LONDON – Heart disease is now known to be both a cause and an effect of serious COVID-19 infection, with more than 1 in 10 patients who have underlying cardiac conditions being killed by the virus, while others with no previous record of cardiovascular problems are suffering significant COVID-19 induced weakening of their hearts.
The scale of cardiovascular complications and the need to identify patients most at risk has prompted the testing of artificial intelligence (AI) systems, in a bid to speed diagnoses and provide a short cut to understanding the characteristics of cardiac involvement.
“To date, there is no way of linking the impact of the virus to predicted patient outcomes,” said Ross Upton, CEO of Ultromics Ltd., which has launched a 500-patient study in collaboration with the Mayo Clinic, to apply its automated AI-driven echocardiogram analysis technology to forecast outcomes.
The aim of the collaboration between Oxford, U.K.-based Ultromics and the Rochester, Minn.-based Mayo Clinic, is to apply the Ultromics AI software Echogo Core, to identify and map the novel cardiac features of COVID-19.
Echogo Core is able to home in on 80,000 different datapoints in each echocardiogram, pattern matching the results with confirmed diagnoses in Ultromics’ image database.
The multisite Mayo Clinic study will look at 500 COVID-19 patients aged 18 to 89, to train the algorithm. The aim is to triage those patients known to have cardiovascular disease and to identify patients without pre-existing disease who are at risk of developing cardiovascular complications.
“We don’t have a good handle on exactly what is happening to the heart in COVID-19 patients,” said Upton. “Our system is trained on lots of patients, so it can quantify the heart and how well it functions, and investigate what happened,” he told BioWorld.
Ultromics hopes to hone the algorithm to pick up the novel aspects of how the virus infects the heart and to develop a specific diagnostic.
“We are looking to do things in two ways; to create something to triage patients with cardiovascular disease at most risk [and] to see can we pick up patients without pre-existing heart disease, and can we pick them up early enough before they fall off a cliff – some patients deteriorate in a matter of hours,” Upton said.
COVID-19 induced microvascular disease and myocarditis have been observed in patients with no history of heart disease. Others have suffered therapy-associated cardiotoxicity as a side effect.
One suggested mechanism underlying these complications is the binding of the spike protein by which the COVID-19 virus enters host cells, to angiotensin converting enzyme 2 receptors expressed in the respiratory tract and myocardium.
“It’s thought that’s a direct pathway where COVID-19 causes cardiovascular disease in patients not previously diagnosed,” said Upton. “It’s not proven, but [the research] could help find out if this is how COVID-19 affects the heart in people without pre-existing conditions.”
Another aspect of COVID-19 pathology on which Echogo Core may be able to shed some light, is why men are more than twice as likely as women to die as a result of COVID-19 infection. “That’s a really interesting one,” Upton said.
In general, men have higher prevalence of cardiovascular disease, but epidemiological research carried out on more than 6,000 people who died from COVID-19 infections in the U.K. has shown factors including heart disease, age, smoking and other comorbidities do not account for the difference in mortality between the sexes
“Understanding why [this is the case], is part of the reason why we need to look further into cardiovascular disease in COVID-19,” Upton said.
Given the ongoing level of infection, Upton does not anticipate any problem in recruiting 500 patients at Mayo Clinic. The company also is looking to extend the analysis to centers in the U.K.
Simon Rasalingham, CEO of Behold.ai, also thinks using AI-based diagnostics could make a significant difference in managing the pandemic. The London-based company has announced it is partnering with Apollo Hospitals Group in India to use its instant triage Red Dot algorithm to analyze chest X-rays of COVID-19 patients, after showing it can automatically detect abnormal chest X-rays with 90% accuracy.
Behold.ai says the partnership with Apollo, which has a network of 71 hospitals, will be the world’s largest deployment of a radiology-based diagnostic for COVID-19. To date, the company has approval for use of Red Dot in two indications, having secured FDA 510(k) clearance in the automatic diagnosis of pneumothorax (collapsed lung) in March, and CE marking for detecting lung cancer in April.
When COVID-19 emerged, Behold.ai tested Red Dot on the chest X-rays of patients being treated for COVID-19 in hospitals in the U.K., showing it could pick up abnormalities in 30 seconds. The company subsequently signed an agreement with Wellbeing Software, which runs radiology information systems at over 700 U.K. hospitals, to combine their technologies and speed up analysis of the significant number of X-rays, which are being used as a key diagnostic test for triaging COVID-19 patients.
“Our technology has been proven to rapidly diagnose chest X-rays as normal or abnormal accurately,” said Rasalingham. There is potential “for faster diagnosis of individuals with COVID-19 and the better management of resources in healthcare systems already stretched by the increase in hospital admissions” he said.