A Control-Theoretic Model of Damage Accumulation and Boundedness in Biological Aging

TL;DR

A compact control-theoretic formulation is developed in which total organismal burden is decomposed into two lesion classes with distinct controllability properties: regulatable damage, whose accumulation and clearance are modulated by endogenous systemic repair, and information-limited damage, whose detection or correction is inaccessible to physiological control.

Abstract

Aging interventions frequently improve function and healthspan without arresting long-term deterioration, indicating that existing frameworks do not fully specify the control conditions required for bounded organismal aging. A compact control-theoretic formulation is developed in which total organismal burden is decomposed into two lesion classes with distinct controllability properties: regulatable damage, whose accumulation and clearance are modulated by endogenous systemic repair, and information-limited damage, whose detection or correction is inaccessible to physiological control. Under mild dynamical assumptions, a sufficiency theorem is established: sustained boundedness of total damage is achieved if and only if endogenous repair persistently exceeds production of regulatable damage and information-limited damage is actively bounded or removed by engineered interventions. Deterministic phase diagrams identify distinct bounded, drifting, and runaway regimes separated by a nontrivial control boundary. A global Latin-hypercube sensitivity analysis with partial rank correlations shows that production of information-limited lesions dominates the asymptotic aging rate, whereas increases in physiological repair capacity have weak marginal influence beyond saturation. Stochastic extensions reveal threshold and sequencing effects relevant to oncogenic risk. The framework yields testable predictions and operational guidance for intervention ordering, biomarker selection, and experimental design in aging research. All conclusions are statements about the dynamical model defined here; biological translation requires empirical identification of observables corresponding to the model variables.

Figures (3)

• Figure 1: Phase diagram of aging dynamics. Long-time behaviour of the damage variable $D(t)$ is classified across damage amplification $\beta$ (x-axis) and repair capacity $\mu$ (y-axis). For each grid point the initial condition $D(0)=D_0$ was integrated to $T_{\mathrm{final}}=200$ time units and classified as: stable homeostasis (tail variance $<10^{-3}$), aging drift (slow increasing tail slope), or runaway damage ($D(T_{\mathrm{final}})>10$). Colours denote regimes: stable homeostasis (purple), aging drift (teal), and runaway damage (yellow). The diagonal boundary indicates a critical trade-off between damage creation and repair: for each $\beta$ a minimum $\mu_c(\beta)$ is required to maintain bounded damage; conversely, for any $\mu$, there exists a maximum tolerated $\beta_c(\mu)$.
• Figure 2: Global sensitivity of the asymptotic aging rate (LHS--PRCC). Partial Rank Correlation Coefficients (PRCC) between model inputs and the long-term slope of total damage $D(t)$ from $N=3000$ Latin-hypercube samples. Deterministic trajectories were integrated to obtain asymptotic slopes; error bars show bootstrap 95% confidence intervals obtained with 1000 resamples. PRCCs were FDR-corrected (Benjamini–Hochberg). The production rate of information-limited lesions ($\beta$) is the strongest positive predictor of the aging slope.
• Figure 3: Identifiability of the effective damage removal rate. Synthetic longitudinal trajectories with measurement noise were generated under removal-limited dynamics $S(t)=\exp(-\beta t)$. The removal rate $\beta$ was inferred from the slope of $\log S(t)$. Estimated values closely matched the ground truth across the tested range, demonstrating that the model parameter corresponds to an observable biological quantity.