New findings on the investigation of accelerated biological aging in patients after acute myocardial infarction
I Nemere, R Skoda, M Nagy, B Deak Bardos, K Hegyi, B Kiraly, I Fejes, A Benczur, I Hizoh, Cs Kerepesi, B Merkely, D Becker, Gy BarcziAbstract
Introduction
The incidence of cardiovascular diseases increases with age. However, biological age may differ from chronological age, and biological age may be more relevant in the prognosis of acute myocardial infarction.
Objective
Our aim was to determine the biological age and accelerated aging of patients with acute myocardial infarction (AMI) using AI-based models combining facial image and laboratory parameters. The developed models were compared, and the relationship between biological age and traditional cardiovascular risk factors was analyzed.
Methods
In the prospective VMAJOR-MI-BIOAGE registry, the biological age of AMI patients was estimated using a combination of three portrait-based (FamousAge-VGG and ResNet, PrisonerAge) and two laboratory-based artificial intelligence models. Accelerated aging was identified based on the difference between biological and chronological age (<5 vs. ≥5 years). Groups were compared according to demographic, lifestyle, and clinical characteristics, and subgroup analyses were performed based on the number of cardiovascular risk factors (<3 vs. ≥3).
Results
To date, 342 patients have been enrolled in the study (36% female, 64% male).Patients with previous PCI showed significantly accelerated aging according to the VGG-Blood (p=0.024) and Prisoner-Blood models (p=0.013). Patients with clustered comorbidities aged significantly faster compared to those with fewer than three risk factors according to the Prisoner-Blood model(p=0.02). Occasional exercisers demonstrated significantly slower biological aging compared to physically inactive patients in both the VGG-Blood (p=0.047) and Prisoner-Blood models(p=0.016). Based on sleep duration analysis, patients sleeping ≥8 hours per day showed significantly lower biological age compared to those with shorter sleep duration in the RES-NET-Blood model(p=0.002). Accelerated biological aging was associated with increased early all-cause mortality in the Blood model (p=0.0038). Using a 0-year cutoff, significant survival differences were observed in the Hunblood (p=0.0367) and Prisoner-Hunblood (p=0.0452) models.
Conclusion
AI-derived biological age is associated with early mortality in AMI and may provide incremental prognostic information beyond traditional cardiovascular risk factors.survival analysis and age prediction