apnea, a condition where breathing repeatedly pauses or becomes restricted during sleep.
“A finding that stood out to me was how much sleep quality affects our body,” Adam Gabet, a computer scientist at the Weizmann Institute of Science and study author, told ScienceAlert.
Gabet noted that earlier movement research focused primarily on leg biomechanics, often neglecting the upper body. Integrating full-body movement data into existing models that already tracked age, height, body mass index, and internal body fat improved overall predictive accuracy for conditions such as diabetes, depression, and insomnia.
Study Scope and Limitations
Despite the promising results, researchers emphasized that the technology does not diagnose medical conditions or prove that health problems directly alter how a person walks. Because the observational study collected participant data at a single point in time, it could not determine whether physical changes or health conditions occurred first.
The study population was also relatively uniform, consisting mostly of healthy Ashkenazi Jewish adults in Israel, meaning the findings may not translate directly to diverse global populations. Additionally, the motion capture system utilized equipment and software manufactured by Newton Tech, whose staff co-authored the study.
Before such technology could be integrated into clinical practice or routine health screenings, independent studies in broader populations are needed. Gabet said that clinical trials showing high specificity and clear separation will be required prior to medical deployment. Researchers plan to investigate whether tracking movement changes over time can help catch health conditions before physical symptoms fully emerge.



