The Syracuse University researcher whose dissertation taught older adults to fall on purpose explains what a safe landing looks like, why a relaxed body gets hurt less than a braced one, and how AI is starting to measure falls that happen outside the lab.
Key Takeaways
- The research began with a family story. Dr. Moon’s father, who learned Taekwondo as a teenager in 1960s South Korea, tripped while hiking downhill in his 60s, two days before her wedding, rolled through the fall, and walked away uninjured.
- The tuck-and-roll is a sideways-fall technique with four parts: rotate the body, flex the knee and sit down, tuck the chin, then roll. The aim is to land on the buttocks, the most cushioned area of the body. That spreads the energy out instead of driving it into the hip.
- Forward and backward falls call for different adjustments. Dr. Moon’s guidance is to block with the arm but let it absorb energy rather than lock out straight, which is how wrists get fractured. Real-world fall video complicates this, and the article says so.
- Relaxing matters more than technique. A stiffened body falls like a stick; a relaxed one dissipates energy. Dr. Moon’s view is that the main benefit of training is a changed mindset, not a graceful landing.
- Older adults hit their heads during fall trials even when they were fit and athletic. Younger participants did not. Dr. Moon suspects neck muscles weaken with age because almost nobody trains them, but she is clear this has not been tested.
- Her current work uses open-source AI pose estimation to pull hard numbers out of ordinary video of real falls. That sidesteps the ethics problem that limits lab studies to the healthiest older adults.
Disclaimer: This article is general information, not medical advice. Decisions about medical conditions should be made with a doctor or physical therapist. If a fall may have caused an injury, call 911.
Yaejin Moon, PhD, is an assistant professor of exercise science in the David B. Falk College of Sport at Syracuse University, where she directs the Systems Motor Control and Biomechanics (SyMBio) Laboratory. The lab studies how the body controls everyday movement, using biomechanics equipment and neuro-stimulation to examine how bones, muscles, and the nervous system work together to keep a person upright, and how that coordination changes with age or disease. She earned her master’s and PhD in kinesiology in the Department of Kinesiology and Community Health at the University of Illinois Urbana-Champaign, since renamed the Department of Health and Kinesiology. From 2018 to 2021 she was a postdoctoral research fellow at Northwestern University, based at the Shirley Ryan AbilityLab in Chicago, where she studied spinal cord stimulation for gait recovery after stroke and spinal cord injury. As a PhD student she led the safe landing work that became her 2016 dissertation, later published in 2019. She is lead author of both the systematic review that identified the tuck-and-roll as the most effective protective landing strategy and the trial that tested whether older adults could learn it. She is a biomechanist rather than a clinician: her training is in the physics and motor control of human movement, so she studies how bodies move and what happens to them on impact, and she does not diagnose or treat patients. We spoke about where the research came from, what a safe fall actually involves, and where the measurement technology is headed.
This conversation has been edited for length and clarity.
On the fall that started the research
Dr. Jacob Sosnoff is now associate dean for research in the School of Health Professions at the University of Kansas Medical Center, where he is also professor of physical therapy, rehabilitation science, and athletic training. He was a professor at the University of Illinois Urbana-Champaign when he supervised this work. We interviewed him separately. He told us the idea traced back to a family story. Moon’s father had a martial arts background, and had once used it to protect her during a fall when she was an infant. Moon’s own account differs on one point, and she wanted to correct the record.
Dr. Sosnoff mentioned the idea came from something that happened with your father. Could you tell it in your own words?
Let me give you my version of the story. My dad learned Taekwondo when he was a teenager. He is about 70 now, and he was born in 1953, so it was back in the 1960s when he learned the tuck-and-roll, the Taekwondo strategy of falling. When I was a baby, he would have been in his 30s, and I would not be surprised if he did a tuck-and-roll back then. But the fall I am talking about happened when he was in his 60s.
