AI can help predict adverse CV events

Using artificial intelligence (AI) to quantify blood flow in the heart can predict adverse cardiovascular (CV) outcomes over and above other traditional CV risk factors, a new study has shown.

“We have tried to measure blood flow manually before, but it is tedious and time-consuming, taking doctors away from where they are needed most — with their patients,” said principal investigator Professor James Moon of UCL* Institute of Cardiovascular Science, London, UK. “AI is moving out of the computer labs and into the real world of healthcare, carrying out some tasks better than doctors could do alone.”

Using AI to quantify blood flow imaging data from CMR** — known as perfusion mapping — allows automatic quantitation to be done at scale, and thus, feasible for application in routine clinical practice, Moon highlighted.

Measures of blood flow in the heart, MBF and MPR*** obtained by CMR perfusion mapping, were independently associated with risk of death and MACE^#. [Circulation 2020;doi:10.1161/CIRCULATIONAHA.119.044666]

For every 1mL/g/min decrease in stress MBF, the risk almost doubled for death (adjusted hazard ratio [HR], 1.93; p=0.028) and MACE (HR, 2.14; p<0.0001), after adjustments for age and comorbidity.

Similarly for MPR, each unit of decrease was significantly associated with an increased risk of death (HR, 2.45; p=0.001) and MACE (HR, 1.74; p<0.0001).

“Quantitative myocardial blood flow provides incremental prognostic information in patients with suspected coronary artery disease [CAD] above traditional CV risk factors,” said Moon and co-authors. “Impaired global perfusion may be a targetable CV risk factor.”

In the two-centre study, 1,049 patients (mean age 60.9 years, 67 percent male) with suspected or known CAD were assessed on myocardial perfusion using CMR. Imaging data were analysed automatically using AI. After a median follow-up of 605 days, deaths and MACE were reported in 4 percent and 16.6 percent of patients, respectively. 

“Even in patients without regional perfusion defects, absolute perfusion is prognostic,” the researchers noted.

Among the 783 patients who showed no regional perfusion defects and no known macrovascular CAD, each unit decrease in MPR remained independently associated with MACE (HR, 1.65; p=0.008) and death (HR, 2.22; p=0.015), consistent with the overall cohort.

“As there is no user input and no ionising radiation, early disease and microvascular disease can be studied at scale … [Also,] the spatial resolution is superior to other functional imaging modalities,” Moon and co-authors pointed out on the advantage of using CMR imaging. 

“The predictive power and reliability of the AI was impressive and easy to implement within a patient's routine care. The calculations were happening as the patients were being scanned, and the results were immediately delivered to doctors,” said lead author Dr Kristopher Knott, also from UCL.

“As poor blood flow is treatable, these better predictions ultimately lead to better patient care, as well as giving us new insights into how the heart works,” he added.