AI research forms the foundation for tomorrow's health applications

30 Sept 2021


Artificial Intelligence (AI) is ostensibly one of the most discussed but least understood technologies today. It often raises substantial questions such as: Why is AI necessary in our lives? What positive contribution can AI make to our society and our economy? Does AI provide added value or is it a threat? What can we expect from AI in the coming years? What impact will AI have on our daily lives? To answer these and other questions, Sabine Demey, Program Director of the Flanders AI Research Program, talks about the start of the program and the breakthrough results so far.

The Flanders AI Research Program is initiated by the Flemish government, namely the Department of Economics, Science, and Innovation (known by the Dutch acronym “EWI”). A unique, intensive cooperation has been established by and between universities and research institutes within the confines of the program. Each of them cooperates in one or more research projects with the goal of developing AI methods that can be developed further in our society. A unique cooperation by and between our AI experts (known to us and to the rest of the world) whom we could simply call the 'Red Devils or Red Lyons of AI': our A-team that combines expertise with great passion and commitment to boost AI research. Sabine Demey elucidates the importance of the joint research program: "The research we conduct is of strategic importance to boost the application of artificial intelligence in Flanders. We focus on the complex and generic questions for which there are currently no ready-made solutions. So, we need to develop new knowledge that can then be converted into concrete AI applications. This research forms the foundation for applications of tomorrow. Under the Flanders AI Research Program, some 300 doctoral students, senior researchers and professors engage in strategic basic research, applied in three areas: healthcare, Industry 4.0, and applications for the government and its citizens. In this article, we focus on the impact of AI in healthcare.” 


During the COVID-19 pandemic we have seen a great impact of strategic AI research in the healthcare sector. When the Flanders AI Research Program was established, we had never heard about COVID-19. But at the start of the pandemic, the research groups brought their knowledge and AI methods to bear to help vulnerable groups of patients and to come up with answers to urgent and fundamental questions in healthcare.  
The research teams of the KU Leuven and the VUB focused on AI-based software for analyzing CT images of lungs against the background of the COVID-19 pandemic. These medical images provide useful information to gain a better understanding of the COVID-19 syndrome and to monitor its development in patients. Thanks to the AI software, abnormal areas in the lungs can be identified more quickly and more accurately on CT images. This has led to a rewarding breakthrough in the healthcare sector. Additional grants have enabled the research to continue in hospitals and research centers. The results were recorded and integrated into the cloud-based image processing service of icometrix NV. Meanwhile, this AI technology is improving the performance of hospitals here and in many other countries. More than 800 hospitals in Europe use the service daily to process CT images. 


A few vulnerable groups of patients were also alerted when the pandemic started. For example, patients with multiple sclerosis (MS) wondered whether they were more susceptible to COVID-19, whether they could still undergo their treatment, whether the virus would trigger a new MS attack, and so on. To be able to provide correct medical advice, the MS Data Alliance, with UHasselt [University of Hasselt] as its coordinator, launched a study in record time. A database was built together with researchers from the Flanders AI Research Program to host the health data of MS patients from more than 80 countries. These data were collected and analyzed to adjust the COVID-19 advice for people with MS based on data-driven insights. This medical advice is now available in 14 different languages around the world. 


The Flanders AI Research Program will naturally cover many more medical applications. Sabine Demey explains the focus: "We are geared to methods that can support and improve healthcare in the coming years. Thanks to AI, people can get the right diagnosis and treatment faster, plus a qualitative follow-up that monitors the patient in all circumstances, in the hospital and at home."  

One high-profile area of this research is single-cell technology. To gain a better understanding into how our bodies work and to be able to map out the immune system in detail, this technology focuses on the study of tissues and cells in unprecedented detail. This can lead to revolutionary results through ‘bespoke' treatment for patients in oncology, neurology, immunology, and stem cell research. You can make that step towards precision medicine if you manage to turn the detailed insights into new treatment methods. Single-cell technology can thus have an extraordinary impact on the pharmaceutical and diagnostic industry. But... to gain substantiated insights, you first need to input and analyse a huge volume of data. And that is a real bottleneck. That is why the UGent/VIB research group is working within the Flanders AI Research Program on the development of new methods to analyse single-cell data.  

Some 60,000 people have epilepsy in Belgium. Despite their medical treatment, 1 in 3 suffers from sudden and unexpected seizures. Most epileptic seizures occur at home, not in the presence of caregivers and hospital equipment. The goal within the program is therefore to develop a method that can help the patient outside the hospital environment. KU Leuven, Ghent University, and Hasselt University are working together with UZ Leuven to develop an AI method that predicts as accurately as possible when a patient will have the next epileptic seizure or when the risk of a seizure increases significantly. Patients can monitor themselves better with this information, while the treating neurologist can get a better idea of the patient's medical condition. 


