CoChat
Sign Up
Cited by 7Genetics, Aging, and Longevity in Model OrganismsEvolution and Genetic DynamicsTelomeres, Telomerase, and Senescence

The DOI matches its Crossref record; 69 of the 69 references it cites were found in a registry or an index; 14 of the 72 sources its claims cite were checked for support.

A damage accumulation model identifies distinct aging regimes across species

Open access
Raz, Naveh, Yang, Yifan, Pridham, Glen, Shenhar, Ben, Frenkel, Shachaf, Levy, Tomer, Tlusty, Tsvi, Mayo, Avi, Alon, Uri·2026·Nature Aging
DOI 10.1038/s43587-026-01138-7PMC PMC13278975

Published version of a preprint

Open this paper in CoChat — free accountOpen PDF

Abstract

Abstract Different species age in similar ways but their lifespans differ by orders of magnitude. It is not clear how these similarities and differences arise from the accumulation of damage that underlies aging. Does long lifespan arise from reduced damage production, increased removal, or enhanced robustness to damage? To address this we apply a stochastic model of damage accumulation and removal, known as the saturating removal model, and fit it to survival data from well-studied species. Several parameters have near-universal values including removal rate, noise amplitude and death threshold. Lifespan differences arise primarily from the damage production rate, which spans seven orders of magnitude. We identify two distinct aging regimes: ballistic aging where damage production outpaces removal, characterising yeast, nematodes, flies and mice, and quasi-steady-state aging where damage production and removal are balanced until old age, characterizing humans, dogs, and cats. These principles offer a mechanistic basis for comparative aging.

Overview

AI snapshot

Written from the paper’s full text.

TL;DR

The paper fits a stochastic saturating-removal (SR) damage-accumulation model to survival curves across species and finds that damage production rate is the strongest single predictor of lifespan. It identifies two aging regimes—ballistic aging in short-lived species and quasi-steady-state aging in longer-lived mammals—while noting that the inferred parameters are model constructs awaiting experimental validation.

Key findings

  • The damage production parameter η varied by seven orders of magnitude between species and was the best single predictor of lifespan; SHAP mean importance was η = 2.85, versus threshold Xc = 1.53, removal β = 1.38, and noise ε = 0.33.
  • Two aging regimes emerged: ballistic aging, where production outpaces removal, in yeast, C. elegans, flies, and mice; and quasi-steady-state aging, where damage tracks a moving set point of balanced production and removal, in humans, dogs, guinea pigs, and cats.
  • Several parameter combinations were nearly universal in mammals: the noise timescale to threshold Tε = Xc²/ε ≈ 600 (within a factor of 3), and the removal-to-noise timescale ratio βXc/ε ≈ 10 (within about a factor of 2).
  • Species fell on a loglinear relationship in dimensionless parameters, approximately εL/Xc² = 0.17(ηL²/Xc · βL/Xc)^0.5, with humans at one extreme and C. elegans at the other; dogs, cats, guinea pigs, starving E. coli, and Drosophila strain 853 were closest to humans.

Methods

  • The authors assembled high-quality survival/mortality datasets for E. coli (n = 4,744), yeast (n = 24,512), C. elegans (n = 2,908), Drosophila strains (n ~ 1,000 each), mice (n ~ 3,000), dogs (n > 1,000 per breed), cats, guinea pigs, and humans; sample sizes per species ranged from 674 to ~80,000.
  • They simulated the SR stochastic differential equation dx = (ηt − βx/(κ+x))dt + √2ε dW as an Itô process, with death as first passage of damage x across threshold Xc, using 20,000 simulated individuals per parameter set and Bayesian MCMC (Emcee) with Gaussian kernel density likelihoods and log-uniform priors.
  • Hazard curves were classified as Gompertz-like or Weibull-like using BIC, and parameter importance for lifespan was quantified with KernelSHAP across 19 independently calibrated parameter sets.

Limitations

  • The paper states that both species’ lifespans and SR parameters are estimated from the same survival curves, so inferred parameters such as damage production and removal are mathematical constructs rather than directly measured biological quantities; the model does not independently validate unique biological causation and awaits experimental verification.
  • The paper reports an identifiability challenge: multiple parameter sets fit well, forming elongated valleys in parameter space, largely due to uncertainty in the death threshold Xc; simulated data indicated at least n = 500 individuals are needed for errors below twofold for well-defined parameter combinations and below fivefold for most others.
  • The paper notes that the model does not capture the decrease in mortality at old ages in C. elegans, possibly due to phenotypic heterogeneity between plates and/or individuals not modeled; more broadly, the text makes evident that the one-dimensional scalar damage assumption is a simplifying hypothesis whose biochemical identity can differ by species.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1038/s43587-026-01138-7

Cites

69 of 69 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

14 of 72 cited sources checked · 8 supported · 0 flagged · accuracy not yet measured

  • Findingcitation

    In the quasi-steady-state regime, the hazard curve resembles a Gompertz exponential rise over several decades that slows down at very old ages beyond the median lifespan (Fig. 2k).

