An AI model that predicts biological age by analyzing electrocardiogram (ECG) data has uncovered a significant link between ECG-derived age and cognitive functioning.
According to preliminary findings set to be discussed at the American Stroke Association’s International Stroke Conference 2025 in Los Angeles, from February 5-7, 2025, electrocardiogram assessments combined with artificial intelligence (AI) may one day be utilized to identify early signs of aging and cognitive decline. This conference is a premier global event focusing on stroke research and brain health.
Stroke can lead to cognitive decline as we age, impacting both life quality and daily activities. An electrocardiogram (ECG) records the heart’s electrical activity. Each heartbeat is triggered by an electrical impulse that travels through the heart. Researchers developed a deep neural network (DNN) model to estimate the biological age of individuals—essentially the age of their body’s cells and tissues—using their ECG data.
According to Bernard Ofosuhene, B.A., the study’s lead author and clinical research coordinator at UMass Chan Medical School in Worcester, Massachusetts, “While chronological age measures how many years a person has lived, ECG-age indicates the heart’s functional condition and possibly the health of the entire body at the tissue level, offering insights into aging and wellness.”
Earlier studies have demonstrated that ECG-age can be a predictor of heart disease and mortality. However, prior to this research, little insight existed regarding ECG-age’s effect on cognitive function.
The research team examined data from the UK Biobank, which includes over 63,000 individuals who volunteered to participate in a long-term study that began with more than 500,000 participants aged 40 to 69. The participants completed several cognitive assessments, which were timed to coincide with their ECG testing, thus providing a precise evaluation of their cognitive state relative to their ECG-age.
From the ECG-age results, participants were categorized into three groups: those experiencing normal aging, those with accelerated ECG aging (older than their chronological age), and those with decelerated ECG aging (younger than their chronological age).
The analysis revealed that those in the decelerated group outperformed the normal aging group in 6 out of 8 cognitive tests, while individuals in the accelerated aging group scored lower on 6 of the 8 tests.
- Participants younger than their chronological age excelled in 6 of the 8 cognitive assessments.
- Participants older than their chronological age performed poorly in 6 of the 8 tests.
Ofosuhene emphasized, “There is a wealth of ECG data available for stroke management, and I urge healthcare providers to utilize this information to identify early signs of cognitive decline. This could facilitate earlier diagnosis and timely intervention.”
There are several limitations to this study. The age range of the 43 to 85 years analyzed may not represent findings applicable to other age groups. Because this is a cross-sectional study, it does not track changes in cognitive performance over time. Furthermore, results obtained from this specific population may not be applicable to other demographics.
“Future studies will aim to explore potential gender differences in the relationship between ECG-age and cognitive performance. Additionally, as many participants in the UK Biobank identify as European ancestry, we are keen to test if our results can be validated in more diverse groups,” said Ofosuhene.
“The connection between cardiovascular and brain health is increasingly acknowledged. This study illustrates that AI can discern a relationship between higher biological age—derived from ECG data—and lower cognitive performance. Employing ECG data to evaluate cognitive abilities sounds like an innovative concept. If validated, this study can lead to significant implications. For instance, ECG readings taken during a doctor’s visit or through wearables could provide cognitive assessments in home settings or in rural areas where neuropsychiatric specialists are scarce. Additionally, analyzing ECG data with AI may be faster and more objective compared to traditional neuropsychological evaluations. One critical question still exists: can ECG data predict future cognitive decline? If answered affirmatively, this could lead to crucial treatments since certain ECG abnormalities are treatable,” remarked Fernando D. Testai, M.D., Ph.D., FAHA, who chairs the upcoming American Heart Association’s scientific statement on the relations between cardiac health and brain performance and is also a professor at the University of Illinois College of Medicine in Chicago, though he was not involved in this research.
Details about the study’s design and background:
- The study involved 63,800 participants (average age 65, 52% female) over a period from August 2023 to July 2024. Most were of white European descent participating in the UK Biobank study, which began enrolling volunteers between 2006 and 2010.
- Participants were excluded if they had missing or invalid ECG or cognitive data.
- The normal aging group included 15,563 individuals, the accelerated group comprised 24,671, and the decelerated group included 23,566 people.
- A total of eight cognitive tests were analyzed. While some UK Biobank participants underwent additional evaluations, the cognitive test results were compared across the three groups, adjusting for chronological age, gender, and education.
