Is artificial intelligence applied to the diagnosis and treatment of cardiovascular diseases, leading or assisting?

 

With the advent of the medical big data era, artificial intelligence has been widely used in many aspects of medicine. The ESC 2021 conference "The latest science in the field of artificial intelligence and digital health" has more related research announcements. This journal specially invites Professor John Cleland from the University of Glasgow, the chairman of this session, to talk about his understanding of artificial intelligence technology.

International Circulation: With the advent of the medical big data era, artificial intelligence has been widely used in many aspects of medicine, providing new methods for precise diagnosis and treatment. Could you please share us what are the more mature technologies currently used in disease diagnosis and treatment?

Professor John Cleland: Probably the most mature use of artificial intelligence applied to clinical medicine is in the field of imaging. I am a cardiologist, so particularly cardiac imaging is of interest, which includes echocardiography and cardiac MRI. The intention here is to use artificial intelligence to access cardiac volumes, cardiac function, and ultimately act as a diagnostic aid for the clinician. Ultimately, the clinician takes the responsibility for making the diagnosis, but in terms of accurate measurement of cardiac volumes, the velocity signals and so on from echocardiograms and cardiac MRI, it may be that artificial intelligence is able to do this much faster, more reproducibly, and possibly more accurately than a human looking at the images. So I see it as a diagnostic assistant. Soon I think we will see artificial intelligence applied to the ECG. I think it is a little bit more difficult there because ECG is a much more mature technology – one hundred years old or more. There is a huge amount of human intelligence in ECG. But going back to the digital ECG, it is possible we are going to see new applications where, rather than the ECG being used for diagnosis, it is used to predict what will happen in the future, which is not what we do with an ECG today. So there would be a change in use of the ECG.

International Circulation: At present, which aspects of the field of cardiovascular medicine are mainly used in artificial intelligence technology?

Professor John Cleland: I summarized that above. I think it is mainly in the cardiac imaging area, and in the future, the ECG. When it comes to identifying treatment patterns and clinical diagnosis, we are a long way away from finding a good use for artificial intelligence. Ultimately, we have to have the courage to not treat a patient on the basis of the information supplied by the artificial intelligence. To give an example, I have a patient with high cholesterol, who the guidelines say should have a statin, but the artificial intelligence says the patient is at very low risk for events, not by conventional risk factors but by the artificial intelligence program. Am I really not going to treat that patient with statins because the artificial intelligence told me not to? I think we are decades, maybe more than decades away from being able to use artificial intelligence in that way. Ultimately, artificial intelligence can be used to help doctors make decisions in terms of treatment, but it doesn’t make decisions for them.

International Circulation: In the special session of "Late Breaking Science in Artificial Intelligence & Digital Health" of this conference, Professor Yutao Guo from China introduced the “Upstream AF risk control with digital tool” ,What is your opinion about this study?

Professor John Cleland: I thought it was a very intelligent study. I think Professor Guo had thought very carefully about how the information from the Smartwatch could be developed. Importantly, as I was saying earlier, with ECG the most important use of artificial intelligence may not be in diagnosis but in predicting future events. And in this case, Professor Guo was trying to predict the probability of developing atrial fibrillation, and then to advise a change either in lifestyle (maybe taking more exercise, losing weight, altering diet) or taking medication prophylactically to prevent the possibility of atrial fibrillation. The Smartwatch could also detect the onset of atrial fibrillation using a technique called PPG (photoplethysmography). This was the way the Smartwatch could continuously monitor the patient for atrial fibrillation. But finally, and very interestingly, we know that PPG may have some inaccuracies, so the ability to also be able to do a confirmatory single lead ECG, which would be the trigger for treatment, was very important. The fact that the watch was not trying to do one thing, but doing three quite separate tasks, makes it a very interesting intervention.

International Circulation: In the future, what is the application status of new artificial intelligence tools in disease diagnosis and treatment? Dominant or as an auxiliary tool for doctors?

Professor John Cleland: Almost certainly, it is going to be an auxiliary tool for quite some time. I think we should think about man and machine working together so human intelligence and artificial intelligence are complementary. I think the relationship will start out with the human being 90% and the artificial intelligence being 10%. But over time, probably decades, that relationship will change and a new balance will be found. Maybe in a hundred years time, it will be 10% human and 90% artificial intelligence. Maybe there is just that possibility that the artificial intelligence will take over completely, but I think that will take a long time.