    “The hazard curve resembles Gompertz exponential rise over several decades which slows down at very old ages beyond the median lifespan (Fig. 2k)2,19,29.”
    Quote in the paperSubstantial claim1 of 3 sources checked
    • cites [2] Alon, U. Systems Medicine: Physiological Circuits and the Dynamics of Disease (CRC Press, 2024). source
    • cites [19] Karin, O., Agrawal, A., Porat, Z., Krizhanovsky, V. & Alon, U. Senescent cell turnover slows with age providing an explanation for the Gompertz law. Nat. Commun. 10, 5495 (2019). source
    • cites [29] Avchaciov, K. et al. Unsupervised learning of aging principles from longitudinal data. Nat. Commun. 13, 6529 (2022). source
  • Premisecitation

    Cellular error rates inverse to mammalian species lifespan include somatic mutations, epigenetic dysregulation, transcription and translation errors.

    “Cellular error rates that have been found to be inverse to mammalian species lifespan include somatic mutations12, epigenetic dysregulation14, transcription and translation errors13.”
    Quote in the paperNot a substantial claim2 of 3 sources checked
    • cites [12] Cagan, A. et al. Somatic mutation rates scale with lifespan across mammals. Nature 604, 517–524 (2022). source
    • cites [14] Crofts, S. J. C., Latorre-Crespo, E. & Chandra, T. DNA methylation rates scale with maximum lifespan across mammals. Nat. Aging 4, 27–32 (2023). source
    • cites [13] Ke, Z. et al. Translation fidelity coevolves with longevity. Aging Cell 16, 988–993 (2017). source

      The citation is attached to transcription and translation errors, not to somatic mutations or epigenetic dysregulation. The Results and discussion report that mistranslation at the first and second codon positions correlates negatively with species maximum lifespan, but say the assay's results are unlikely to reflect transcription errors. The paper therefore does not establish the claimed lifespan relationship for transcription errors. (LLM, conf=0.88)

      Experimental — accuracy not yet measured
    • Rests on an assumption: The damage accumulation model can apply to different forms of damage x in different organisms.
  • Interpretationcitation

    Reduced production of senescent cells in long-lived mammals can be due to enhanced repair of cellular damage, such as double-stranded breaks.

    “Reduced production of senescent cells in long-lived mammals can be due to enhanced repair of cellular damage, such as double-stranded breaks3,9–11,58.”
    Quote in the paperNot a substantial claim3 of 5 sources checked
    • cites [3] Lombard, D. B. et al. DNA repair, genome stability, and aging. Cell 120, 497–512 (2005). source

      The Introduction links DNA-repair defects to premature cellular senescence and identifies double-strand breaks as a type of damage repaired by cellular pathways. The Conclusions predict that enhanced DNA maintenance could retard aging, but the paper does not establish that long-lived mammals produce fewer senescent cells because they repair double-strand breaks more effectively. (LLM, conf=0.84)

      Experimental — accuracy not yet measured
    • cites [9] Popov, A. A., Petruseva, I. O. & Lavrik, O. I. Activity of DNA repair systems in the cells of long-lived rodents and bats. Biochem. Mosc. 89, 1014–1023 (2024). source
    • cites [10] Ma, S. et al. Cell culture-based profiling across mammals reveals DNA repair and metabolism as determinants of species longevity. eLife 5, e19130 (2016). source

      The Results section (“Genes showing positive correlation with lifespan”; page unavailable) links longer lifespan to expression of DNA-repair genes, including one involved in double-strand break repair. The Discussion suggests that longer-lived species may have better repair mechanisms. The paper does not establish that this reduces production of senescent cells. (LLM, conf=0.86)

      Experimental — accuracy not yet measured
    • cites [11] Tian, X. et al. SIRT6 is responsible for more efficient DNA double-strand break repair in long-lived species. Cell 177, 622–638 (2019). source
    • cites [58] Debès, C. et al. Ageing-associated changes in transcriptional elongation influence longevity. Nature 616, 814–821 (2023). source
    • Rests on an assumption: The damage accumulation model can apply to different forms of damage x in different organisms.
  • Findingown data

    Several SR model parameters have near-universal values across species, including ratios of removal rate, noise amplitude and death threshold.

    “Several parameters have near-universal values including ratios of removal rate, noise amplitude and death threshold.”
    • Also stated in: § Main · plain
      “We find that several parameter combinations are nearly universal between species”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: The damage that causes aging can be summarized by a scalar x.
    • The authors state this as an assumption: Life cannot persist above a certain level of damage x.
    • Rests on an assumption: The biochemical nature of x can be different in each species.
    • Rests on an assumption: Damage producing units (DPUs) are produced at a constant rate and their number grows linearly with age.
    • Rests on an assumption: The saturation of removal at high damage is a common feature of biological processes.
  • Findingown data

    There are two distinct aging regimes: ballistic aging and quasi-steady-state aging.

