Related Experiment Video
Updated: May 22, 2025
Author Spotlight: Automated Lifespan Monitoring – Discovering Aging Dynamics with the Lifespan Machine
Published on: January 26, 2024
Biological age prediction and NAFLD risk assessment: a machine learning model based on a multicenter population in
Lianrui Deng1, Jing Huang2, Hang Yuan3
1Affiliated Rehabilitation Hospital of Nanchang University, Nanchang, China.
Background:
The objective was to develop a biological age prediction model (NC-BA) for the Chinese population to enrich the relevant studies in this population. And to investigate the association between accelerated age and NAFLD.
Methods:
On the basis of the physical examination data of people without noninfectious chronic diseases (PWNCDs) in Nanchang, Jiangxi, China, the biological age measurement method was developed via three feature selection methods (all-subset regression, LASSO regression (LR), and recursive feature elimination) and three machine learning algorithms (generalized linear model (GLM), support vector machine, and deep generalized linear model (deep GLM)). Model performance was evaluated by the coefficient of determination (R²) and mean absolute error (MAE). National Health and Nutrition Examination Survey (NHANES) data were used to verify the model's generalizability. The standardized age deviation (SAD) was calculated to explore the associations between age acceleration and the risk of morbidity and mortality from NAFLD.
Results:
The physical examination data of 26,356 PWNCDs were collected in Nanchang. Among the 26 biomarkers, 26 and 24 biomarkers were associated with chronological age in the male and female groups, respectively (P < 0.05). The model combining the LR and deep GLM algorithms provided the most accurate measurement of chronological age (r = 0.58, MAE = 5.33) and was named the Nanchang-biological age (NC-BA) model. The generalizability of the NC-BA model was verified in the NHANES dataset (r = 0.57, MAE = 7.12). There was a significant correlation between NC-BA and existing biological age indicators (Klemera-Doubal method biological age (KDM-BA), PhenoAge, and homeostatic dysregulation (HD), r = 0.42-0.66, P < 0.05). The physical examination data of 1,663 and 1,445 patients with NAFLD from the Nanchang population and NHANES, respectively, were obtained. The SAD values of NAFLD patients were significantly greater than those of PWNCDs (P < 0.001). The SAD values of NAFLD patients with younger chronological ages were greater (P < 0.001). Higher SAD values were associated with a greater risk of all-cause mortality (HR = 1.73, P = 0.005).
Conclusions:
This study provides a new model for biological age measurement in the Chinese population. There is a clear link between NAFLD and age acceleration.
