A person walks, stands, and turns while cameras quietly record the body in motion, with no sensors strapped on and no clinic required. A new study in Nature Medicine suggests that these ordinary movements carry traces of the liver, the metabolism, sleep, and mood, legible to a machine that has learned to read them. The work was led by researchers in a large Israeli-led cohort and co-authored by Newton Tech, a company that sells the tools that made the measurements. It is a striking hint that the way we move may hold more of our inner state than we have known to ask about, and it arrives with a
Movement Patterns Carry Health Signals Across 18 Body Systems, New Nature Medicine Study Reports
Liver health leaves a detectable imprint on how people move.
Mark
Start with the headline number. 1,887 of 3,210 measures. Is that as big as it sounds?
Mimi
It's a wide net. The model touched 18 body systems, and the associations held up after adjusting for age, BMI, visceral fat and height.
Luke
But an association isn't a diagnosis. With 3,210 measures tested, a lot of hits is partly the nature of testing many things. The summary doesn't tell us how much each one actually predicts.
Mark
The liver result seems the most surprising.
Mimi
Yes. Movement matched or beat a full panel of standard liver blood tests at improving prediction of liver elasticity, beyond age and BMI.
Luke
Improving prediction of elasticity, though, not detecting fibrosis in patients. That's a narrower claim.
Mark
And the depression and diabetes figures, 81% and 83% more cases identified?
Mimi
Those come from a screening analysis in men, and the numbers are striking.
Luke
They're relative increases, and we aren't given false-positive rates or absolute counts. Also, antidepressants affect movement, so some depression signal may be medication.
Mark
Does it matter that Newton Tech co-wrote it?
Luke
It matters. The company built the data pipeline and sells the product. That doesn't make the findings wrong, but independent replication is the missing piece.
Mimi
The NervGen result is similar: 20 people, post hoc, exploratory. Promising, not proof.
Il Polso
- An AI model linked 3D movement patterns to 1,887 of 3,210 health measures across 18 body systems in 3,328 adults.
- Movement data improved predictions beyond age, BMI, visceral fat and height, even reaching liver stiffness as well as a full panel of standard liver blood tests.
- Screening analyses reported large jumps in identified cases, with depression detection up 81% and diabetes up 83% in men.
- The study's co-author Newton Tech sells the platform and pitches it to drug developers, a commercial stake that invites scrutiny.
- Independent replication and disease-specific trials in conditions like ALS, Parkinson's and Duchenne are still needed before movement becomes a trusted endpoint.
A person walks, stands, and turns while cameras quietly record the body in motion, with no sensors strapped on and no clinic required. A new study in Nature Medicine suggests that these ordinary movements carry traces of the liver, the metabolism, sleep, and mood, legible to a machine that has learned to read them. The work was led by researchers in a large Israeli-led cohort and co-authored by Newton Tech, a company that sells the tools that made the measurements. It is a striking hint that the way we move may hold more of our inner state than we have known to ask about, and it arrives with a reminder that the one who profits from reading the signal is also among those vouching for it.
A person walks, stands and moves through five standardized motor tasks while cameras record. No markers on the body, no specialized lab. A new paper in Nature Medicine argues that this footage, once converted into three-dimensional skeletal motion and run through an AI model, carries information about the liver, metabolism, sleep and mood.
The study, led by Adam Gabet within the Human Phenotype Project, analyzed 17,589 movement sequences from 3,328 adults. The model found significant associations with 1,887 of 3,210 health measures across 18 body systems. Even after accounting for age, body mass index, visceral fat and height, movement still sharpened predictions in nearly every system.
Some results reach well beyond traditional gait analysis. Movement predicted liver stiffness, the main noninvasive measure of fibrosis, as well as a full panel of standard liver blood tests did. It also helped identify reported depression and visceral fat, the fat wrapped around internal organs, with a correlation of 0.83.
Newton Tech, which co-authored the paper and captured the data with its Movement Lab technology, has an obvious commercial interest. Its MovementOS platform is designed for drug developers, and its CEO argues that movement holds far more biological information than standard tests record. An outside researcher at Spaulding Rehabilitation Hospital called the work an important step toward movement as a multisystem biomarker.