    “We identify two distinct aging regimes: ballistic aging where damage production outpaces removal, characterizing yeast, nematodes, flies and mice, and quasi-steady-state aging, where damage tracks a moving set point of balanced production and removal, characterizing humans, dogs, guinea pigs and cats.”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: The damage that causes aging can be summarized by a scalar x.
    • The authors state this as an assumption: Life cannot persist above a certain level of damage x.
    • Rests on an assumption: The biochemical nature of x can be different in each species.
    • Rests on an assumption: Damage producing units (DPUs) are produced at a constant rate and their number grows linearly with age.
    • Rests on an assumption: The saturation of removal at high damage is a common feature of biological processes.
  • Probably not a substantial claim. 31
  • It may state two separate claims as one. 35

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Premisecitation

    The SR equation was developed by model selection from longitudinal damage data.

    “The SR equation, developed by model selection from longitudinal damage data19, is dx/dt=ηt−βx/(κ+x)+√2ϵξ.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [19] Karin, O., Agrawal, A., Porat, Z., Krizhanovsky, V. & Alon, U. Senescent cell turnover slows with age providing an explanation for the Gompertz law. Nat. Commun. 10, 5495 (2019). source

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Premisecitation

    Long-lived animals share enhanced DNA repair, reduced molecular error rates and other processes.

    “Long-lived animals share enhanced DNA repair3,5,9–11, reduced molecular error rates and other processes6,12–14.”
    Quote in the paperNot a substantial claim5 of 9 sources checked
    • cites [3] Lombard, D. B. et al. DNA repair, genome stability, and aging. Cell 120, 497–512 (2005). source

      The Introduction (page unavailable) reports some evidence that DNA-repair activity differs across species and may influence the rate of age-related decline. That supports a link between repair and lifespan, but not the stronger assertion that long-lived animals generally share enhanced DNA repair. This citation is attached to DNA repair, not to the sentence’s separate claim about molecular error rates. (LLM, conf=0.83)

      Experimental — accuracy not yet measured
    • cites [5] Firsanov, D. et al. Evidence for improved DNA repair in the long-lived bowhead whale. Nature 648, 717–725 (2025). source
    • cites [9] Popov, A. A., Petruseva, I. O. & Lavrik, O. I. Activity of DNA repair systems in the cells of long-lived rodents and bats. Biochem. Mosc. 89, 1014–1023 (2024). source
    • cites [10] Ma, S. et al. Cell culture-based profiling across mammals reveals DNA repair and metabolism as determinants of species longevity. eLife 5, e19130 (2016). source
    • cites [11] Tian, X. et al. SIRT6 is responsible for more efficient DNA double-strand break repair in long-lived species. Cell 177, 622–638 (2019). source
    • cites [6] Rechsteiner, C., Morandini, F., Kim, S. J., Seluanov, A. & Gorbunova, V. Unlocking longevity through the comparative biology of aging. Nat. Aging 5, 1686–1703 (2025). source
    • cites [12] Cagan, A. et al. Somatic mutation rates scale with lifespan across mammals. Nature 604, 517–524 (2022). source
    • cites [13] Ke, Z. et al. Translation fidelity coevolves with longevity. Aging Cell 16, 988–993 (2017). source
    • cites [14] Crofts, S. J. C., Latorre-Crespo, E. & Chandra, T. DNA methylation rates scale with maximum lifespan across mammals. Nat. Aging 4, 27–32 (2023). source

  • It may state two separate claims as one.

    Premisecitation

    The SR model explains many quantitative patterns of aging including Gompertz and Weibull hazard curves, distributions of human frailty index, disease incidence curves and heritability of lifespan.

    “The SR model explains many quantitative patterns of aging including Gompertz and Weibull hazard curves19, distributions of human frailty index2, disease incidence curves31 and heritability of lifespan32.”
    Quote in the paperSubstantial claim1 of 3 sources checked
    • cites [19] Karin, O., Agrawal, A., Porat, Z., Krizhanovsky, V. & Alon, U. Senescent cell turnover slows with age providing an explanation for the Gompertz law. Nat. Commun. 10, 5495 (2019). source

      The Results section, “The SR model quantitatively recapitulates the Gompertz law” (page unavailable), explicitly says the model reproduces the Gompertz law. The citation to this paper is attached to “Gompertz and Weibull hazard curves,” but our copy does not identify a Weibull hazard curve; searches for “Weibull hazard curves” and “Weibull” found no supporting passage. The sentence assigns the frailty-index, disease-incidence, and lifespan-heritability claims to other citations. (LLM, conf=0.86)

      Experimental — accuracy not yet measured
    • cites [2] Alon, U. Systems Medicine: Physiological Circuits and the Dynamics of Disease (CRC Press, 2024). source
    • cites [31] Katzir, I. et al. Senescent cells and the incidence of age-related diseases. Aging Cell 20, e13314 (2021). source

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Premisecitation

    Fedichev et al. proposed a stochastic ODE model derived from expansion of modes in an interacting system, with similarities to the SR model, and noted that mice and humans have quasi-stable versus stable dynamics.

