CARRIER Projectupdate 2023
Coronary Artery Disease: Risk Estimations and Interventions for Prevention and Early Detection
– a Personal Health Train Project
The CARRIER team is developing personalized lifestyle interventions to reduce the risk of developing cardiovascular diseases (primary prevention) or to prevent them from worsening (secondary prevention). They aim to tailor lifestyle interventions to the preferences of patients, as personalized advice has a greater chance of effectively changing behavior. Adherence to the lifestyle intervention will be promoted and monitored through a digital multimedia lifestyle coach (eCoach), as shown in the image.
We interviewed Prof. Dr. Ir. André Dekker. André is originally a clinical physicist. Through radiotherapy, he subsequently ended up at Maastro Clinic, among other roles as Head of IT. There, he recognized the importance of IT for the healthcare sector and initiated a research line. André is now a Professor of Clinical Data Sciences. The Clinical Data Science group (40+ members) includes multiple entities such as Maastricht University, Maastricht UMC+, and Maastro Clinic. They work in oncology, physiotherapy, cardiology, diabetes, gastrointestinal diseases, respiratory diseases, rheumatology, and Alzheimer's.
AI for Sustainable Lifestyle Change
In CARRIER, two models are being developed: a prognostic model aimed at identifying high-risk patients and a predictive model aimed at predicting the effect of a specific lifestyle intervention. What is your risk of cardiovascular disease, and what is the most effective individual action to reduce this risk? Citizens with an increased risk and patients can gain more control over preventing coronary heart diseases and other lifestyle-adjustable diseases.
Personalized Lifestyle Advice: Nudging
“In CARRIER, we adopt a holistic human model, considering humans as a whole, where physical, psychological, social, and spiritual aspects are interconnected and mutually influential. Our hypothesis is that by tailoring lifestyle advice - making it clear to people what a very specific intervention would mean for their health - people are more likely to sustain lifestyle changes.”
Data Technology
Developing AI requires a substantial amount of data. The predictive model in CARRIER utilizes data from the CBS (Central Bureau for Statistics), general practitioners, and hospitals. Combining data from these three sources and making this infrastructure FAIR (Findable, Accessible, Interoperable, and Reusable) is complex, André explains. “In the project, we are addressing the legal and ethical aspects associated with unlocking and using this data. This must be done responsibly and securely. We use the Personal Health Train (PHT), which is a federated learning framework. The essence of the PHT approach is that the research question travels to the 'stations' of the data source instead of transporting data from different sources to the research question. Sensitive data remains where it is.” The Personal Health Train aims to allow healthcare innovators and researchers to work with health data from various sources. It can provide controlled access to data while ensuring privacy protection and optimal involvement of individual patients and citizens.
Partners
In addition to Maastro, Maastricht UMC and Maastricht University are the academic partners. Other partners include the CBS, the eScience center, Huis van de Zorg (advisor in citizen participation), Huisartsen OZL (a group of 116 general practitioners in the eastern South Limburg region), and Sananet zorg (responsible for eCoach development). Sananet focuses on developing eHealth solutions, particularly telemonitoring. These online guidance programs provide healthcare professionals with real-time insight into the well-being of their patients.
Challenges
The development of the technology itself is not the biggest challenge in the project, André explains. “The two main challenges are correctly organizing the project legally and the challenging market model in healthcare. Setting up the legal Christmas tree takes the longest. You have data from general practitioners, hospital data, and CBS data. How do you ensure that all parties - taking into account the appropriate legislation - eventually sign off? Additionally, investing in the development of an eCoach for a company like Sananet is not immediately profitable. It is not a direct cost-saving but an investment in health, and who pays for prevention then? The current way of financing healthcare is less aligned with healthcare innovations focusing on prevention, on the future. We should invest much more in keeping healthy people healthy.”
Future
“With CARRIER, we have taken significant steps in developing complex models involving multiple parties, which we had doubts about beforehand. We have also open-sourced all this infrastructure, which I am quite proud of! We have received additional provincial subsidies to delve deeper into how to have meaningful conversations and make AI understandable for the citizens we want to reach. And discussing the place of AI in our society with each other, this is very educational for us as data professionals.”