In 2020, 1 in 5 people living in Flanders was 65+. And that number will increase in the coming years. The percentage of hospital admissions increases with age, so we expect the pressure on our hospitals to rise as well. The top priority in the health sector is to provide patients with optimal care. More and more hospitals are consequently looking at AI to plan the treatment of their patients better, so that patients spend less time waiting for their treatment, surgery, diagnostic admission, etc. It is important for hospitals to make the best possible use of the time patients spend on their premises. The length of stay is an important parameter for the planning and organizing of operations and patient treatment, staffing and the use of technology and logistics. An optimal patient flow can therefore lead to greater patient satisfaction as well as cost savings for the hospital.  
Under the Flanders AI Research Program, the KU Leuven team, in cooperation with UZ Leuven, is looking for an AI-based approach that takes the patient's specific needs, condition and development into account to chart an optimal hospital planning.  



We end where we started, with two issues that have dominated our lives in the past year and a half: COVID-19 and data. To deal with the pandemic and to create an exit strategy, we rely on data insights. At the tip of the data iceberg, we find daily overviews of the number of infections, hospital admissions, the status of the vaccination campaign, and so on. Also, our call data are used to gain an insight into home working. Citizens can rarely decide what data they want to share, with whom, and for what purpose, however. 
Therefore, VITO [the Flemish Institute for Technological Research] and the universities of Hasselt, Ghent, and Brussels aim to develop a platform for citizen-managed health data based on Solid technology. The technology is linked to careful principles about the ethical use of data and AI algorithms. You are at the helm of your personal data vault that is filled with your health data. You manage the data and other institutions and/or companies may use your data (anonymized or not) only with your prior consent. You can consequently choose to share your data for personal services, or for research and development. In that respect, sharing your health data can eventually contribute to better preventive – rather than just curative – care. 
For Sabine Demey, the challenge ahead is clear: "Living longer in a healthy manner is the ambition of most of us. In case of health-related issues, it is important to get the correct diagnosis as fast as possible and a personalized treatment, fit to the specific situation of the patient. Telemonitoring with medical wearable devices will monitor the patient at home, and if admission to hospital is necessary, the stay should go smoothly as well. So far, several results of the research are already in use to help patients worldwide, and we hope that more applications will follow in the coming years. We can also make a difference on a global scale through our specific focus and smart specializations. We do this not only in the medical applications at issue here, but also in the manufacturing industry, in industry 4.0 applications." 



  • Information about events organized by the Flanders AI Research Program can be found on our LinkedIn page
  • For information about the research on CT images of lungs in COVID-19 patients, click here.
  • If you would like to know more about the research on MS patients during the pandemic, click here.
  • If you would like to know more about the research on single cell technology, click here.
  • For more information about the research on epileptic seizures, click here.
  • For more information about data vaults for personal health data, click here.  




Dr. Sabine Demey is the Program Director of the Flanders AI Research Program. In this function, she coordinates the research activities of ten consortium partners (Flemish universities and research centers) under the ambitious Flanders AI policy plan. She believes it is important for technological developments, such as AI, to have a positive impact on society and the economy. Artificial intelligence was already addressed in various forms in her computer engineering degree program and in her PhD in Robotics in the 90s. Sabine has many years of experience in the industrial field, and she has been involved in various research and business activities such as medical applications of 3D printing and software in the field of Industry 4.0. AI was an important component in many of these activities. 


Health care is a highly diverse application domain, featuring unique properties that require a complementary set of novel AI methods to yield impact. These include dealing with sensitive information (e.g., patient information), complex and diverse data types that should be adequately modelled and combined, and the need for hybrid modelling approaches that combine data-driven approaches with human expertise. Furthermore, models should be interpretable by domain experts to build trust, e.g., when using AI-assisted approaches for clinical decision support.  
To cover all these different challenges, the Flanders AI Research Program selected six use cases or applications. These cases will on the one hand be used to stimulate novel methodological research and developments and will also be used to demonstrate and validate the developed methodology in real-world scenario’s where data is dirty, missing, of low quality or scarce.  
The use cases are grouped into four broad themes: precision medicine, clinical decision support, hospital decision support and the management of health data, and feature both general, technological aspects as well as specific disease themes. 
More insights on the healthcare themes:  

  • The precision medicine theme features two use cases: single-cell technologies and Multiple Sclerosis. These cases feature high dimensional data (e.g., omics technologies), many complementary data types that need to be integrated and combined with human expert knowledge to predict individual patient progression and follow-up, and the combination of sensitive patient data from multiple centers.
  • The clinical decision support theme will tackle both medical image processing with a focus on radiological images (such as analysis of CT images of lungs), and the case study of (tele)monitoring of patients with epilepsy - epileptic seizure detection at home using sensors measuring data from different modalities. Challenges here concern amongst others. learning with flexible supervision, interpretable models, and multi-view learning.  
  • The hospital decision support theme will focus on optimal patient flow and a prediction of the Length-of-Stay to improve the efficiency of hospital resource allocation. This use case will consider many patient characteristics during hospitalization and offers many challenges for AI-assisted data wrangling and optimization of resource planning.
  • The citizen-managed health data theme tackles the challenge of dealing with sensitive personal health information distributed over multiple places and with the citizen in control of the permissions to use the data.