    “Recently Fedichev et al. proposed a stochastic ODE model derived from expansion of modes in an interacting system, with similarities to the SR model, and noted that mice and humans have quasi-stable versus stable dynamics21,22.”
    Quote in the paperNot a substantial claim1 of 2 sources checked
    • cites [21] Fedichev, P. & Gruber, J. A minimal model explains aging regimes and guides intervention strategies. Preprint at BioRxiv https://doi.org/10.1101/2025.08.25.671954 (2025). source
    • cites [22] Pyrkov, T. V. et al. Longitudinal analysis of blood markers reveals progressive loss of resilience and predicts human lifespan limit. Nat. Commun. 12, 2765 (2021). source

      The Results section (“Physiological state fluctuations and loss of resilience”; page unavailable) describes DOSI fluctuations with a stochastic Langevin equation. The Discussion (page unavailable) describes a metastable state in healthy humans and cites mouse studies suggesting dynamically unstable states, but does not describe a quasi-stable-versus-stable species distinction. The passages do not derive the equation from an expansion of modes in an interacting system or compare it with an SR model. (LLM, conf=0.85)

      Experimental — accuracy not yet measured

  • Probably not a substantial claim.

    Findingown data

    Quasi-steady-state aging is a regime where damage tracks a moving set point of balanced production and removal.

    “We identify two distinct aging regimes: ballistic aging where damage production outpaces removal, characterizing yeast, nematodes, flies and mice, and quasi-steady-state aging, where damage tracks a moving set point of balanced production and removal, characterizing humans, dogs, guinea pigs and cats.”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: The damage that causes aging can be summarized by a scalar x.
    • The authors state this as an assumption: Life cannot persist above a certain level of damage x.
    • Rests on an assumption: The biochemical nature of x can be different in each species.
    • Rests on an assumption: Damage producing units (DPUs) are produced at a constant rate and their number grows linearly with age.
    • Rests on an assumption: The saturation of removal at high damage is a common feature of biological processes.

  • It may state two separate claims as one.

    Findingown data

    The parameter that changes the most between species is the damage production parameter η, which is inverse to lifespan.

    “We find that the parameter that changes the most between species is the damage production parameter η, which is inverse to lifespan.”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    The ratio between removal and noise timescales Xcβ/ε is approximately 10, up to about a factor of 2, and is conserved across mammals.

    “The ratio between removal and noise timescales Xcβ/ϵ≈10 up to about a factor of 2 in mammals.”
    • Also stated in: § Discussion · plain
      “Ratios between noise and removal timescales are conserved across mammals”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper may give no reason why it holds.

    Findingnone given

    The death threshold Xc is on the same order of magnitude in most species, Xc ∼ 101.

    “As Xc is on the same order of magnitude in most species (Xc∼10¹), we expect that η∼1/L².”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Findingown data

    Guinea pigs, German Shepherd dogs and Drosophila strains 853 and 441 had a smaller gap (50 < ΔBIC < 100) and have the closest margins in our dataset.

    “Some species, such as humans, mice and yeast had a large gap between the models (ΔBIC > 1,000); the gap was smaller for Guinea pigs, German Shepherd dogs and Drosophila strains 853 and 441 (50 < ΔBIC < 100) and they have the closest margins in our dataset.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Findingown data

    There are two aging regimes: ballistic aging, in which production dominates, and quasi-steady-state aging, in which production is balanced by removal until very old ages.

    “The finding of two aging regimes—ballistic aging and quasi-steady-state aging—can help resolve a debate around the shape of hazard curves.”
    • Also stated in: § Discussion · plain
      “We find two regimes of aging: ballistic aging, in which production dominates, and quasi-steady-state aging, in which production is balanced by removal until very old ages.”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingreasoning

    Plotting the production parameter η against lifespan is consistent with the predicted inverse-quadratic and inverse-linear relationships (Fig. 3a).

    “Plotting the production parameter η versus lifespan is consistent with these predictions (Fig. 3a).”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingnone given

    Starving E. coli have quasi-steady-state aging parameters and a Gompertz-like hazard curve quite similar to humans.

    “Not all short-lived organisms are ballistic; starving E. coli and some Drosophila strains have quasi-steady-state aging parameters and a Gompertz-like hazard curve quite similar to humans.”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    The hazard curves show a diversity of intercepts and shapes, ranging from power-law Weibull-like curves to exponential Gompertz-like curves with appreciable extrinsic mortality (Fig. 2j,k).

    “These show a diversity of intercepts and shapes ranging from power law Weibull-like curves (Fig. 2j) to exponential Gompertz-like curves with appreciable extrinsic mortality (Fig. 2k).”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    An analytical approximation of the SR model with β=0 shows an asymptotic hazard at late times of h(t)=3/16(η/ε)2t2+O(1), in reasonable agreement with the observed yeast hazard curve.

    “An analytical approximation of the SR model with β = 0 shows an asymptotic hazard at late times of h(t)=3/16(η/ϵ)²t²+O(1) (Supplementary Note 5) in reasonable agreement with the observed hazard curve (Fig. 4b).”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    The other model parameters show smaller trends than η and generally decrease more mildly with lifespan.

    “The other model parameters show smaller trends and generally decrease more mildly with lifespan (Extended Data Fig. 1 and Extended Data Tables 2 and 3).”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Findingreasoning

    In longer-lived mammals and one Drosophila strain, removal balances production and tracks it over most of the organism's lifespan (quasi-steady state).

    “In contrast, in longer-lived mammals and one Drosophila strain, we find that removal balances production and tracks it over most of the organism’s lifespan—they are at quasi-steady-state.”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingreasoning

    In the quasi-steady-state regime, production and removal are balanced throughout most of the lifespan and damage tracks the rising production rate (Extended Data Fig. 2).