That was two days before my wedding date. He was hiking downhill in South Korea. I am originally from South Korea, and it was some mountain, I do not know the name. It was not that rough, it was paved, but it was pretty steep. There was something on the ground and he tripped over it. Because it was downhill he accelerated very quickly, but he used the momentum instead of resisting it, and he was able to just roll, and then stand up again. He did not get injured at all.
Because it was two days before my wedding day, he was shocked, and then very relieved that he was not injured. Otherwise, he would not have been able to attend my wedding. He knew I was doing the falls study, so he said: instead of just saying to people, do not fall, do not fall, you really have to teach them how to fall safely. There are falls that are inevitable. That is how we got started.
And that study was your own dissertation work?
Yes. I led the FAST study for my dissertation. That was back in 2016.
Editor’s note
A note on the name. Dr. Moon uses “FAST” here for her own dissertation study, following the label our interviewer brought to the conversation. They are two separate projects. Her dissertation work was registered as Safe Landing Strategy in Older Adults at the University of Illinois Urbana-Champaign and published in 2019. FAST, short for FAlling Safely Training (ClinicalTrials.gov NCT05260034), is the later randomized trial led by Dr. Jacob Sosnoff at the University of Kansas Medical Center, which built on her findings and carried safe-falling training to older adults at higher risk of injury. Dr. Neil Alexander, whom we also interviewed, is a co-investigator on it.
Two papers came out of that period, and they do different jobs. The first pooled 13 existing studies covering 217 participants. It compared seven protective landing strategies against unprotected falls, and the tuck-and-roll came out ahead, cutting hip impact force by 28 percent.1 The second put the question directly to older adults. Fourteen healthy participants, averaging 64 years old, were randomly assigned to tuck-and-roll training or an active control. They were tested at baseline, right after training, and again a week later. The trained group cut hip impact force by 33 percent against 16 percent in the control group, the skill transferred to the untrained side, and the effect held at one week.2 The control group improved too, which matters: they were doing something, not nothing. In the later FAST trial the control arm trained balance through the modified Otago program, and hip acceleration fell in both groups. What separated them was the head, where the trained group had lower acceleration and far lower odds of head impact. The tentative reading is that balance work lowers how hard you land, while the falling technique is what protects the head.3
On what the tuck-and-roll actually is
For a reader who has never heard the term, can you break the tuck-and-roll down into its parts?
The most important components are in the name, tuck and roll, but there are others. It is one of the martial arts techniques for falling, and it is specifically for the sideways fall. When you fall sideways, that has the highest risk of hip fracture, because you hit the side part of your hip, which is the most vulnerable part, so you can get fractured.
So instead of getting hit on the side, you rotate your body so that you can land on your buttocks. Rotate your body. Next, instead of resisting the fall, flex your knee, just sit down. Then tuck your chin to protect your head, and roll, instead of falling like a stick.
So sideways it is rotate, flex the knee, tuck, roll. What changes for a forward or a backward fall?
I did a systematic review of fall strategies based on falling direction.1 For the forward fall, what people were studying was blocking with your arm. If you do not block, you are going to hit your head right away.
Though not with a fully outstretched arm.
Right. That is to prevent a wrist fracture. So instead of blocking too hard, you absorb the energy with your arm.
And for the backward fall it is the same as the tuck-and-roll, after rotating onto your back. You are starting from a backward fall, so it is the same thing. It is a tuck-and-roll for the back.
On where the energy goes
Part of the roll is about distributing the momentum, I think, though I am not sure momentum is the right word.
I would say energy. Distribute the energy.
Distributing it across a large, cushioned area rather than one spot. Is that the principle?
Yes. Land on your buttocks. That is the instruction we give, because that is the most cushioning area of the body. And then instead of just staying there, roll with it, so that you can dissipate the energy.
