Study Shows AI Could Offer Accurate and Reliable Prognosis for Cardiovascular Patients

Adopting artificial intelligence in the diagnosis and prognosis of disease could help to prolong the lifespan of people and at the same time offers considerable savings for the NHS.

This is as stated by the scientists from Cardiff University who have offered convincing evidence proving the advantages offered by modern methods to risk assessments in patients.

In the latest research published in PLOS One, the research team has illustrated how artificial intelligence can offer an equally reliable and accurate prognosis for patients with cardiovascular disease when compared to conventional techniques.

Since there was no need for human interaction or expertise for the machine-learning methods they used, a major difficulty in the process was solved.

If we can refine these methods, they will allow us to determine much earlier those people who require preventative measures. This will extend people’s lives and conserve NHS resources.

Professor Craig Currie, School of Medicine, Cardiff University

Professor Craig Currie is also the co-author of the study.

In the age of evidence-based medicine, the use of statistics has become an integral part of predicting the risks of some types of disease.

Conventionally, statisticians and clinicians have dealt with this task by manually formulating mathematical equations. However, artificial intelligence offers methods that can reveal complex associations in the data.

Although we already have reliable methods of forecasting people according to their degree of risk of serious heart events, artificial intelligence promises new ways of interrogating data and the likelihood of more reliable classification of risk.

Professor Craig Currie, School of Medicine, Cardiff University

In their study, the team trialed a method called genetic programming (GP), which is a technique motivated by the evolution in nature through which computer programs are encoded as a set of genes that are then iteratively evolved or modified.

GP is beneficial over algorithms developed by humans in the sense that it minimizes bias and the chance of human error, and simultaneously allows for any variations in the environment to be automatically incorporated into mathematical formulas.

One benefit of this particular technique is that it is possible to make the complex associations revealed by artificial intelligence from the data transparent to the clinicians, that is, they don’t have to deviate from their current practice.

In the research, the scientists used GP to evaluate the future risks of a cardiovascular event, such as non-fatal stroke, non-fatal myocardial infarction, or cardiovascular death, in more than 3800 cardiovascular patients, aged between 19 and 83, over a period of 10 years.

A total of 25 predictors obtained from patient data, including sex, age, blood pressure, BMI, alcohol, and smoking use, were used by machine-learning algorithms.

The results demonstrated that the machine-learning algorithms could perform similarly to conventional techniques when estimating the risk associated with individual patients.

The ability to interpret solutions offered by machine learning has so far held the technology back in terms of integration into clinical practice. However, in light of the recent resurgence of neural networks, it is important not to side line other machine learning methods, especially those that offer transparency such as genetic programming or decision trees. After all, we are looking to use artificial intelligence to aid human experts and not to take them out of the equation altogether

Professor Irena Spasić, School of Computer Science and Informatics, Cardiff University

Professor Irena Spasić is also the co-author of the study.