    “In the second regime—quasi-steady-state aging—production and removal are balanced throughout most of the lifespan and damage tracks the rising production rate (Extended Data Fig. 2).”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    Mice, C. elegans, Drosophila strains 707, 441 and 217 and yeast are closer to the ballistic regime and have different mortality patterns than humans.

    “Mice; C. elegans; Drosophila strains 707, 441 and 217 and yeast are closer to the ballistic regime and have different mortality patterns than humans.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Interpretationreasoning

    The failure to capture the old-age mortality decrease in C. elegans may be due to phenotypic heterogeneity between plates and/or individuals that the model does not represent.

    “This may be due to phenotypic heterogeneity between plates and/or individuals we do not model.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Premisecitation

    The SR model was applied to invertebrates to explain rapid shifts in hazard curves in Drosophila and scaling of survival curves in C. elegans.

    “It was applied to invertebrates to explain rapid shifts in hazard curves in Drosophila and scaling of survival curves in Caenorhabditis elegans2,19.”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [2] Alon, U. Systems Medicine: Physiological Circuits and the Dynamics of Disease (CRC Press, 2024). source
    • cites [19] Karin, O., Agrawal, A., Porat, Z., Krizhanovsky, V. & Alon, U. Senescent cell turnover slows with age providing an explanation for the Gompertz law. Nat. Commun. 10, 5495 (2019). source

  • Probably not a substantial claim.

    Interpretationreasoning

    Biologically, the rise of production rate with age may represent the accumulation of damage producing units (DPUs) that are not removed.

    “Biologically, the rise of production rate with age may represent the accumulation of damage producing units (DPUs) that are not removed.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Damage producing units (DPUs) are produced at a constant rate and their number grows linearly with age.

  • Probably not a substantial claim. It may be too vague to check against evidence.

    Interpretationown data

    This analysis provides insight into the ways that evolution has sculpted strategies to resist aging-related decline across the tree of life.

    “This analysis provides insight into the ways that evolution has sculpted strategies to resist aging-related decline across the tree of life, and provides a framework to compare damage production, removal and noise in different organisms.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Premisecitation

    The SR model was applied in mice to parabiosis, compression of morbidity and combinations of longevity interventions.

    “In mice it was applied to parabiosis33, compression of morbidity and combinations of longevity interventions34.”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [33] Karin, O. & Alon, U. Senescent cell accumulation mechanisms inferred from parabiosis. GeroScience 43, 329–341 (2021). source
    • cites [34] Yang, Y. et al. Compression of morbidity by interventions that steepen the survival curve. Nat. Commun. 16, 3340 (2025). source

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductcitation

    The choice of additive Gaussian noise is for convenience, because other types of noise such as Poisson noise or multiplicative noise terms give similar quality of fits to aging patterns.

    “The choice of additive Gaussian noise is for convenience, because other types of noise such as Poisson noise or multiplicative noise terms give similar quality of fits to aging patterns (Supplementary Note 1).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Supplementary Note 1

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • It may state two separate claims as one.

    Premisecitation

    Senescent cells are removed by the immune system, with a damage half-life of days in mice.

    “Senescent cells are removed by the immune system, so that a damage half-life of days is plausible and supported experimentally in mice19.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [19] Karin, O., Agrawal, A., Porat, Z., Krizhanovsky, V. & Alon, U. Senescent cell turnover slows with age providing an explanation for the Gompertz law. Nat. Commun. 10, 5495 (2019). source
    • Rests on an assumption: The damage accumulation model can apply to different forms of damage x in different organisms.

  • Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.

    Interpretationnone given

    Biologically, Xc describes robustness mechanisms that allow an organism to survive at a given level of damage x.

    “Biologically, Xc describes robustness mechanisms that allow an organism to survive at a given level of damage x.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Interpretationreasoning

    Removal rates seem roughly constant across mammals, whereas repair affects production and is improved in long-lived organisms.

    “Removal rates seem roughly constant across mammals, whereas repair affects production and is improved in long-lived organisms.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The damage accumulation model can apply to different forms of damage x in different organisms.

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductnone given

    No statistical method was used to predetermine sample size.

    “No statistical method was used to predetermine sample size. Instead, we maximized the available power by including all eligible cohorts that met our pre-established quality and homogeneity criteria.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductown data

    There exist several sets of parameters that provide good fits.

    “An identifiability challenge we encountered is the existence of several sets of parameters that provide good fits.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Study design and conductown data

    One needs at least n = 500 individuals to obtain errors smaller than twofold for the well-defined parameter combinations and less than fivefold for most other parameters.

    “Using simulated data with known parameters we find that one needs at least n = 500 individuals to obtain errors smaller than twofold for the well-defined parameter combinations and less than fivefold for most other parameters.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Premisecitation

    Classical Strelher-Mildvan models posit linearly reducing vitality and death caused by exponentially distributed events.

