AI and machine vision addresses falls prevention

VirtuSense uses machine vision and AI to address the multi-million falls prevention market opportunity.

According to the U.S. Centers for Disease Control and Prevention, falls are the leading cause of fatal and non-fatal injuries among the elderly in the United States. Falls result in more than 800,000 hospitalisations and the total cost of fall injuries is greater than $50 billion annually. [1]

Longevity.Technology: Sometimes, our mapping of the Longevity landscape is questioned. Surely it’s the research and development of rejuvenation therapies that’s proper Longevity? Well, not really, if you consider any intervention that extends life and ensures the quality of that extended life, then Longevity is actually quite broad (we’re stopping short of classifying the Mediterranean Diet, or a pair of trainers, as Longevity technologies though) … here’s a great example of this:Β 

Given such damning statistics associated with the sometimes life-changing effects of a fall in laters years, it’s little wonder that fall prevention is a key concern for healthcare organisations, care home providers and society in general. And technology definitely has a part to play.

Illinois-based start-up VirtuSense is at the forefront of this sector with an AI platform for fall risk analysis and prevention. The company’s founder, Deepak Gaddipati, was compelled to start the company when, in 2009, he had first-hand experience of how falls can have tragic consequences.


“My grandmother was 68 years old, she felt broke her hip, and she died within 10 days after the fall …”


β€œMy grandmother was 68 years old, when she fell and broke her hip, and she died within 10 days after the fall,” he recalls. β€œAnd it really struck me afterwards that someone had to fall to become a fall risk, which sounded really backwards to me.”

Most people might just have accepted this as one of the perils of aging, but Gaddipati was better qualified than most to actually do something about it. He had built a career in machine vision and developed the first commercial full-body, automated scanning system that is widely deployed today across most U.S. airports. Crucially, he had also worked on machine learning projects for the US Army that involved motion tracking. He realised that this technology might be exactly what was needed to combat falls in the elderly and licensed it from the Army in order to start VirtuSense. The company was built on a relatively simple idea.

β€œFirst, you need to know who’s a fall risk, why they’re a fall risk, and what can you do to prevent them from falling,” says Gaddipati.

VSTBalance uses machine vision and AI to assess fall risk.

VirtuSense’s first product, VSTBalance, combines a 3D sensing hardware element with the company’s proprietary AI technology, which care staff or therapists can use to assess a person and objectively identify deficits in balance, gait, and function – the three leading indicators of fall-risk.

β€œThink of it like a CT scanner that uses infrared,” says Gaddipati. β€œIt reconstructs the whole area in 3D and assesses how people move in that.”


… the VSTBalance system uses evidence-based testing to objectively figure out if someone is at risk …


In about two minutes, the VSTBalance system uses evidence-based testing to objectively figure out if someone is at risk and automatically generates reports to help care professionals create more effective care plans. The assessments being conducted by the system are all existing evidence-based assessments and simply enhanced by the use of technology.

β€œIn the past, these assessments were conducted by someone observing with their eyes,” explains Gaddipati. β€œThe human eye is good, but not at observing fast changes. If someone walks 10-15 feet, you only have about 10 to 15 seconds to see what is happening.”

VSTBalance tracks 26 different joints on a person in real-time 3D – without any markers attached to the body – and assesses how the joints are moving when they’re doing a set of specific tasks: walking or getting up from a chair, for example. Based on this data, the system is able to categorise the person as either a low, medium or high fall risk and the appropriate care plan is then provided to help them improve, ranging from exercise through to physical therapy or even primary care for high risk patients.

VSTBalance automatically generates reports to help create more effective care plans.

Now deployed in 60-70 health systems across the Unites States, VSTBalance has already been used to assess hundreds of thousands of older people, both in primary care and senior living environments, and the results are compelling.


“… 12 months after implementing the technology … there was a 74.3% reduction.”


β€œWe did a big case study with 37 different nursing homes with about 4,000 patients in 2018,” says Gaddipati. β€œWe recorded the number of falls with injuries before they implemented the technology and then again, 12 months after implementing the technology. And there was a 74.3% reduction.”

β€œThese are CMS-reported outcomes. And now we are in a position where CMS is directly looking at our data – they’re even buying a few millions dollars’ worth of product from us and putting it in nursing homes to prevent falls.”

But VirtuSense did not stop at risk assessment. Gaddipati recognised that the technology could also be adapted to improve fall alerting as well. The most common technologies currently used for fall alerting are pressure pads that aren’t predictive and create a lot of false alarms.

β€œIf you’re a poor nurse who’s in charge of 30 beds in a ward, you’re probably listening to somewhere between 400 to 600 false alarms a day,” says Gaddipati. β€œSo they become desensitised to them.”

We addressed this issue in our very insightful interview with Dr Lorraine Morley.

VSTAlert uses AI to predict when someone is potentially going become a fall risk.

For this issue, VirtuSense has developed its VSTAlert system, which is basically a sensor technology that goes on the wall – typically next to the TV in nursing homes or hospitals. And it uses AI to identify the intent of someone getting up.

β€œOur AI knows when a high fall risk patient is getting up on average 30 to 65 seconds before they actually do,” says Gaddipati. β€œAnd then, rather than just beeping, it talks to them and says β€˜Please don’t get up – someone is coming to help you.’”

At the same time, the system automatically alerts the appropriate staff to come and take care of the patient.

β€œAs a result, we’ve been able to reduce falls with injuries by 65% to 70% for these critically sick patients,” says Gaddipati.

Looking to the future, VirtuSense is preparing to launch an in-home version of its alerting system, which will assess much more than just fall risk. Currently in beta testing, the system will also assess factors such as eating and drinking, taking medication, gait speed, frequency of toileting and so on.

β€œIt does all that in an automated fashion, without the need for them to touch or hold or do anything,” says Gaddipati. β€œAnd once they start deviating from their normal pattern, our AI is monitoring all this data and can alert the appropriate people.”

The in-home system is expected to launch in late 2020 or early next year and, impressively, having secured β€œa few million” in funding from the Ziegler Link-Age Fund in 2018, Gaddipati is not expecting to have to raise more in the near term.

[1] https://www.cdc.gov/injury/features/older-adult-falls/index.html