Accelerated epigenetic aging and inflammatory/immunological profile (ipAGE) in patients with chronic kidney disease

Yusipov, Igor; Kondakova, Elena; Kalyakulina, Alena; Krivonosov, Mikhail; Lobanova, Nadezhda; Bacalini, Maria Giulia; Franceschi, Claudio; Vedunova, Maria; Ivanchenko, Mikhail

doi:10.1007/s11357-022-00540-4

Cited by 29 publications

(27 citation statements)

References 61 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…All three biomarkers (FGF21, GDF15, CXCL9) were included in the resulting model with optimal parameters. Results: We demonstrated that ESRD is associated with the significant acceleration of biological age compared with the control group [2]. There was a statistically significant (p < 0,001) increase in the content of all biomarkers (FGF21 GDF15, CXCL9) presented in the ESRD group compared to the controls.…”

mentioning

confidence: 77%

Comprehensive biomarkers of accelerated aging and mortality risk in end-stage renal disease

2022

The Thirteenth International Multiconference

View full text Add to dashboard Cite

mentioning

confidence: 77%

Comprehensive biomarkers of accelerated aging and mortality risk in end-stage renal disease

2022

The Thirteenth International Multiconference

View full text Add to dashboard Cite

“…In this paper, we propose a small SImAge immunological clock that predicts a person’s age based on a limited set of immunological biomarkers. The best known studies in this field include such models as IMM-AGE 30 , iAge 31 , ipAge 32 , which are based on different cytokine datasets and show quite good prediction results.…”

Section: Discussionmentioning

confidence: 99%

“…Various metrics such as MAE 32,42,44,45,51,52,52–55 , RMSE 26,44 , correlation between chronological age and predicted age 28,45,47 , median error 27,50 , and others are used to evaluate the efficiency of machine learning models in solving age estimation problems. The most popular among them is MAE.…”

Section: Methodsmentioning

confidence: 99%

“…Traditionally, one of the most common methods for constructing age predictor models (clocks) for different types of data is Elastic Net. This model is used as the basis for multiple age predictor models based on epigenetic [26][27][28][29] , immunological 32,41 , transcriptomic 42 , metabolomic [43][44][45] , microRNA 46 , and proteomic 47 data. More advanced are the methods of gradient boosting, which are also actively used to construct age predictors, particularly from epigenetic 48 or gut microbiome 49 data.…”

Section: Introduction 1backgroundmentioning

confidence: 99%

See 1 more Smart Citation

Small immunological clocks identified by Deep Learning and Gradient Boosting

Kalyakulina

Yusipov

Kondakova

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Background. The aging process affects all systems of the human body, and the observed increase in inflammatory components affecting the immune system in old age can lead to the development of age-associated diseases and systemic inflammation. Results. We propose a small clock model SImAge using a limited number of immunological biomarkers. To solve the problem of regressing chronological age from cytokine data, we first use a baseline Elastic Net model, gradient-boosted decision trees models, and several deep neural network architectures. On the full dataset for 46 immunological parameters, LightGBM, DANet, and TabNet models showed the best results. Dimensionality reduction of these 3 models with SHAP values revealed the 10 most age-associated immunological parameters, which formed the basis of the SImAge small immunological clock. The best result of the SImAge model has mean absolute error of 6.28 years, it was shown by the DANet deep neural network model. Explicable artificial intelligence methods were used to explain the model solution for each individual participant. Conclusions. We proposed an approach to construct a small model of immunological age, SImAge, using the DANet deep neural network model, which showed the smallest error on the 10 immunological parameters. The resulting model shows the highest result among all published studies on immunological profiles. Since gradient-boosted decision trees and neural networks show similar results in this case, we can consider parity between these types of models for immunological profiles.

show abstract

“…We calculate several epigenetic age estimators, build a linear-regression and elastic-net based immunological age-estimator ipAGE, and implement AI-based age-estimators with individual age-acceleration explained by SHAP values. Results: We demonstrated accelerated aging in terms of the most common epigenetic age estimators in ESRD patients, developed a new clock/predictor of age based on the inflammatory/immunological profile (ipAGE) and identified the differentially expressed inflammatory/immunological biomarkers [1]. IpAGE appeared to be more sensitive than epigenetic clocks in quantifying the accelerated aging phenotype.…”

mentioning

confidence: 99%

Biomarkers of aging and age-related diseases: from linear regression to explainable artificial intelligence

2022

The Thirteenth International Multiconference

View full text Add to dashboard Cite

Motivation and Aim: Discovery of aging biomarkers and development of biological clocks that enable precise estimation of calendar age and sensitive detection of age acceleration, related to the manifestation of age-related diseases remains central to the field. Despite such recent successes as the Horvath and Hannum epigenetic clocks, Phenoage clocks and GrimAge clocks, there is still a keen need to (1) encompassing novel and multi-level biomarkers, (2) elaborating the systems description of aging, (3) developing the age estimators based on advanced Artificial Intelligence tools, including the algorithms that allow for explaining AI based predictions both cohort-wide (population-wide) and individual-wise (personalized). Methods and Algorithms: We investigate simultaneous DNA methylation (Illumina EPIC) and immunological data (multiplex analysis) from human peripheral blood samples for the control group and patients with end-stage renal disease (ESRD) receiving hemodialysis. We calculate several epigenetic age estimators, build a linear-regression and elastic-net based immunological age-estimator ipAGE, and implement AI-based age-estimators with individual age-acceleration explained by SHAP values. Results: We demonstrated accelerated aging in terms of the most common epigenetic age estimators in ESRD patients, developed a new clock/predictor of age based on the inflammatory/immunological profile (ipAGE) and identified the differentially expressed inflammatory/immunological biomarkers [1]. IpAGE appeared to be more sensitive than epigenetic clocks in quantifying the accelerated aging phenotype. Moreover, we developed the mixed epigenetic-immunological AI-based age estimators and explain AI models, identifying epigenetic and immunological factors responsible for age acceleration in each single participant. Conclusion:The obtained results uncover and confirm the deep relation between aging and remodeling of the human immune system, previously conceptualized as the inflammaging theory. The developed immunological clocks allow for a direct inference of immune system aging and shed light on the underlying mechanisms. The most to-date advanced methods of Explainable Artificial Intelligence not only rank the most important populational biomarkers of immunological and epigenetic aging, but allow for the first time to differentially identify specific biomarkers that are responsible for age acceleration of each particular individual.

show abstract

Accelerated epigenetic aging and inflammatory/immunological profile (ipAGE) in patients with chronic kidney disease

Cited by 29 publications

References 61 publications

Comprehensive biomarkers of accelerated aging and mortality risk in end-stage renal disease

Comprehensive biomarkers of accelerated aging and mortality risk in end-stage renal disease

Small immunological clocks identified by Deep Learning and Gradient Boosting

Biomarkers of aging and age-related diseases: from linear regression to explainable artificial intelligence

Contact Info

Product

Resources

About