We studied the five minutes before 1,039 falls. In 4 out of 5 cases, residents changed their position or actions at least 30 seconds before falling. They moved close to the bed edge, struggled to get up, or stopped to steady themselves against a wall. Spotting these changes could give care staff time to help.
At Teton, we believe this changes how fall prevention should be approached. Knowing who is at risk is not enough. Care staff also need to know when a resident is at risk.
Most falls happen when no one is in the room, leaving the moments leading up to them unseen. Traditional care technology largely responds after a fall has already happened. Teton is built to understand what happens before. Ten times per second, our AI reconstructs the room in 3D. It captures the position of the bed, chairs, and walls, along with how residents move and interact with their surroundings. This enabled us to conduct a large-scale study of residents’ movements in the minutes before a fall.
4 in 5 falls show an early warning sign
lay at the bed edge · ~4 min before
sat on the bed edge · ~2 min before
bent down to pick something up · ~2 min before
used furniture while upright · ~1 min before
No sign detected 30 seconds to 5 minutes before the fall
Fall
The study reveals how everyday movements can develop into situations where residents need help. A resident may drift toward the bed edge, struggle to stand, reach for furniture to steady themselves, or bend down to pick something up. Some of these signs appear minutes before a fall. We explore four recurring situations below, showing what happened and where staff might have had an opportunity to help.
Residents often move toward the bed edge minutes before falling
During the night, residents can slowly move toward the edge of the mattress before rolling or sliding out. Spotting this movement early gives staff an opportunity to help them back to a safer position. In a study of US emergency departments, more than half of bed-related fall injuries in older adults came from falling out of bed [1].
Lying at the edge of the bed is the earliest warning sign of a fall
Lying at bed edge or limb off bed
The two clips below show residents near the bed edge. In the first, a resident moves to the edge and falls out. In the second, a resident lies with their head hanging over the side. This is an opportunity for staff to help the resident back to a safer position.
Many falls begin when a resident gets up
Residents sat on the bed edge before 4 in 10 falls, usually about two minutes before they fell. Sitting up can signal that a resident is preparing to stand, giving staff an opportunity to help before they get to their feet. In our Anatomy of 2,000 Falls, residents getting out of bed were 2.5× more likely to fall [2].
More than 4 in 10 falls come after a resident gets up
Sat on the bed edge
Stood up from bed
Stood up from a chair
Stood up from a wheelchair
Some residents struggle to stand and pause before walking. Others get up easily and walk straight away. These differences help explain a resident’s fall risk. Clinical tests such as the Timed Up and Go [3] and Five Times Sit to Stand [4], [5] assess this by measuring how quickly people stand up and move.
The clips below show two examples. One resident struggles out of a chair and takes a break before walking. The other gets out of bed easily and walks straight away.
Our data shows that slower and less consistent stand-ups are associated with more falls. Before a fall, 29% of stand-ups took at least twice the resident’s usual time, compared with 14% normally. A resident who stands up more slowly than usual is more likely to need help from care staff.
Before a fall, standing up often takes longer than usual
Before a fall
Baseline stand-ups
The chart below shows how fall rates differ according to how consistently residents stand up.
Residents with more variable stand-up times fall up to 4× as often
Variation in sit-to-stand timeLowest variability → highest variability
Each bar represents one fifth of residents, ranked by how much their stand-up time varies.
Residents held on to furniture before nearly half of falls
Before 47% of falls, residents held on to furniture while standing. They reached for chairs, walls, or door frames to steady themselves. Teton’s AI maps the room and tracks these movements, showing when residents use their surroundings for support.
Nearly half of falls come after a resident leans on furniture
Supporting on furniture while standing upright
Standing still for 30 s or more
Walking
The two clips below show residents reaching for support. In the first, a resident stops and leans against a wall. In the second, a resident steadies themselves against the wall, then leans on the bed. Recognizing these moments could help staff see when a resident needs someone by their side.
How residents walk helps us understand their fall risk
Residents started walking before 19% of falls, typically about a minute before they fell. How they walk helps us understand their underlying mobility and balance. The two clips below show different walking patterns. One resident walks steadily across the room. The other moves slowly, taking short steps.
These everyday movements help build a picture of each resident’s baseline fall risk. Understanding their usual walking pattern can help staff plan the support they need, including a walking aid, physiotherapy or someone alongside them.
Residents stood still for prolonged periods before nearly a third of falls
Before 31% of falls, residents stood still for at least 30 seconds. These pauses can give staff an opportunity to check whether a resident needs help.
The two clips below show why the surroundings matter. One resident stands in the middle of the room with nothing to hold on to. Another leans against a door frame for support. Understanding where a resident has stopped and what they can reach helps staff assess the situation.
Residents who bent down more slowly had higher fall rates
Residents bent down before 11% of falls, usually about 2 minutes before they fell. Picking something up from the floor requires them to keep their balance as they bend down and stand back up.
1 in 9 falls comes after a resident bends down
Bending down
The two clips below show different ways of making this movement. A resident slowly picks something up from the floor, while a staff member bends down with ease.
Residents who took longest to bend down had roughly 4 times the fall rate of those who were quickest. The chart below shows how fall rates increased across groups with longer bending times.
Residents who bend down slowest fall 4× as often
Average time spent bending downShortest duration → longest duration
Each bar is one fifth of residents, ranked by their average time spent bending down.
Staff can use this information to make everyday tasks easier. Keeping frequently used items within reach or providing a grabber tool can reduce the need to bend down.
Only 18% of falls were not preceded by a change in movement or posture
We found no change in movement or posture between 30 seconds and five minutes before these falls. Some residents went from sitting or lying down to falling within seconds, leaving little time for staff to respond.
Most falls without a warning sign start in bed
Last seen in bed, without risk flags
Last seen in a chair or wheelchair
Last classified as out of view
Other last-state context
The two clips below show how quickly this can happen. One resident falls from a bed, while another falls from a chair. These sudden movements leave little time to recognize what is happening and offer help.
Understanding situational risk can help prevent falls
A fall happens in an instant, but the situation leading up to it often builds over minutes. Before 82% of the falls we reviewed, the resident’s posture or actions changed at least 30 seconds ahead. Those minutes show when a resident is moving into a risky situation.
At Teton, we believe a resident’s baseline fall risk is only part of the story. A risk score tells staff how likely a resident is to fall, not when. Getting out of bed, steadying against a wall, or struggling to stand after bending down can signal a need for help. Recognizing these moments can help staff see when a resident needs immediate support.
Teton’s Spatial Intelligence makes the room machine-readable while protecting residents’ privacy. The AI runs on a device in the room. No footage is stored or leaves the device. Only the information needed for care is sent on. By understanding how and why falls happen, we believe we can help care staff everywhere prevent more of them.