This manuscript proposes Theoretical Life Science as a research program for a unified description of life. Its central conjecture is that life is not a static entity in three-dimensional space, nor a simple hierarchical sum of molecules, cells, tissues, and organs. Rather, it is an open, non-equilibrium process on a high-dimensional state manifold. The visible morphology, omics patterns, pathological classifications, and physiological phenotypes observed by humans are projections of this process under limited observational coordinates, finite temporal resolution, and coarse-grained scales. The current aim of Theoretical Life Science is not to restore life, but to develop a theoretical language that can jointly describe life states, open fluxes, randomness, relational structures, topological stability, and viability. A preliminary bioinformatic validation based on a public single-cell aging atlas shows that, as the observational representation expands from low-dimensional markers to a high-dimensional transcriptomic state, aging-stage prediction improves and prediction uncertainty decreases. This provides preliminary support for the testable proposition that low-dimensional projection may create part of the apparent randomness observed in life phenomena. Life-state restoration should therefore be understood only as a long-term implication of a mature theory. Only when life states can be sufficiently described, measured, predicted, and controlled can disease reversion, aging intervention, tissue regeneration, and higher-level reconstruction of life states become rigorous scientific questions.
Zhang et al. (Tue,) studied this question.