Editor’s note
This is the part that carries best outside the laboratory. Researchers reviewed video of 2,388 real falls by 658 long-term care residents, average age 84. Injuries occurred in 38 percent of those falls, and about a third of the injuries were to the head. Rotating during the descent turned out to be the protective move. In the 43 percent of forward-directed falls where the resident rotated to land sideways instead, the odds of head injury were half those of everyone else.4
One finding in that study cuts against the advice above, and is worth stating plainly. Bracing with the arms appeared in 58 percent of these real falls, and it was associated with higher odds of injury, not lower. The authors are careful about why: bracing shows up most in forward landings, which are the hardest to survive well, so the arm may be marking the difficulty of the fall rather than causing the injury. Dr. Moon’s point is about how to brace, absorbing rather than locking the elbow, not whether to. But the real-world data is a reminder that an outstretched arm is not a reliable shield.
On why relaxing beats bracing
If someone has no martial arts background, are they naturally inclined to fall in a way that is not safe? Is this something they would have to train for?
My dad’s story is one of my motivations, but he was not the only one who pointed me toward this topic. I used to work with people with multiple sclerosis as research participants, and one of them told me he had fallen the night before but did not get injured, and that he never gets injured from a fall, because he knows how to fall. So, I asked him how he falls. He said: I just go with it.
Editor’s note
Falls are far more common in multiple sclerosis than the condition’s reputation suggests. Across international data, more than half of people with MS fall at least once in a three-month window and roughly a third fall repeatedly, a rate close to that of adults over 80.5 That comparison is about how often falls happen, not about age. Most people with MS are decades younger than 80. Onset before 50 is the rule, and pooled data put late-onset MS, meaning symptoms that start at 50 or later, at about 5 percent of all cases.6 The disease usually begins well before that, in young adulthood, though published estimates of the peak range vary somewhat.7 Balance and gait problems often appear early in the disease course, which is why people with MS feature so heavily in fall-injury research.
Instead of resisting the fall, just go with it. Which means, similarly, bend your knee, relax your body, and go with the energy instead of stiffening up and tensing your body, which results in falling like a stick.
And that is trainable?
People may not be able to perform a perfect tuck-and-roll. But if they are used to it, if they get constantly exposed to falling, then their mindset changes. At the moment of the fall they feel more relaxed, and they feel more confident that they are not going to get injured. When they relax, they can fall safely.
People in my research also ask me, do you think I can really use this strategy in real life? My answer to that is yes and no. Like my dad did, it is possible they perform a perfect tuck-and-roll and never get injured. But even if they do not perform a perfect, graceful tuck-and-roll, the point is not getting injured, not performing perfectly, like figure skating. This type of intervention is meaningful not because they can perform a perfect tuck-and-roll, but because it changes their mindset. They are more prepared, and they at least know that if they relax more, they will not get injured. Relaxing their body ends up with reduced injury risk.
On why older adults hit their heads
Dr. Sosnoff has described head impact as the part of his fall data he still cannot explain. Older participants hit their heads far more often than younger ones. The obvious explanations did not hold up. We put the same question to Dr. Moon, who saw the pattern in her own trials.
Did you see the same thing, that older adults hit their heads more?
I agree with him. When I did the data collection for the falls, I was worried they were going to complain about body aches afterward, from repeatedly hitting the mat across the fall trials. But when we gave them a questionnaire asking how they felt about the fall study, the thing they were most concerned about was that they kept banging their head.
When I did the pilot study with young participants, I did not see that trend. They were all able to hold their head up, and they did not bang their head. It was only after I did it with very healthy older adults that the banging of the head on the mat happened.
Editor’s note
Why the pilot group was younger. Deliberately tipping a person onto a crash mat is only ethical once the protocol is known to be safe, so the sequence runs young and healthy first, then older and healthy. That order is also what made the head-impact contrast visible, since the same protocol produced different behavior in the two age groups. The pattern was later written up formally, with Dr. Moon as a co-author, comparing head impact and head acceleration during experimentally induced sideways falls in 15 younger and 10 older adults. Across 147 analyzed falls, head impact occurred in 11.4 percent of the younger group’s falls against 34.5 percent of the older group’s, a statistically significant gap, and falls involving head impact produced markedly greater head acceleration than those without.8
Even in a healthy older adult, not only a frail one.
Yes, exactly.
Dr. Sosnoff’s hypothesis is weakening of the neck muscles. Do you have the same one, or a different one?
