“AI precisely detects when a nursing home resident is at risk of falling.”

As a computer scientist, Harro Stokman worked on image recognition. He is now applying the technology in healthcare, to detect when the elderly fall, and prevent it. ‘With our image recognition, scarce healthcare personnel can be better utilized.’

What is a chair? For a human being, that is a dead simple question. Four legs or not, armrests or not, when you see one, you know it. For a computer, it’s a completely different story. It took decades before programmers finally managed to create a program that could distinguish a chair from other objects with some accuracy.

It seems a somewhat trivial issue to have whole armies of bright minds devote their time to. Harro Stokman (56) was one of those masterminds as a doctoral student in computer vision at the University of Amsterdam.

With a few colleagues, Stokman started the company Euvision. That made headlines with the app Impala, which could classify photos on a phone: dogs, cats, sailboats – and, for the enthusiast, chairs. Such a smart algorithm on a cell phone, that had not been seen before. US-based Qualcomm bought the company in 2014.

Relevant application

After a few years employed by the tech giant, it was time for something different. Artificial intelligence, meanwhile, had gotten a lot “smarter. It was now possible to infer from moving images what people were doing. But it was searching for a socially relevant application.

The eye fell on elder care. “They did work with cameras and sensors in that sector, but it was all very old-fashioned and they gave a lot of false alarms,” Stokman says. His new company, Kepler Vision Technologies, uses image recognition to record whether people are lying down, (almost) falling, walking around, sitting on the floor, or leaving the room.

Again, a touch trivial? Not at all, says Stokman. Nursing homes with 100 residents often need three nurses during a night shift. These come into action when a false sensor goes off. Often it turns out to be a false alarm. Or protocol dictates going into each room a few times a night to see if everything is still okay. Both scenarios are common as well as disruptive. For the nursing home resident because of the broken night, for the caregiver who is called up unnecessarily and for society because this way of working takes a lot of manpower.

Working more efficiently

Kepler Vision hangs a smart sensor in the room, which will sound an alarm if anything threatens to go or goes wrong. A timely alarm can help prevent much misery. The system continues to learn, almost never misses an incident and false alarms are very rare: one per sensor every three months. This allows clients to better utilize scarce healthcare personnel.

A sensor in your bedroom that is always on, isn’t that a privacy nightmare? No, says Stokman, because the system does not recognize people, only activities. And the sensor does not send images, only a text message when someone falls. In addition: a caregiver no longer enters the room every few hours.

Healthcare is often cautious about adopting new techniques, Stokman notes. Its systems now monitor more than 4,500 clients. Kepler is almost breaking even, Stokman said. He wants to see sales, about €1 mln in 2023, grow by a factor of three each time over the next three years.

To make that happen, Kepler is working with major technology suppliers who can market the product to their customers. A loan of nearly €2 mln secured in January from the regional development company InWest, among others, provides room for growth. That there is now competition, Stokman thinks, is a good sign.

Read the full article: https://fd.nl/tech-en-innovatie/1502187/ai-ziet-precies-wanneer-een-verpleeghuisbewoner-dreigt-te-vallen