Self-learning model to predict risk of coronary heart disease

Million euro grant for an individualised approach based on big data

Researchers from Maastricht University and Maastricht UMC+ are receiving a grant of nearly two million euros from NWO for the development of a self-learning eHealth application that aims to prevent coronary heart disease. The development of the digital application, led by Digital Society professor Andre Dekker, uses massive amounts of data from hospitals, general practitioners and the Central Bureau of Statistics (CBS), among others. Based on that ‘big data’, a personal risk profile can be formulated and starting points for intervention can explored.

Coronary heart disease is caused when blood flow to the heart is restricted. In most cases, this is due to a narrowing of the coronary artery. Smoking, diet, exercise and other lifestyle factors play an important role in this. That means that the risk of coronary heart disease can be reduced by stimulating a healthier lifestyle. However, that is easier said than done. The Maastricht researchers therefore want to move towards a personalised model. However, big data remains the starting point.

Big data

Hospitals and general practitioners have a wealth of medical information. Statistics Netherlands also has data of a socio-economic nature. For example, data from patients with a history of coronary heart disease can be combined with data on lifestyle trends among the population and the physical environment of a person. Based on these combined data, risk assessments can be made for the development of coronary heart disease. That knowledge is bundled in a predictive model. Using this as an aid, it is possible to accurately estimate who is at risk on the basis of his or her individual characteristics. This data will again be used for the refinement of the eHealth application.

eCoach 

In addition to the risk assessment, the researchers want to take it a step further. “Ultimately, the goal is to minimise and prevent the risk of coronary heart disease”, says Prof. André Dekker, professor of Clinical Data Science. “But in small, feasible steps. Adopting an integrated lifestyle in one go is very difficult for people and is not realistic.” Because of this, the researchers use their predictive model to first look at where the most benefits can be achieved. “And that’s where we apply the lifestyle interventions. This also encourages people to actually improve their health. The eCoach that we are going to develop will support this.”

The research project is called CARRIER and is being carried out in collaboration with MAASTRO Clinic, Statistics Netherlands, huisartsen Oostelijk Zuid-Limburg and Sananet.

 

Bron: Maastricht University 
Foto: Stefan Oostwegel Fotografie