    “Classical Strelher–Mildvan models posit linearly reducing vitality and death caused by exponentially distributed events23,64.”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [23] Strehler, B. L. & Mildvan, A. S. General theory of mortality and aging: a stochastic model relates observations on aging, physiologic decline, mortality, and radiation. Science 132, 14–21 (1960). source
    • cites [64] Li, T. & Anderson, J. Shaping human mortality patterns through intrinsic and extrinsic vitality processes. Demogr. Res. 28, 341–372 (2013). source

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductown data

    Different noise types produced similar results.

    “We used normal Gaussian noise, although different noise types produced similar results (Supplementary Note 1).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. It may state two separate claims as one.

    Premisecitation

    Systems with the same dimensionless parameter groups tend to behave in a similar manner and respond similarly to perturbations.

    “In engineering this principle is called similitude analysis54,55, where systems with the same dimensionless parameter groups tend to behave in a similar manner and respond similarly to perturbations.”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [54] De Rosa, S. & Franco, F. Analytical similitudes applied to thin cylindrical shells. Adv. Aircr. Spacecr. Sci. 2, 403–425 (2015). source
    • cites [55] Casaburo, A., Petrone, G., Franco, F. & De Rosa, S. A review of similitude methods for structural engineering. Appl. Mech. Rev. 71, 030802 (2019). source

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Premisecitation

    The queue theory model posits constant production and removal that drops with age, a mechanism that Karin et al. found to be at odds with damage trajectories in mice.

    “This model posits constant production and removal that drops with age, a mechanism that Karin et al. found to be at odds with damage trajectories in mice19.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [19] Karin, O., Agrawal, A., Porat, Z., Krizhanovsky, V. & Alon, U. Senescent cell turnover slows with age providing an explanation for the Gompertz law. Nat. Commun. 10, 5495 (2019). source

  • It may state two separate claims as one.

    Study design and conductown data

    The parameter sets with the best-fit scores formed elongated valleys in parameter space, denoting a continuum of parameters with a good fit.

    “The parameter sets with the best-fit scores formed elongated valleys in parameter space, denoting a continuum of parameters with a good fit (Supplementary Note 2).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Interpretationreasoning

    Selection to delay aging may push death to occur far beyond the saturation of removal (Xc/κ ≫ 1).

    “One reason may be that selection to delay aging pushes death to occur far beyond the saturation of removal (Xc/κ ≫ 1).”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Interpretationown data

    There are two aging regimes: a ballistic regime found in short-lived species where damage production dominates removal, and a quasi-steady-state aging regime found in mammals where damage tracks a moving set point determined by the balance of rising production and saturating removal.

    “We identify two aging regimes: a ballistic regime found in short-lived species where damage production dominates removal, and a quasi-steady-state aging regime found in mammals where damage tracks a moving set point determined by the balance of rising production and saturating removal.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Aging results from the accumulation of damage over time.

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductown data

    Using the unnormalized values did not change the main result presented in Fig. 3a.

    “Using the unnormalized values did not change the main result presented in Fig. 3a, (Extended Data Fig. 1, black markers).”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Premisecitation

    Reducing production alone would stretch the lifespan but also the sickspan (time of poor health at end of life) by the same factor.

    “Recent analysis indicates that reducing production alone would stretch the lifespan but also the sickspan (time of poor health at end of life) by the same factor34.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [34] Yang, Y. et al. Compression of morbidity by interventions that steepen the survival curve. Nat. Commun. 16, 3340 (2025). source

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.

    Study design and conductreasoning

    These sample sizes are sufficient to provide statistical power for the results presented in this paper.

    “These sample sizes are sufficient to provide statistical power for the results presented in this paper.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Premisecitation

    Other frameworks include failure models in networks of components.

    “Other frameworks include failure models in networks of components65–67.”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites [65] Gavrilov, L. A. & Gavrilova, N. S. The reliability theory of aging and longevity. J. Theor. Biol. 213, 527–545 (2001). source
    • cites [66] Sun, E. D., Michaels, T. C. T. & Mahadevan, L. Optimal control of aging in complex networks. Proc. Natl Acad. Sci. USA 117, 20404–20410 (2020). source
    • cites [67] Mitnitski, A. B., Rutenberg, A. D., Farrell, S. & Rockwood, K. Aging, frailty and complex networks. Biogerontology 18, 433–446 (2017). source

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductown data

    Repeating the simulation without heterogeneity produced qualitatively similar results.

    “We repeated the simulation without heterogeneity and obtained qualitatively similar results (see full results in git in Supplementary Information).”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Premisecitation

    Hazard curves are Weibull-like (power law) in some organisms and Gompertz-like (exponential) in others.

    “Hazard curves are Weibull-like (power law) in some organisms and Gompertz-like (exponential) in others15,59.”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [15] Wilson, D. L. The analysis of survival (mortality) data: Fitting Gompertz, Weibull, and logistic functions. Mech. Ageing Dev. 74, 15–33 (1994). source
    • cites [59] Oswal, N., Martin, O. M. F., Stroustrup, S., Bruckner, M. A. M. & Stroustrup, N. A hierarchical process model links behavioral aging and lifespan in C. elegans. PLoS Comput. Biol. 18, e1010415 (2022). source

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Premisecitation

    Senescent cells are a broad category of damaged cells that enter growth arrest and cause inflammation and stem cell exhaustion.

