Digital innovation in healthcare: entrepreneurial lessons from Kepler Vision Technologies

Few sectors are as in need of digital innovations as healthcare. In elderly care, the already pressing staff shortage will increase by another 20% over the next five years, partly because the demand for care is only growing due to an aging population.

Kepler Vision Technologies has developed the Kepler Night Nurse: a system that uses smart sensors to very precisely detect patients’ movements. How did the company, originally a UvA spinoff, transform an innovation into a successful product for the healthcare market? ROM InWest asked CEO Harro Stokman for an explanation to inspire other digital ‘healthcare entrepreneurs’. (Dutch version here).

“A gigantic problem is looming in healthcare,” says Harro Stokman of Kepler. “A quarter of healthcare staff will retire in the next five years, and the number of elderly people continues to increase. It’s a double aging issue. A few months ago, the NRC already spoke of a ‘care infarction’. If we don’t intervene, we’re really going to get stuck – perhaps first in elderly care. The sector needs efficiency more than ever.”

Image recognition app

In 2010, Harro encountered a modern problem: an endless stream of photos on his iPhone where he struggled to find specific moments. Together with his colleagues from UvA – where he was working on his PhD in Computer Vision – he developed Impala, an app that can recognize objects and actions in photos. “By distinguishing dogs, cats, a sunset, or a scene, we ‘taught’ the app to categorize images,” Harro explains. “But AI developments quickly progressed. AI also began recognizing human actions. So whether a person is sitting, standing, smoking, or eating. The technology fascinated me immensely, and I soon thought we should use it for something societal. The staff shortage in healthcare gave me the idea to use the software for camera footage in nursing homes.”

Puzzle pieces

Switching to the healthcare sector brought challenges. Harro: “We had a few puzzle pieces to put together. For instance, healthcare mainly uses cameras with a so-called fisheye lens. These not only look straight ahead but also to the sides, capturing much more in view. Because existing computer vision technology didn’t align with this, we had to adapt it to its technical core first. A second problem was the reliability of the AI. When the software once raised an alarm during the holidays, the healthcare staff found not an injured patient but baby Jesus in the nativity scene. By now, we have advanced to the point where the technology analyzes patients’ lying positions and can generate reports based on the data.”

Scalable solution

Harro asserts that the technology is now rock solid. “Healthcare workers receive alerts based on image analyses not only in the event of an accident but also to prevent accidents – for example, if a client spends too long in the bathroom or sits on the edge of their bed. Because the sensors recognize the uniforms of healthcare workers, the alarm only goes off for clients.” Thanks to the investment from ROM InWest, we can boost marketing and sales to further introduce the innovation of Kepler Night Nurse to the market. My goal is for the software to watch over a million clients by 2030. The ultimate aim is to keep healthcare affordable and of high quality.”

Tips for entrepreneurs

What lessons can other entrepreneurs in digital healthcare solutions learn? Harro advises shadowing healthcare staff. “They know best what they need. And reassure your target group – healthcare professionals – that they are not being replaced by technology. Smart technological solutions are there to support staff, enabling them to do their work more efficiently and effectively.”

About Kepler Night Nurse

The already extreme staff shortage in healthcare will increase by another 25% over the next five years, while the demand for care continues to grow due to the aging population. Better detection is crucial to relieve the staff: with existing sensors in nursing homes, it often takes five minutes for someone to arrive when a client falls from their chair or bed. Many alerts also turn out to be false alarms, such as when someone accidentally presses the alarm button. The smart sensor of Kepler Night Nurse sends an alert within 10 seconds of a fall and prevents accidents by precisely detecting patients’ movements – such as getting up from a chair or sitting on the edge of the bed. This allows staff to arrive sooner and reduces false alarms by 99%.

Kickstart at UvA

Kepler Vision Technology is being led by CEO Harro Stokman, former UvA researcher in Computer Vision. It was founded in 2018 with intellectual property generated by Cees Snoek and Arnold Smeulders. Kepler got a kickstart with an initial investment from UvA Venture. The initial setup was to explore computer vision cases. Later on, the idea came to focus on care home fall detection.