revealing fundamental invariants and transitions of complex multiscale patient data: life-span study – AUDACITIES

Physical theories are based on stable mathematical structures, based on regularities and symmetries. In these theories, objects are defined and understood thanks to invariants and transformations preserving the invariants. These invariants allow the synthesis of physical laws, useful for making predictions. In contrast, biological organisms exhibit variability, contextuality, memory effects, where their unique trajectories involve a cascade of changes in their symmetries and a continuous 'reshaping' of existing phenotypes and genotypes, a process that depends on rare events ( i.e. a change of rules). The present proposal hypothesizes that quasi-invariant laws and associated transitions exist in multi-omics multi-phenotypic multi-scale data. We postulate that biological quasi-invariants should be synthesized via a systems approach, not only by numerical summaries (e.g. statistical quantifiers), but also in crucial combination with dynamic summaries, as well as with topological and geometric invariants simultaneously. We will search for quasi-invariant biological laws, their transitions and the emergence of aging contained in unique anonymized human lifespan data (i.e. longitudinal, from 20 years to centenarians) that combine multi-omics and multiple scales. Indeed, we have unique access to two complementary databases: BLSA (Baltimore Longitudinal Study of Aging, NIH, USA) and SLAS (The Singapore Longitudinal Aging Studies, National University of Singapore), which offer us a privileged place to advance this research. We will analyze this data using several tools: geometric and topological data analysis, machine learning, deep learning and structural recurrence analysis.