I agree with him. I thought about that as well. Why the neck? They are very athletic, even more athletic than regular young adults. So why the neck? When you go to the gym you do a lot of exercise that strengthens your legs and arms, but not to strengthen your neck muscles. So, I thought, although they are very healthy older adults, they never do neck muscle exercise itself.
That is just a thought I had. In order to prove that assumption you really have to do the neck exercise and then compare. So, I cannot say this is correct. That was my assumption.
If you were setting up that experiment tomorrow, what exercise would you have them do?
I really did not search for what it would be. But I know that martial arts practitioners do strengthen their neck. They put some kind of weight on, and there are weight training machines for it. I have not looked into it.
Editor’s note
This is the kind of idea that travels fast once it is in print, so it is worth stating plainly. Two biomechanists arrived at the same hypothesis on their own. It is a hypothesis, not a finding. Nobody has run the neck-strengthening trial that would test it.
On the three phases of a fall, and what balance protects
Balance predicts more than falls. Dr. Claudio Gil Araújo, whom we also interviewed, runs the CLINIMEX exercise cohort in Rio de Janeiro, which has tracked participants since 1994. In one analysis of 1,702 adults aged 51 to 75, those who could not hold a 10-second one-legged stance had about an 84 percent higher risk of death from any cause over a median of seven years. That held after adjusting for age, sex, body mass index, and existing conditions.9 We asked Dr. Moon where balance sits relative to the impact she studies.
Could you talk about reactive versus anticipatory balance, and about balance training generally?
I was mainly focused on the impact phase of the fall. There are three phases:
Pre-fall is anticipatory.
During the fall is reactive.
The third phase is the impact.
Both the anticipatory and the reactive phases are about preventing the fall, preventing hitting the ground. My part is the last one, the impact part. I did not really do much on reactive and anticipatory. I just know there are studies on it.
When I apply that concept to my own topic: in the laboratory setting, although we made it as unexpected as possible, they already know the fall is going to happen. So, they do a lot of anticipatory movement before they really fall. It could be a little different in a real-life scenario, where they cannot do any anticipatory movement.
But having better balance to begin with would help even at impact?
Yes, definitely. Fall injury is related to the fall impact severity, and impact severity can be impact force and impact velocity. With the correct anticipatory or reactive strategy you can reduce that, reduce the impact velocity, from the beginning.
Impact force and impact velocity are related but not the same thing. Velocity is how fast the hip is traveling when it reaches the ground.?10? Force is how hard the landing hits once that motion is stopped, which depends on how much of the body absorbs it and over how long.11 Better balance slows the descent, so it lowers velocity. The tuck-and-roll works on the other side of the equation, spreading the same energy over a wider area and a longer moment, which lowers force. Dr. Moon’s studies measure both, because either one can be what tips a fall into a fracture.
On teaching people who cannot break a fall with their arms
Dr. Sosnoff mentioned wanting to look at people with neurological conditions. Beyond the participant with MS, have you worked with that group on falling safely?
Not really, not that I know of. But here is what I found. When I worked in the rehabilitation hospital as my postdoc, right after I finished my dissertation, some people say this is a crazy idea, how can we teach people how to fall safely. Some people think that way.
But what I observed at the rehab hospital, where they have all kinds of people with injuries, was a patient with an arm amputation. What I found was that the physical therapist was actually teaching them how to fall safely, because they cannot block the fall with their arm. They were teaching the same strategy. If you are falling, do not go forward, twist your body, land on your back.
So, I have not done research in a systematic way on which other populations are exposed to this kind of safe falling training, but I believe there are some community-based interventions, or attempts.
On the ethics problem that pushed her toward AI
You are building technology to measure movement. Can you explain what it does?
What I figured out when I did the study was that we can only recruit very, very healthy older adults. I want to make sure this is an older adult study, but because of research ethics we have to ensure this is a safe protocol for the participants. So, I can only recruit very healthy, athletic older adults, and they are not actually at risk of getting injured from the fall.
