Predicting Long-Term Biological Dynamics from Early-Time State-Space Geometry

The trajectories of complex biological systems are commonly inferred from long-term observations of recovery or deviation after perturbation. We suggest that early-time state-space geometry could contain information enough to anticipate system trajectories before recovery. This hypothesis is informed by extensions of the quantum adiabatic theorem suggesting that under fast, nonadiabatic perturbations, a system prepared in its ground state within the same phase retains the largest overlap with the post-perturbation ground state. Translating to biological systems, we consider cellular functional identity as a stable attractor in a high-dimensional state space where abrupt perturbations like brief inflammatory pulses do not induce regime transitions. Our simulations suggest that post-perturbation states distribution is biased toward the original attractor, reflecting persistence of structural alignment rather than uniform exploration of accessible configurations. Early-time overlap with the baseline attractor, attractor dominance and state-space entropy could stand for operational metrics for inferring system fate. Higher initial overlap should correspond to increased return probability and reduced dispersion, whereas reduced overlap may indicate proximity to regime boundaries. We predict that system fate can be inferred from initial post-perturbation configurations without requiring long-term observation. Potential applications of our framework include fast assessment of cellular resilience, early identification of instability preceding disease transitions and optimization of intervention strategies based on early system responses.