Forecasting health patterns

How Artificial Intelligence can improve the health of people with intellectual disabilities

Department of Social Sciences

Is there a link between eating peanut butter and aggression in people with Down's syndrome? Can electronic patient records allow early detection of breast cancer in women with intellectual disabilities? These and other questions are researched by Wageningen University & Research’s PhD candidate Joep Tummers, who is building an AI-driven platform to improve the health and well-being of people with intellectual disabilities.

This research contributes to Sustainable Development Goal 3: Good health and well-being.

About 0.85% of the Dutch population has an intellectual disability, meaning they have an IQ below 70. The life expectancy of these people is lower than that of the average population, in part due to them not receiving the required care because signals may have been missed. PhD candidate Joep Tummers is building an artificial intelligence-driven platform that could change this.

Within the project Sterker op eigen benen (Stronger on your own feet) led by Radboud University, Joep Tummers of the Information Technology (INF) group at Wageningen University & Research is building a platform that makes responsible and safe use of data from care and medical systems to identify illness and other problems in a timely manner.

“People with a mental disability often live in care institutions. They have to deal with a multitude of care providers, such as counsellors for daily care, specialised doctors, dentists, and general practitioners. Each of these care providers report on the care provided and the condition of the client in their own digital systems. All that information is therefore scattered across various information systems that all have their own language and format,” Joep Tummers explains.

We know, for example, that the length of daily reports can be a predictor of the likelihood of incidents

These systems don’t communicate with each other, which is a pity, according to Tummers, because all those files are an unprecedented potential source of information to improve care for people with intellectual disabilities if we could combine that data to discover patterns in them. “We know, for example, that the length of daily reports can be a predictor of the likelihood of incidents: as healthcare providers’ daily reports become longer, it often turns out that an aggressive episode is imminent. Another example is that changes in eating patterns or people who become introverted could be a signal for illness. It’s important that diseases are diagnosed in time. It’s also well known that people with intellectual disabilities are less likely to come to hospital for oncological care, for example,” says Tummers.

Machine learning

Machine learning, artificial intelligence, natural language processing, and text mining are key terms in Tummers’ work: “It’s key to move from data to information and then to knowledge. Anonymity and confidentiality of those data are, of course, of the utmost importance,” he emphasises. “In our group, we work a lot in the field of agricultural data. Yet whether you’re talking about people or cucumbers: the technologies and methodologies are rather similar. Privacy, on the other hand, is a completely different story.”


  • Many different systems with each their own language and format
  • No communication between systems
  • Information is scattered across platforms


  • One integrated platform for healthcare providers and clients alike
  • Platform that can combine data to discover patterns

Want to know more about AI and machine learning?

Big data platform

“The ultimate goal of the project is to create a big data platform where healthcare professionals, interested parties, researchers, and people with intellectual disabilities can address all kinds of questions related to illness and behaviour of people with intellectual disabilities. These questions can range from finding out whether there is a link between eating peanut butter and aggression in people with Down’s syndrome to whether electronic patient records allow early detection of breast cancer,” says Tummers. Research questions are submitted via Crowdience, an easy-to-read and accessible web platform that was launched at the end of 2019.

Tummers says that there is a need in the healthcare sector for a platform like this. It proved to be necessary at the time of the Covid lockdown: healthcare institutions wanted to know whether there were more or fewer incidents in institutions during the lockdown as a result of the Coronavirus outbreak. “Within two weeks, we were able to provide feedback, letting them know that the number of incidents during the lockdown initially decreased by 20% and then gradually returned to the old level. Doctors and healthcare providers can analyse these results and learn from them. Thus, we use information technology to help improve the quality of life and care for people with intellectual disabilities.”

Do you have questions about using artificial intelligence in healthcare? Ask our expert:

Joep Tummers

PhD candidate Information Technology

Share article