Researchers at Brigham and Women’s Hospital in collaboration with Keio University in Japan have unveiled the potential of an AI model to revolutionize the detection of a heart condition known as atrial septal defect through analyzing electrocardiogram readouts, presenting a major leap towards enhanced early screening methods for heart health.
An atrial septal defect (ASD) refers to the formation of a hole between the upper chambers of the heart—known as atria. The condition can cause heart failure and is often overlooked due to a lack of symptoms before the appearance of fatal complications. Reporting in eClinicalMedicine, researchers have now developed a deep learning AI model able to efficiently detect ASD in electrocardiograms (ECGs).
“If we can deploy our model on a population-level ECG screening, we would be able to pick up many more of these patients before they have irreversible damage,” said Shinichi Goto, MD, PhD, instructor in the Division of Cardiovascular Medicine at Brigham and Women’s Hospital and co-author on the paper in a press statement.
The research team harnessed the power of deep learning technology by training the AI model on ECG data sourced from an extensive pool of over 80,000 patients aged 18 and above. These patients had all undergone both ECG and echocardiogram tests as part of the effort to identify ASD. Within this diverse dataset, 857 individuals were diagnosed with ASD, highlighting the prevalence of the condition within the examined population.
The data collection spanned across three medical facilities, including the prominent educational institutions BWH and Keio University, as well as the community-oriented Dokkyo Medical University Saitama Medical Center. The AI model’s subsequent evaluation at Dokkyo demonstrated its consistent efficacy, even within a more general patient population not specifically targeted for ASD screening.
Remarkably, the AI model exhibited greater sensitivity compared to conventional diagnostic methods, specifically when using recognized ECG abnormalities as indicators. The model showcased an impressive 93.7% accuracy in identifying cases of ASD, surpassing the 80.6% accuracy achieved through the use of known abnormalities.
“It picked up much more than what an expert does using known abnormalities to identify cases of ASD. The model’s performance was retained even in the community hospital’s general population, which suggests that the model generalizes well,” Goto said.
While acknowledging certain limitations, such as the AI model’s training on samples from academic institutions with a focus on rare diseases, the study’s results ignite optimism for its potential application in real-world health care settings. By harnessing AI’s capabilities, researchers and medical professionals move closer to transforming how heart conditions are detected and managed, potentially saving lives through earlier interventions.
“Perhaps this screening could be integrated into an annual PCP appointment or used to screen ECGs taken for other reasons,” Goto concluded in a press statement.