In a Hot Line session yesterday, Professor Geoffrey Strange (University of Notre Dame - Sydney, Australia) presented results from the AI-ENHANCED AS study, which investigated whether an artificial intelligence (AI) algorithm, developed from routine echocardiographic parameters, could identify moderate-to-severe and severe aortic stenosis (AS) phenotypes associated with an increased risk of mortality.
The proprietary AI-Decision Support Algorithm (AI-DSA) was trained using data from the National Echo Database of Australia (NEDA), which contains more than 1,000,000 echocardiograms from over 630,000 patients and is linked to mortality information. The algorithm was also trained to ensure all guideline-defined severe AS was detected. Training was performed using 70% of the NEDA data, which were randomly selected. Using the remaining 30% of NEDA data, 5-year death rates in patients with the moderate-to-severe and severe AS phenotypes were compared with 5-year death rates in the reference group, patients without significant risk of severe AS.
Among 179,054 individuals, 2.5% were identified by the AI-DSA as having severe AS and 77.2% of these individuals met the guideline criteria for severe AS. The AI-DSA also identified 1.4% of individuals as having moderate-to-severe AS.
The 5-year mortality rates were 56.2% for moderate-to-severe AS, 67.9% for severe AS and 22.9% for the reference group. Compared with the reference group, the age- and sex- adjusted odds ratios for all-cause mortality were 1.82 (95% CI 1.63 to 2.02) for moderate-to-severe AS and 2.80 (95% CI 2.57 to 3.06) for severe AS. Within the severe AS phenotype identified by the AI-DSA, the 77% who met current guidelines had a 5-year mortality rate of 69.1%. The additional population identified by the AI-DSA with a severe phenotype, but who do not meet current guidelines, had a mortality rate of 64.4%.
Thus, in addition to all patients within current guidelines, the AI algorithm picked up patients with a high risk of dying that may be missed by conventional definitions, according to Prof. Strange. He comments, “The findings suggest that the AI algorithm could be used in clinical practice to alert physicians to patients who should undergo further investigations to determine if they qualify for aortic valve replacement. Given the rising prevalence of AS and its impact on mortality, it is time to revisit the practice of watchful waiting and consider more proactive attempts to identify those at risk. More research is needed to determine if aortic valve replacement improves survival and quality of life in patients identified by the AI-DSA as having a high risk of mortality, but who do not meet current guideline definitions.”