Kinometrix Uses Predictive AI to Reduce Dangerous Falls in the Hospital

CEO Devina Desai is leveraging her experience in patient safety, and close ties to health systems, to develop a fall risk platform that’s designed to reduce unnecessary injuries, curb physician burnout, and save hospitals millions.

Challenge

Falling down might not seem like a big health issue. But even for seemingly fit and healthy people, falls can lead to fractures, head injury, fear of falling, long-term disability, or death. You might think that patients already in the hospital would be protected from dangerous falls that could aggravate their condition or cause new health concerns. But counter-intuitively, in the one place where you hope you receive constant care, there is a risk of adding “injury to injury” from falls.

Falls are the most common preventable safety events in hospitals because facilities are often ill-equipped to predict and prevent them. Research shows that each year, about 700,000 to one million patient falls happen in hospitals in the United States. These falls lead to about 250,000 injuries and 11,000 deaths.

Falls present a considerable cost for hospitals too. A 2023 study published in JAMA Health Forum estimates that each patient fall costs about $62,521.

Not to mention that treatment from a fall in the hospital doesn’t get reimbursed because associated injuries or harm are recognized as a “never event.” The Centers for Medicare and Medicaid Services (CMS) established the concept of “never events” to prevent “clearly identifiable, preventable, and serious” hospital-acquired conditions (HACs) and promote patient safety. And as of 2008, they stopped reimbursing hospitals for caring for these conditions.

Human sitters — healthcare workers that stay with patients to ensure their safety — take up already limited resources, and more evidence is needed to show they work well at preventing falls. Camera or thermal bed monitors work precisely as intended; they monitor and alert clinicians when a fall occurs. They cannot predict falls and may even contribute to clinician fatigue. Manual fall risk assessment methods like Johns Hopkins Fall Risk Assessment Tool (JHFRAT), St Thomas’s Risk Assessment Tool in Falling elderly inpatients (STRATIFY), and Falls Risk Assessment Tool (FRAT) are subjective, undependable, and imprecise.

Addressing this persistent health challenge using predictive AI is Kinometrix President and CEO Devina Desai.

Origin Story

For many years Devina Desai worked in patient safety at Inova, a large healthcare system in Northern Virginia. Her interest in patient safety grew while doing clinical service and nursing research at Inova Fairfax. Her research role helped her better recognize the real and heavy cost of patient safety issues, particularly venous thromboembolic events (VTEs) at the hospital.

During her tenure at the hospital, she worked with a vendor called Kinometrix to validate their physical therapy device and data analytics platform. As part of her department’s validation process, Desai discovered the device did not significantly impact outcomes. However she was impressed with the data analytics side of the platform. She and her team became curious if this could be used to address a more pressing concern — inpatient falls.

Kinometrix needed to pivot, and that meant identifying a leader who could guide a new vision. Innova and Kinometrix tapped Desai because of her deep understanding of patient safety, her work at a large health system, and her close working relationships with nurses and other providers.

Desai transitioned from Inova to Kinometrix and developed its first product, which now addresses the shortcomings of the traditional fall risk assessment systems and delivers a more accurate prediction of a person’s fall risk.

Today, Inova and Kinometrix have a collaborative relationship. The hospital has provided years of inpatient data to help the startup develop and validate its machine-learning platform and will likely be one of the first health systems to use the product.

Under the Hood

Hospitals gauge a person’s susceptibility to falls using manual assessment tools like the Johns Hopkins fall risk assessment, which uses a checklist to determine a patient’s fall risk score. This risk assessment method is subjective and time-consuming, resulting in inaccuracy. It also cannot dynamically capture a person’s circumstances and risks as they change.

Kinometrix uses real-time data from a person’s electronic health record (EHR), with information like a person’s demographics and medications, to create an objective score of a person’s fall risk and risk drivers of the score. It also updates this risk score as information in a person’s EHR changes.

With the help of the Kinomterix fall risk assessment platform, clinicians can appropriately allocate fall prevention resources based on each patient’s personalized risk score.

Kinometrix operates on a machine-learning module. It’s constantly fed additional data to make it robust and more comprehensive. The more data that runs through it, the better it works.

Specifically, Kinometrix uses more variables than subjective assessment tools and about half a million patient records for its machine learning module for higher precision and accuracy than other risk assessment tools. Currently, it’s greater than 95% accurate at determining a person’s fall risk. For comparison, that’s about as accurate as a home pregnancy test. On the other hand, the accuracy of a manual fall risk assessment tool relies solely on the person using the tool. And that changes depending on how well the person uses it.

Plus, Kinometrix is easy to set up and runs instantly. It integrates seamlessly and works with any EHR vendor.

Fortunately, this fall-risk assessment platform is the first of the many patient-safety innovations the founders hope to build with Kinometrix.

Pressure injuries are another never event and a much more expensive problem than falls, costing healthcare systems around $35 billion annually, compared to falls, which is $7 billion. Desai predicts that the next Kinometrix innovation will be a risk assessment platform that works better than and will replace subjective manual risk assessment methods like the Braden scale.

Why We’re Proud to Invest

We’re proud to back Kinometrix because hospitals need a better way to predict and prevent falls. The toll of the status quo — in injury, death, and cost — is just too high. Falls and other areas of patient safety have plagued the healthcare system seemingly forever, and there’s been too little innovation applied to the sector.

But not just any startup could tackle such an entrenched problem. It’ll take a founder who intimately understands patient safety, hospital administration, and healthcare provider needs. We find that trifecta of experience in Devina Desai.

Finally, we’re excited to back the Kinometrix team because fall risk assessment is just the beginning. The founders’ goal is to develop a suite of products that promotes patient safety, reduces clinician workload, and conserves hospital resources and money.

Kinometrix protects lives, saves the healthcare system billions of dollars, prioritizes clinician needs and time, and lowers the burden of catering to never events. Join us in welcoming Kinometrix to the StartUp Health family!

→ Connect with the Kinometrix team via email.

Passionate about breaking down health barriers? If you’re an entrepreneur or investor, contact us to learn how you can join our Health Equity Moonshot.

Investors: Contact us to learn how you can back Health Transformers and Health Moonshots.

Digital health entrepreneur? Don’t make the journey alone. Learn more about joining StartUp Health.

Follow us on social media for daily updates on Health Transformers: Twitter, LinkedIn, Facebook, and Instagram.

Kinometrix Uses Predictive AI to Reduce Dangerous Falls in the Hospital was originally published in StartUp Health on Medium, where people are continuing the conversation by highlighting and responding to this story.

healthcare
health
hospital
device
digital health
ai
electronic health record
machine learning
data analytics