    “In mammals, one candidate for x is senescent cells—a broad category of damaged cells that enter growth arrest and cause inflammation and stem cell exhaustion56,57.”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [56] Kumar, N. & Walia, H. K. in Cellular Senescence, Age-Related Disorders, and Emerging Treatments (eds Qamar, I. & Maurya, P. K.) 103–133 (Springer Nature, 2025). source
    • cites [57] Liu, Y. & Chen, Q. Senescent mesenchymal stem cells: disease mechanism and treatment strategy. Curr. Mol. Biol. Rep. 6, 173–182 (2020). source
    • Rests on an assumption: The damage accumulation model can apply to different forms of damage x in different organisms.

  • Probably not a substantial claim.

    Interpretationreasoning

    Our inferred parameters-such as damage production and removal-are mathematical constructs rather than directly measured biological quantities.

    “Consequently, our inferred parameters—such as damage production and removal—are mathematical constructs rather than directly measured biological quantities.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Premisecitation

    Another recent stochastic damage accumulation model based on queue theory provides a similar framework for the Gompertz law.

    “Another recent stochastic damage accumulation model based on queue theory provides a similar framework for the Gompertz law24.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [24] Ledberg, A. Exponential increase in mortality with age is a generic property of a simple model system of damage accumulation and death. PLoS ONE 15, e0233384 (2020). source

  • It may state two separate claims as one.

    Interpretationreasoning

    Because η varies widely while removal and noise vary less, the production rate ηt is much larger than removal in some organisms and much smaller in others.

    “Because the production parameter η varies widely, whereas removal and noise vary less, the production rate ηt is much larger than removal in some organisms, and much smaller in others.”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Premisecitation

    Long-living mammals have enhanced repair and reduced error rates in several processes.

    “Long-living mammals have indeed been found to have enhanced repair and reduced error rates in several processes (some of which may not drive aging).”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites [12] Cagan, A. et al. Somatic mutation rates scale with lifespan across mammals. Nature 604, 517–524 (2022). source
    • cites [14] Crofts, S. J. C., Latorre-Crespo, E. & Chandra, T. DNA methylation rates scale with maximum lifespan across mammals. Nat. Aging 4, 27–32 (2023). source
    • cites [13] Ke, Z. et al. Translation fidelity coevolves with longevity. Aging Cell 16, 988–993 (2017). source
    • Rests on an assumption: The damage accumulation model can apply to different forms of damage x in different organisms.

  • It may state two separate claims as one.

    Interpretationown data

    Organisms seem to lie on a continuum between two extremes; at one extreme all parameters are small in units of threshold and lifespan (C. elegans) and at the other extreme all parameters are large (humans).

    “Organisms thus seem to lie on a continuum between two extremes; at one extreme all parameters are small in units of threshold and lifespan (C. elegans) and at the other extreme all parameters are large (humans).”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Premisecitation

    University of Michigan site was found to have longer lifespan for male mice than the other sites.

    “For male mice we used ITP data from the Jackson Laboratory and the University of Texas and excluded data from University of Michigan as this site was found to have longer lifespan for male mice then the other sites71.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [71] Strong, R. et al. Evaluation of resveratrol, green tea extract, curcumin, oxaloacetic acid, and medium-chain triglyceride oil on life span of genetically heterogeneous mice. J. Gerontol. A 68, 6–16 (2013). source

  • It may state two separate claims as one.

    Interpretationreasoning

    In the quasi-steady-state regime, production rises so slowly that removal adapts and becomes more saturated, so that damage rises slowly.

    “Damage production rate rises so slowly that removal can adapt to it, become more saturated and damage rises slowly accordingly.”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Interpretationcitation

    At quasi-steady state dx/dt ≈ 0, production approximately balances removal (ηt − βx/(κ+x) ≈ 0), and mean damage rises with age as ⟨x⟩ ∼ (κηt+ε)/(β−ηt), diverging at tc = β/η.

    “Roughly speaking, at quasi-steady-state there is no change dx/dt≈0, and hence production approximately balances removal ηt−βx/(κ+x)≈0, and therefore mean damage rises with age 〈x〉∼(κηt+ϵ)/(β−ηt) diverging at tc=β/η(ref. 19).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [19] Karin, O., Agrawal, A., Porat, Z., Krizhanovsky, V. & Alon, U. Senescent cell turnover slows with age providing an explanation for the Gompertz law. Nat. Commun. 10, 5495 (2019). source

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Study design and conductreasoning

    Certain parameter combinations have well-defined best fit values, whereas uncertainty in certain parameters led to large CIs.

    “This resulted in certain parameter combinations having well-defined best fit values, whereas the uncertainty in certain parameters led to large CIs.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Interpretationreasoning

    The time for removal processes working at full speed to remove damage equal to the threshold is on the order of months and is ten times faster than the time for noise to reach the threshold.

    “The time for removal processes working at full speed to remove damage equal to the threshold (on the order of months) is ten times faster than the time for noise to reach the threshold.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductreasoning

    Our model does not constitute independent validation of unique biological causation.

