Artificial intelligence (AI) tools could help doctors to detect transthyretin amyloid cardiomyopathy (ATTR-CM) earlier and more fairly across different racial groups, according to a recent study published in the journal JACC: Advances.
ATTR-CM is often diagnosed late — especially in Black patients, who are also more likely to develop the disease than people from other racial backgrounds. These delays can lead to worse health outcomes and missed chances for early treatment.
To see how well different predictive models could spot the disease and whether they worked fairly, researchers looked at data from the electronic health records of more than 3,300 patients, including 176 patients with confirmed ATTR-CM. They compared four tools: a model based on medical insurance claims (called the “Huda model”), a scoring system from the Mayo Clinic and two AI tools that analyze heart ultrasound images (called the “Echo models”). The team checked how accurately each tool diagnosed ATTR-CM and also looked at whether the tools performed differently or unfairly across racial groups.
None of the tools showed clear racial bias. They were all just as likely to miss the condition in Black patients as in patients from other racial groups.
When it came to accuracy, the medical claims-based model performed the worst, while the two AI-based Echo models were the most accurate.
Read more about ATTR-CM testing and diagnosis
“Machine learning techniques have shown promise when applied to imaging and electronic health record data to enhance the early detection or correct the misdiagnosis of diseases, such as cardiac amyloidosis,” the researchers said. “The use of such AI-assisted diagnostic tools in ATTR-CM could improve racial and ethnic disparities through the early identification and uptake of novel therapies in these patients.”
Still, they cautioned that for AI models to make a real difference in patient care, more high-quality research, agreement between experts and other practical steps are needed.
Sign up here to get the latest news, perspectives, and information about ATTR-CM sent directly to your inbox. Registration is free and only takes a minute.
