A deep-learning model combining MRI scans with blood markers, imaging-derived biomarkers, and lifestyle data successfully predicted organ-specific biological age, according to poster data presented May 14 at the International Society of Magnetic Resonance in Medicine (ISMRM) meeting.
The model also identified accelerated aging in participants who went on to develop Alzheimer’s disease and myocardial infarction, wrote a team led by Veronika Ecker of the University Hospital of Tuebingen in Germany.
“Biological age has the potential to capture the combined effects of genetic, lifestyle, and health-related factors on both individual- and organ-specific aging, but remains difficult to quantify,” Ecker and colleagues noted. “Integration of MRI with other health-related data may improve estimation and provide insights into early disease risks.”
MRI offers detailed information on both structure and function of the human body, capturing patterns that reflect individual differences in the aging process, the group explained, writing that “these patterns allow estimation of biological age which may diverge from chronological age due to genetic, environmental, and lifestyle factors.”
The investigators developed a deep-learning model that combined 3D MPRAGE brain MRI and 2D+t cardiac cine MRI results with imaging-derived biomarkers, blood-based measures, and lifestyle information and applied it to 70,000 UK Biobank participants between the ages of 44 and 83 in an effort to predict biological age of the brain and heart and to identify accelerated aging. Because no ground truth for biological age exists, the model was also trained on a subcohort of healthy individuals in which biological and chronological age were assumed to approximate one another, the group explained.
Overall, Ecker and colleagues reported that the model detected a mean brain age gap of 1.98 years and a heart age gap of 0.81 years in disease subgroups compared with healthy controls – a result which suggests that it captures pathological aging processes before clinical diagnosis, they noted.
Predicted age gaps (= predicted age – chronological age) of the brain and heart across chronological age for a healthy test set (blue, upper row) and in comparison with a diseased subcohort (orange, lower row). Diseased subgroups are defined per organ (brain: patients developing Alzheimer’s disease; heart: patients developing myocardial infarction).Veronika Ecker and ISMRM
“Integrating … complementary data sources can strengthen the robustness and interpretability of biological age prediction,” the team wrote, concluding that “the model revealed consistent age-related embeddings, outperformed single-modality approaches, and captured accelerated aging in participants at higher disease risk.”
Check out AuntMinnie.com’s full coverage of ISMRM 2026 on our ShowCast.