Researchers at the Federal University of Minas Gerais in Brazil trained a deep neural network on ECGs from more than 1.5 million patients to estimate a different number: the age the heart's own electrical signal looks like, which does not always match the calendar.
People whose ECG-estimated age came in more than 8 years above their real age had a 79 percent higher death rate during follow-up. People whose ECG age came in more than 8 years below their real age had a 22 percent lower death rate, even when their actual exam looked normal to a cardiologist.
The pattern held up in two outside groups of Brazilian patients, more than 15,000 people combined, which let the team rule out a fluke of one hospital's equipment. An ECG that looks unremarkable to a trained eye can still carry a mismatch between the heart's apparent age and the person's real one, and that mismatch alone predicts who is more likely to die soon.
Read the original study
Lima et al., Nature Communications, 2021 · doi.org/10.1038/s41467-021-25351-7
The heart keeps its own calendar.
Source
- Lima, E. M., Ribeiro, A. H., Paixão, G. M. M., Ribeiro, M. H., Pinto-Filho, M. M., Gomes, P. R., Oliveira, D. M., Sabino, E. C., Duncan, B. B., Giatti, L., Barreto, S. M., Meira, W., Schön, T. B., & Ribeiro, A. L. P. (2021). Deep neural network-estimated electrocardiographic age as a mortality predictor. Nature Communications, 12(1), 5117. https://doi.org/10.1038/s41467-021-25351-7