So there is a dilemma there. I thought, what is the safe and ethical way to know how people are falling, how older people are falling? These days AI has advanced so much. People can put a video into an AI tool and do all kinds of analysis that we used to need a motion capture system for, which is the research-grade system only available in our laboratory setting.
But there are videos of real life. If I put a real-life fall that already happened into the AI tool, it can track what kind of movement, what kind of hip velocity, what kind of body configuration they are in at the moment of the fall. That is the technology I am pursuing, to understand how older people who really are at risk of fall injury are falling in real life.
Editor’s note
What the tool is built from. Dr. Moon did not name the software on the call, but her published work identifies it. Her group uses OpenPose, an open-source pose estimation system that locates 25 body landmarks in each frame of ordinary video, then runs the resulting hip positions through custom code to work out how fast the hip was moving at impact. Newer models such as WHAM track movement in three dimensions rather than two, which opens up joint angles during a fall.
How well it works. The method was tested on 110 videos of 13 older adults falling sideways, checked against laboratory motion capture. It measured two different things, and handled them very differently. Hip impact velocity, meaning how fast the hip was traveling when it reached the floor, came out close: average error near 7 percent, about 0.17 meters per second, with no systematic tendency to run high or low. Hip impact acceleration, meaning how abruptly that motion was arrested, came out much weaker, off by roughly 26 percent, and it consistently underestimated the hardest landings, those above 3 g. A g here is a unit of acceleration rather than force, and one g is the acceleration gravity gives a falling object. A hip decelerating at 3 g is being brought to a stop at three times that rate, which loads it with roughly three times its own weight. Those are the severest landings in this data set, so the tool is softest exactly where the stakes are highest, which is why the validation matters rather than being a technicality. The algorithm could still tell the three landing patterns apart: a stick-like fall hit hardest, a tuck-and-roll less so, and a knee block least of all.12
Where it is going. This work is funded rather than speculative. Dr. Moon holds a National Institutes of Health grant to study the biomechanics of real-life falls using pose detection algorithms, and a longer-running award to predict fall injuries from real-life falls using artificial intelligence. Both are validation and mechanism studies, not product development. There is no consumer app.
On what the tool can and cannot do yet
Is there a consumer product, or is it still in testing?
It is the validation stage. We want to see whether that AI algorithm can have similar accuracy to the motion capture system, which is the gold standard of motion tracking.
Would a single phone camera be enough?
It is just a single phone camera. But it is not at the phone app stage yet. We have to go through a very difficult installation process with a high version of the graphics card, a powerful computer.
Did you build the algorithm yourself?
There is an open source program. This is not something developed just for falls. It is developed for all kinds of movement analysis, and the program algorithm is shared open access. You download the program and install it on your own computer. I downloaded and installed it for my own purpose, which is fall analysis.
Do you see this ending up on a phone that people use at home, or more in a doctor’s office?
That is a really good question. That is actually the question I am facing right now. My analysis of the falling strategies, the falling movement behavior, is very accurate, and how am I going to translate that to clinical usage? That is the question I am facing right now.
Are you talking to your tech transfer office?
Yes. That could be one of the ways. Let’s say I was able to track the behavior during the fall correctly. Then I might develop an app that can measure how they fall in real life, and inform them: you are falling, but you are falling too hard, something like that.
Editor’s note
A tech transfer office is the unit inside a university that helps turn research into something the public can use. It handles patents, licensing, and agreements with companies, so a finding that works in a laboratory has a route to becoming a product or a clinical tool instead of staying in the journals. When Dr. Moon says the question she faces is translation, this is the machinery she means.
There is already existing technology for fall detection. As of now it is only monitoring whether they fall or not. It is more like a binary. We do not know the detailed content of how people are falling. I think if we gather that data, it can be the data that is useful for researchers and clinicians to understand the real way people are falling.
Do you have active participants now?
Yes, I have active participants. I have been able to collect the data at the impact phase, and what I am trying to do is apply the AI to see the reactive and anticipatory movement during the perturbation phase. That is an ongoing study right now.