    “Although our model describes how these theoretical variables track lifespan variation, it does not constitute independent validation of unique biological causation.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductown data

    Both species' lifespans and the SR parameters are estimated from the same survival curves.

    “A limitation of our approach is that both species’ lifespans and the SR parameters are estimated from the same survival curves.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Study design and conductown data

    The paper provides comprehensive error analysis for its methodology, including dependence of error on sample size, and reports 95% CIs for all primary results.

    “To ensure the robustness of our findings, we provide comprehensive error analysis for our methodology in Supplementary Note 2, including dependence of error on sample size, and report 95% CIs for all primary results.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductown data

    All results are nearly indistinguishable from those using median lifespan calculated from data.

    “All results are nearly indistinguishable from those using median lifespan calculated from data (Extended Data Tables 2 and 3).”
    Quote in the paperNot a substantial claim
58 of 58

  • The paper relies on this without support.

    Different species age in similar ways but their lifespans differ by orders of magnitude.

    “Different species age in similar ways but their lifespans differ by orders of magnitude.”
    “Different animals age in a roughly similar way, with loss of function and a sharp increase in death probability with age. However, their median lifespans vary by many orders of magnitude.”

  • The authors state this as an assumption.

    The damage that causes aging can be summarized by a scalar x.

    “The SR model is based on the simplifying hypothesis that the damage that causes aging can be summarized by a scalar x and that life cannot persist above a certain level of x.”

  • The authors state this as an assumption.

    Life cannot persist above a certain level of damage x.

    “The SR model is based on the simplifying hypothesis that the damage that causes aging can be summarized by a scalar x and that life cannot persist above a certain level of x.”

  • The paper relies on this without support.

    The biochemical nature of x can be different in each species.

    “The biochemical nature of x can be different in each species.”

  • The paper relies on this without support.

    Damage producing units (DPUs) are produced at a constant rate and their number grows linearly with age.

    “The DPUs are produced at a constant rate and therefore their number grows linearly with age.”

  • The paper relies on this without support.

    The saturation of removal at high damage is a common feature of biological processes.

    “The saturation of removal at high damage in the SR model is a common feature of biological processes.”

  • The paper relies on this without support.

    The SR model had not been used to compare the aging of different species systematically.

    “However, the SR model has not been used to compare the aging of different species systematically.”

  • The paper relies on this without support.

    Aging results from the accumulation of damage over time.

    “Aging is thought to result from the accumulation of damage over time.”

  • The paper relies on this without support.

    The damage accumulation model can apply to different forms of damage x in different organisms.

    “The damage accumulation model we use can apply to different forms of damage x in different organisms.”

  • The paper relies on this without support.

    Removal in mammals is due to the immune system, which has a circadian rhythm.

    “Removal in mammals is thought to be due to the immune system, which has a circadian rhythm.”
10 of 10

Citations

Cites

Flagged first, then by citations

Research works printed but not linked: 2

  1. Gompertz, B. On the nature of the function expressive of the law of human mortality, and on a new mode of determining the value of life contingencies. Abstr. Pap. Print. Philos. Trans. R. Soc. Lond. 2, 252–253 (1833).
  2. Lundberg, S. M. & Lee, S.-I. A unified approach to interpreting model predictions. In Proc. Advances in Neural Information Processing Systems Vol. 30 (eds Guyon, I. et al.) (Curran Associates, 2017).

Cited by

Sorted by citations

Related

Metadata

Authors
Raz, N., Yang, Y., Pridham, G., Shenhar, B., Frenkel, S., Levy, T., Tlusty, T., Mayo, A., & Alon, U.
Affiliations
  • Weizmann Institute of Science (IL)
  • Westlake University (CN)
  • Ulsan National Institute of Science and Technology (KR)
Corresponding author
  • Raz, Naveh
  • Alon, Uri
ORCID
Venue
Nature Aging
Publisher
Springer Science and Business Media LLC
Published
2026-06-09
Volume
6
Issue
6
Pages
1330-1340
ISSN
2662-8465 (ISSN-L)
DOI
10.1038/s43587-026-01138-7
PMC
PMC13278975
OpenAlex
W7164018153
Semantic Scholar
ac9ff74aae9ff81a75c8aa72b56e7bb9b6936e89
Cited by
7
References
68
Citations by year
2026: 6
FWCI
7.27
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Aging (publishedVersion, cc-by-nc-nd)
  • Research Square (acceptedVersion, cc-by)
  • PubMed (publishedVersion)
  • PubMed Central (publishedVersion, cc-by-nc-nd)
Funders
  • EC | Horizon 2020 Framework Programme (856487)
  • Zuckerman STEM Leadership Program
  • Sagol Institute for Longevity Research
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Aging › Genetics, Aging, and Longevity in Model Organisms
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Genetics › Evolution and Genetic Dynamics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Cell Biology › Telomeres, Telomerase, and Senescence
Keywords
damage accumulation, aging, lifespan variation, damage removal, comparative aging, survival data
MeSH terms
Aging, Animals, Cats, Dogs, Guinea Pigs, Humans, Longevity, Models, Biological, Species Specificity, Stochastic Processes, Mice

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

A language model drew the claims from the paper on 2026-11-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-11-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review