Using an AI-based method, researchers at Cardiff University in Cardiff, Wales, developed a clinical prediction model, according to a study published in PLOS One. The model showed it can provide an accurate and reliable prognosis for patients with cardiovascular disease when compared to traditional methods.
"If we can refine these methods, they will allow us to determine much earlier those people who require preventative measures,” Craig Currie, PhD, study co-author and professor at Cardiff University's School of Medicine, said in a statement.
A clinical prediction model uses a variety of characteristics to predict a diagnostic or prognostic outcome. Typically, clinicians and statisticians use manually tuned Cox regression models to estimate the risk of various diseases.
For the study, researchers used genetic programming (GP) to develop the clinical prediction model and assess a patient’s future risks of a cardiovascular event, like cardiovascular death, non-fatal stroke or non-fatal myocardial infarctions. GP is described as an AI-based method where computer programs are encoded into a set of genes that are then modified using an evolutionary algorithm.
The goal of the study was to determine the utility of GP for the automatic development of clinical prediction models and compare its predictability performance against traditional methods. Researchers used a dataset of more than 3,800 cardiovascular patients for the study.
The study showed that both models' predictability performance were comparable, with the GP model demonstrating “considerable potential as a method for the automated development of clinical prediction models for diagnostic and prognostic purposes,” Christian A. Bannister, PhD, a research associate at the Cardiff University School of Medicine, et al. wrote.
“The ability to interpret solutions offered by machine learning has so far held the technology back in terms of integration into clinical practice,” Irena Spacic, study co-author and professor at Cardiff University’s School of Computer Science and Informatics, said in a statement.
“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.”