References
1. Moon, Y. & Sosnoff, J. J. Safe Landing Strategies During a Fall: Systematic Review and Meta-Analysis. Arch. Phys. Med. Rehabil. 98, 783–794 (2017).
2. Moon, Y., Bishnoi, A., Sun, R., Shin, J. C. & Sosnoff, J. J. Preliminary investigation of teaching older adults the tuck-and-roll strategy: Can older adults learn to fall with reduced impact severity. J. Biomech. 83, 291–297 (2019).
3. Zanotto, T. et al. Strategies to Minimize Fall-related Injuries in Older Adults at Risk of Falls: The Falling Safely Training Study. J Gerontol A Biol Sci Med Sci 80, glaf076 (2025).
4. Robinovitch, S. N. et al. Protective responses of older adults for avoiding injury during falls: evidence from video capture of real-life falls in long-term care. Age Ageing 51, afac273 (2022).
5. Coote, S. et al. Falls in People with Multiple Sclerosis: Risk Identification, Intervention, and Future Directions. Int. J. MS Care 22, 247–255 (2020).
6. Naseri, A., Nasiri, E., Sahraian, M. A., Daneshvar, S. & Talebi, M. Clinical Features of Late-Onset Multiple Sclerosis: a Systematic Review and Meta-analysis. Mult. Scler. Relat. Disord. 50, 102816 (2021).
7. Reich, D. S., Lucchinetti, C. F. & Calabresi, P. A. Multiple Sclerosis. N. Engl. J. Med. 378, 169–180 (2018).
8. Wood, T. A., Moon, Y., Sun, R., Bishnoi, A. & Sosnoff, J. J. Age-Related Differences in Head Impact during Experimentally Induced Sideways Falls. Biomed Res. Int. 2019, 6804614 (2019).
9. Araujo, C. G. et al. Successful 10-second one-legged stance performance predicts survival in middle-aged and older individuals. Br. J. Sports Med. 56, 975–980 (2022).
10. van den Kroonenberg, A. J., Hayes, W. C. & McMahon, T. A. Hip impact velocities and body configurations for voluntary falls from standing height. J. Biomech. 29, 807–811 (1996).
11. Nasiri Sarvi, M. & Luo, Y. Sideways fall-induced impact force and its effect on hip fracture risk: a review. Osteoporos. Int. 28, 2759–2780 (2017).
12. Michaels, R., Barreira, T. V., Robinovitch, S. N., Sosnoff, J. J. & Moon, Y. Estimating hip impact velocity and acceleration from video-captured falls using a pose estimation algorithm. Sci. Rep. 15, 1558 (2025).
Dr. Eliezer (Eli) Lichter
Dr. Eliezer (Eli) Lichter brings a rare combination of hands-on clinical experience and rigorous scientific training to his work as a medical writer at Medical Guardian.
Before entering research, Eli served as an emergency medical technician in Yonkers, New York, one of the state’s busiest urban EMS systems, where he was recognized with multiple Lifesaving Citation Awards. That work put him on the front lines of exactly the crises Medical Guardian exists to prevent, including fall responses involving older adults. He later served in an administrative role at a large skilled nursing facility in upstate New York, where he managed the full operational complexity of senior care, including fall prevention protocols and incident response.
That direct, real-world exposure to what aging adults and their families face every day shapes everything he writes.
Eli earned his PhD in Biochemistry and Molecular Biology from the University of Nebraska Medical Center, where his research focused on the genetic mechanisms underlying neurodegenerative diseases, including Alzheimer’s and Parkinson’s disease. He then completed postdoctoral training at Boston University in computational biomedicine, including a research affiliation with the Broad Institute of MIT and Harvard, developing molecular and computational strategies for early detection of cancer, Alzheimer’s disease, and other serious conditions through advanced applications in genomics and epigenetics.
At Medical Guardian, Eli translates complex medical and scientific research into clear, trustworthy content that helps older adults and their families make informed decisions about their health, safety, and independence.
