ABSTRACT
Einleitung The aging of the immune system is an individual process with high variability: Two persons of the same biological age may differ substantially in the response of their immune system towards diseases and further conditions. The goal of ImmunLearning is to identify cytokines that reflect the immune system's status and to subsequently use these biomarkers to assess a person's 'immuno-fitness' at home or at points-of-care. These could be used, for example, for personalized treatment scheduling without the need for elaborate clinical tests. Methoden To serve our objective of promoting the measurement of biomarkers at home, we studied technologies for cytokine measurement from small whole blood samples with low-end technologies at home and at points of care. We identified a small number of cytokines for which such mature technologies exist. To investigate age-related biomarkers we assessed ex-vivo T-cell marker and serum cytokines levels in samples collected from 74 healthy donors. To take account of the impact of the COVID-19 pandemic, we adapted the sampling procedure and the analytics' workflow, to include recovering COVID-19 patients. To investigate the relationship between cytokine levels and biological age we applied exploratory analysis techniques, regression and machine methods. Our machine learning toolbox encompasses classification algorithms that assess the expected (not biological) age of individuals on the basis of biomarkers, especially cytokine levels;workflows for dimensionality reduction, clustering, visualization and inspection of the association between age strata and cytokine levels;a workflow for the study of correlations among biomarkers in subsamples of individuals with different health conditions. Ergebnisse Our present results indicate a large variance in the biomarker profiles and in the relationship between biomarkers and age strata, both among healthy subjects and among subjects recovered from COVID-19. The correlations among different cytokines indicate that immuno-fitness predictors can be built from a small number of carefully cytokines. Schlussfolgerung Studies on larger samples are needed, especially involving subjects with chronic diseases.