The company also points to a separate exploratory trial analysis by NervGen Pharma in spinal cord injury, where a small group of treated patients showed strong responses. The source itself cautions that disease-specific validation is still required. Independent replication and trials in specific conditions are what will determine whether the promise holds.
A person walks, stands and moves through five standardized motor tasks while cameras record. No markers on the body, no specialized lab. A new paper in Nature Medicine argues that this footage, once turned into three-dimensional skeletal motion and run through an AI model, carries information about the liver, the metabolism, sleep and mood.
The study is titled "A Gait Foundation Model Predicts Multi-System Health Phenotypes from 3D Skeletal Motion." Its lead author is Adam Gabet, and it was conducted through the Human Phenotype Project, a large research cohort. Newton Tech, which develops AI-powered movement analysis for clinical research, co-authored the paper. Its Movement Lab technology captured and processed the data, and members of its team are among the authors.
The numbers are broad. In 3,328 adults, researchers analyzed 17,589 movement sequences. The model found significant associations with 1,887 of 3,210 health measures, spanning 18 body systems. Those measures ranged from blood tests and liver imaging to sleep, mental health and clinical diagnoses. After accounting for age, body mass index, visceral fat and height, movement data still improved predictions in all 18 systems in men and in 17 of 18 in women.
Several findings reached beyond what gait analysis is traditionally used for. On the liver, movement predicted liver elasticity, the main noninvasive measure of fibrosis. Adding movement to age and BMI improved that prediction as much as, or more than, adding a full panel of standard liver blood tests, in both sexes. On mental health, movement improved prediction of reported depression, and in a screening analysis it increased the number of depression cases identified by 81% in men. The models also predicted antidepressant use. On metabolism, movement predicted visceral adipose tissue, the fat around internal organs, with a correlation of 0.83, well ahead of conventional gait measures. In the same screening analysis, it raised the number of diabetes cases identified by 83% in men. The authors also report that movement data increased cases identified for peptic ulcer disease, and that the models predicted use of proton pump inhibitors and antihistamines.
The models are explainable, the source says, in that they show which body regions carry which signals. Lower-body movement was especially informative for metabolic health and frailty, while torso movement contributed to predictions about sleep and lifestyle.
Newton Tech's commercial interest is plain. The company sells a platform called MovementOS, which converts video of standard assessments into hundreds of biomechanical markers per patient, and it pitches the technology to drug developers who now rely on single numbers such as a walking time. CEO Yuval Brodsky, a co-author, said movement holds far more biological information than conventional tests record. Paolo Bonato, director of the Motion Analysis Laboratory at Spaulding Rehabilitation Hospital and an associate professor at Harvard Medical School, called the work an important step toward establishing movement as a multisystem biomarker. Spaulding is bringing Newton's Movement Lab into its lab under a multiyear agreement, a lab that serves as the primary clinical gait laboratory for Boston Children's Hospital's cerebral palsy team.
The company also points to a trial use. In May 2026, NervGen Pharma reported an exploratory, post hoc gait analysis of its Phase Ib/IIa CONNECT SCI study of NVG-291 in chronic spinal cord injury, built on Newton's blinded video analysis. Among 20 participants, a combined statistical test favored the drug over placebo (p=0.0197), and 10 of 10 treated participants were classed as responders versus 1 of 10 on placebo. That work was separate from the Nature Medicine paper.
The source itself notes that disease-specific validation will be required before movement can serve as an endpoint in conditions such as ALS, Parkinson's or Duchenne. Because the developer co-authored the paper and sells the product, independent replication and trials in specific diseases are what to watch next.
Citazioni salienti
This study is an important step toward establishing movement as a multisystem biomarker.— Paolo Bonato, Spaulding Rehabilitation Hospital and Harvard Medical School
Movement contains far more biological information than the handful of numbers conventional tests record.— Yuval Brodsky, CEO of Newton Tech and study co-author
Vuoi la storia completa?
Leggi l'originale su Medical Xpress ↗
Health & Medicine