‘Digital heart’ to help doctors make a diagnosis

‘Digital heart’ to help doctors make a diagnosis

Better care thanks to computer models and artificial intelligence

18-01-2022 · Science

‘Smart’ devices and digital aids: it is hard to imagine daily life without them. They are also increasingly being used in the medical world. A Maastricht computer model that simulates the heart should – in the future – be able to help cardiologists make a diagnosis. But not every doctor is jumping for joy about it.

Two generations of Maastricht researchers worked on it for more than 25 years: an accurate computer model of the heart. Not an easy task, as the organ constantly changes shape. A combination of physical formulas and physiological knowledge is used to imitate the complex process of contracting and relaxing of the heart muscle as well as the pumping of blood.

For years now, the model has been used for fundamental research into the way the heart operates. This can be used, for example, to replace animal testing, says biomedical engineer Joost Lumens, who is leading the project. In addition, it is used widely in education. Medical students – in Maastricht, but also more and more at other universities worldwide – use it to get a better understanding of the heart. Since the model has been publicly available on Internet, it was downloaded more than twenty thousand times.

Mix of data

“The simulator shows all kinds of information about a beating, virtual heart,” says Lumens. “The students see data that can also be measured in practice in patients. They can introduce a defect, for example a leaking heart valve or scar tissue caused by a heart attack, in the virtual heart, to study how that influences the all-over functioning, and how they can see this in the diagnostic data. Every defect leaves its own fingerprint in this data. With the model, students are trained to recognise these. Later, as cardiologists, they have to do this to make a diagnosis based on measurements in patients.”

The latter is not always easy in practice, however. A minor heart defect can change the behaviour of the entire system, and patients often have several disorders at the same time. Lumens: “So doctors have to determine various defects from the mix of diagnostic data. This is both complex and time consuming.”

Artificial intelligence

Can artificial intelligence (AI) not help here? This is researched in the MARCIUS project – which started in 2020 and is a collaborative project of the UM, the KU Leuven and the Norwegian branch of technology concern General Electric. “We develop software that independently recognises defects from data measured in patients,” says Maastricht PhD candidate Claudia Manetti, involved in MARCIUS.

However, a large, varied collection of diagnostic data is needed for AI to practice on. “Until now, only information from actual patients was used for this, but in this way it will take too long before you collect the necessary amount and diversity of information,” says Manetti. “That is where our computer model comes into play: with it, we can create a tremendous amount of data from virtual hearts with all kinds of combinations of disorders.” With this, the Norwegian and Belgian colleagues subsequently train the AI in recognising defects.

Should the trained software indeed prove to work as expected, it could serve as an aid to make a cardiologist’s job quicker and easier. “Although it will never replace a doctor,” Lumens emphasises. “Future cardiologists will still be trained in identification. Except that now they can compare their own assessment to that of the computer.” Moreover, AI can detect disorders that a doctor may miss when examining the measurements by eye, Manetti adds. “Certainly when you have been staring at diagnostic data for hours on end, it is helpful. Also, it is easier for AI to look ‘through’ the noise of the measurements.”

Sceptical doctors

Lumens sees more applications for the model. For example, in collaboration with MUMC+ cardiologists and business developers at the Brightlands Campus, work is being carried out on technology that can estimate the blood pressure within the heart – the so-called hyperaemia, without having to carry out an expensive and invasive procedure with a catheter, as is now the case. They are also trying to predict whether individual patients would benefit from a pacemaker, by first testing the device virtually on a patient heart’s ‘digital twin’.

Still, not every doctor is enthusiastic about this ‘digital assistance’. In the medical world in which developments are taking place at record speed, Lumens regularly comes up against scepticism. “Some people set unrealistically high demands for technology; I think it is because they don’t feel any empathy with it. They accept a mistake made by a colleague more easily than a mistake made by a computer.”

So how do you build the necessary trust? According to Lumens, objective measurements of the achievements and uncertainties of the technology is the only way to do so. He thinks that the fact that FHML has a new chair (Lumens has been professor of Computational Cardiology since the end of last year), dedicated to this field of research, is a good sign. “This will help it make headway within the UM. The support and constructive attitude of cardiologists at MUMC+ and researchers at CARIM are crucial for this kind of innovation.”
 

Image: Pixabay

Categories: news_top, Science
Tags: artificial intelligence,ai,cardiology,heart,pacemaker,computermodel,code,circadapt,marcius,carim,fhml,mumc+,biomedical engineering,instagram

Add Response

Click here for our privacy statement.

Since January 2022, Observant only publishes comments of people whose name is known to the editors.