Home Support system AI decision support tool accurately identifies aortic stenosis with low chance of survival

AI decision support tool accurately identifies aortic stenosis with low chance of survival

0

August 28, 2022

2 minute read

Source:

Strange GA, et al. Hot Line 6. Presented at: Congress of the European Society of Cardiology; August 26-29, 2022; Barcelona, ​​Spain (hybrid meeting).

Disclosures:
The National Echo Database Australia (NEDA) and Echo IQ Ltd. funded ENHANCED-AI. NEDA has received research grants from Echo IQ, Edwards, Novartis and Pfizer. Strange reports receiving consulting fees or royalties or owning stock in Echo IQ, Edwards and Medtronic.


We have not been able to process your request. Please try again later. If you continue to have this problem, please contact [email protected]

In a large real-world cohort undergoing routine echocardiography, an artificial intelligence decision support algorithm identified patients with moderate to severe forms of aortic stenosis associated with poor survival in the absence of treatment.

Surgical and transcatheter aortic valve replacement is associated with reduced mortality in all subgroups of aortic stenosis, but data show that less than half of patients with aortic stenosis with an indication or potential indication of valve replacement are treated, Geoffrey A. Strange, PhD, FCSANZ, said at a press conference at the Congress of the European Society of Cardiology. An AI-based decision support system could better identify patients at risk of death, he said.

Heart Matrix_Adobe Stock
Source: Adobe Stock

“The decision support algorithm correctly identified patients at significant mortality risk, so among those with moderate disease – 1.4% of the population – when validated against a large database, we were able to show 56% mortality over 5 years,” Strange said. “In the severe group, we were able to show 67% mortality at 5 years. Within this severe group, there were two cohorts: one that met current guidelines for severe aortic stenosis, and another that fell just outside the guidelines, but the AI ​​detected a similar risk profile within this population.

Formation of a new model

For the ENHANCED-AI study, Strange and colleagues analyzed data from 1,077,145 surveys of 631,824 adults at 23 centers in Australia from 1985 to 2019, with an average follow-up of 7.2 years (mean age, 61 years old). The researchers included 70% of the population in an AI training set and 30% of the cohort in a validation group to assess the performance of the algorithm. All data has been linked to the National Death Index of Australia.

“We had a six-step process to train the algorithm,” Strange said. “The raw data from the million studies, an imputation model to support the discrete nature of echocardiography, the neural network, a Gaussian distribution of the result, an error calculation of this result, and backpropagation.”

The 5-year mortality rates were 22.9% in the low-probability group, 56.2% in those with moderate to severe aortic stenosis, and 67.9% in those with severe aortic stenosis.

Compared to the low probability group, people with moderate to severe aortic stenosis were almost twice as likely to die within 5 years (OR=1.82; 95% CI, 1.63-2.02; P .001). People with severe aortic stenosis were almost three times more likely to die within 5 years compared to the low probability group (OR=2.8; 95% CI, 2.25-3.06; P .001).

“We were able to demonstrate that the algorithm’s ability to place different patients into different risk ‘compartments’ was also associated with long-term survival outcomes,” Strange said.

In analyzes that stratified the two severe aortic stenosis groups, patients who met guidelines for severe aortic stenosis (n=2081) were only 26% more likely to die within 5 years compared to patients identified as high risk by decision support tool (n=711 OR=1.26 95% CI: 1.04-1.53; P= .021).

Predict risk in the workflow

During a Q&A session after the press conference, Strange said the decision support algorithm can be deployed directly into an echocardiography lab database.

“What we plan to do in the future is make sure there’s an alert system in the echo reports themselves,” Strange said. “This algorithm would be deployed either in retrospective time after a week or a month, depending on the establishment’s workflow, but also in real time to signal an alert to the declaring doctor.”