News | 14 August 2019

Self-learning model to predict risk of coronary heart disease

Millions in funding for personalised approach based on big data

Researchers at Maastricht UMC+ and Maastricht University are receiving nearly two million euros in funding from the Netherlands Organisation for Scientific Research (NWO) to develop a self-learning eHealth app to prevent coronary heart disease. The digital tool will use vast amounts of data, for example from hospitals, general practitioners and Statistics Netherlands (CBS), to compile a personal risk profile and seek out potential interventions.

Blockage due to atherosclerosis (iStock)Blockage due to atherosclerosis (iStock)Coronary heart disease occurs when the blood supply to the heart is impaired. In most cases, the cause is a narrowing of the coronary artery (atherosclerosis). Smoking, nutrition, exercise and other lifestyle factors play an important role in this context, and that means that the risk of coronary heart disease can be reduced by adopting healthier habits. But making lifestyle changes is easier said than done. The researchers in Maastricht are therefore looking to develop a model that can be personalised for individual users, although they are basing their work on "big data".

Big data
Hospitals and general practitioners have a wealth of medical data at their disposal. Similarly, Statistics Netherlands collects socio-economic data. Combining data on patients who have a history of coronary heart disease with data on population lifestyle trends and an individual user's physical environment, for example, makes it possible to estimate the risk of developing coronary heart disease. This information will be aggregated into a predictive model that can help to accurately identify those at risk, based on their individual traits. The new data obtained in this manner will then be used to refine the eHealth app.

In addition to risk assessment, the researchers want to go a step further. "The ultimate goal is to minimise the risk of and prevent coronary heart disease," says Dr André Dekker, professor of Clinical Data Science, "but to do that in small, manageable steps. Expecting people to change an ingrained lifestyle in one go is unrealistic because it's so difficult." That is why the researchers will use their predictive model to pinpoint where the greatest gains can be made at the outset. "And that's where we'll implement the lifestyle intervention. It will also encourage people to make real improvements in their health, with the support of the eCoach that we plan to develop."

The research project is entitled CARRIER and has been undertaken in partnership with Maastro, Statistics Netherlands, general practitioners in Eastern South Limburg and Sananet.