Artificial intelligence (AI) was able to predict which patients were likely to develop an irregular heart rhythm, even when physicians interpreted results as normal, and identified patients at increased risk of dying of any cause within 1 year, in this study presented November 16 at the American Heart Association's Scientific Sessions 2019 in Philadelphia.
Researchers used more than 2 million electrocardiogram (ECG) results from more than 30 years of records in the Geisinger Health System to compare AI models that either directly analyzed the raw ECG signals or relied on standard ECG features typically recorded by a cardiologist and commonly diagnosed disease patterns.
The AI model that directly analyzed ECG signals was superior for predicting 1-year risk of death, even in patients deemed to have a normal ECG by a physician.
The researchers also found that within the top 1% of high-risk patients, as predicted by the AI model, one out of three patients were diagnosed with atrial fibrillation within a year.
The findings could completely alter the way ECGs are interpreted in the future, the researchers say.