From a Fitbit for cows and a tag for access control to intelligent language technology for healthcare (more on that later); there are many ideas for AI application in Nedap’s beautiful headquarters in Groenlo. Nedap calls itself a technology company pur sang, and they were one of the pioneers in the development of RFID in the 1970s. This contactless communication technology is now used on a large scale worldwide, including in the OV-chipkaart.
From hardware to software
Director Ruben Wegman: “We were good at putting things together, as an equipment factory. We still do that, but we have increasingly shifted from hardware to software. Artificial Intelligence plays a big role in this.”
“You need entrepreneurship for valuable AI Projects.”
Applying AI requires several elements, says Wegman. “First of all, you need to know what you are talking about when it comes to AI – you need to have sufficient technical knowledge. Secondly, you need domain knowledge; you need to know what happens in a specific market in great detail. You will also get nowhere without enough usable data, so that is the third requirement. Financial resources are also important. And finally: entrepreneurship. You have to be able to come up with scalable propositions and valuable projects.”
The hub as a centre
According to Wegman, AI hub East Netherlands should help parties with meeting these five requirements. “What do parties or projects lack, and how can we help them? The hub should make the right connections; it has to become a place that binds people together. Many entrepreneurs need such a centre for research and support, an accessible place that gives you access to the rest of the AI system – from here up to Brussels. The hub also brings education and business closer together. We notice that professors are getting better at finding us, simply because people on the market are talking about the AI hub.”
“Professors are getting better at finding our company, simply because people are talking about the AI hub.”
Technology in healthcare
Nedap is being led by the market more and more. They are no longer just applying the things they are good at; they are developing what is needed. No management layers, but plenty of freedom and responsibility for employees. They aren’t over-organized, either, according to Wegman: “We are less domain-specific and more and more hybrid.”
The technology company does have a few target groups on the market, of which the healthcare sector is an important one. Data science manager Thomas Markus: “There is increasing pressure on specialized staff in healthcare. This can be partly solved with the right training and partly with technology.”
An example is Ons Wondzorg (Our Wound Care), an application that monitors wounds. “A nurse can use our app to take pictures and analyze the wound instead of having to describe it. The app monitors if the wound is healing properly and indicates if it is necessary to call in a specialist. Leaving this analysis to technology saves time and creates space for more attention and better care, both for the patient and the specialist.”
Discovering trends
Natural Language Processing (NLP) is also being increasingly used in healthcare. Markus: “There’s a lot of data in healthcare, but this data is not always entered in a structured way. Healthcare organizations can more easily discover trends when this data becomes more accessible. Whether the number of diabetes people or people with dementia is increasing or decreasing, for example; whether the care needs are equally distributed across teams and if other automatically identified developments may be related to this.”
“AI tremendously improves healthcare, both for patients and healthcare professionals.”
An example of the previously mentioned intelligent language technology for healthcare is FutureType. “It is a tool for text prediction in reports and patient records. Just like, for example, Google makes suggestions for your Internet search based on several characters or words, FutureType does this for complex medical texts. The suggestions are partly based on a person’s healthcare profile while still preserving their privacy.” Doctors and nurses have had to enter as many as two million fewer words in the short time that the system has been up and running, thanks to the self-learning system. “This saves time and reduces the likelihood of errors.”
Setting boundaries without restrictions
In short: AI helps improve healthcare. But again, all five requirements must be met for a successful application, according to Wegman. “Finding enough usable data can be a challenge. There is plenty of data in healthcare like my colleague just said, but not all of this is available due to privacy concerns. This is the perfect example of something that the hub, and especially the overarching Dutch AI Coalition, can help with. Together, you can determine frameworks and create conditions. We have to set boundaries without restricting innovation too much.”
