pith. sign in

arxiv: 2607.01014 · v1 · pith:IGT347HLnew · submitted 2026-07-01 · 🧬 q-bio.PE · math.DS

Approximating Peak Prevalence in Multistage SIR Epidemics

Denis Tverskoi , Andrew Gothard , Grzegorz A. Rempala This is my paper

Pith reviewed 2026-07-02 01:32 UTC · model grok-4.3

classification 🧬 q-bio.PE math.DS
keywords multistage SIR modelspeak prevalenceErlang scalingfactor-two approximationdelay differential equationsmoving averagesincidenceepidemic modeling
0
0 comments X

The pith

Under Erlang scaling multistage SIR models converge to delay equations in which prevalence is twice the triangularly weighted stage maximum.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper examines the link between true peak prevalence and the maximum of a weighted sum over infectious stages in deterministic multistage SIR models. The relationship changes with the scaling chosen for the stage-progression rate as the stage count grows. Naive scaling makes the two peaks asymptotically the same, while Erlang scaling that fixes the mean infectious period produces a delay-equation limit. In that limit prevalence equals the unweighted moving average of incidence and the weighted functional equals the triangular moving average, so their peaks differ by a factor of two. This supplies error bounds and curvature corrections that improve the approximation when waves are sharp.

Core claim

In deterministic SI(k)R models under Erlang scaling that preserves the mean infectious period, as k increases the model converges to a delay differential equation formulation. In this limit the prevalence becomes the unweighted moving average of incidence over the infectious period, and the weighted stage functional becomes the triangularly weighted moving average of incidence. Consequently the factor-two approximation relating their maxima becomes accurate, with error bounds derived from curvature and with curvature-based and parameter-based corrections that improve accuracy across parameter ranges.

What carries the argument

The Erlang scaling of stage-progression rates, which produces convergence to a delay formulation where prevalence and the weighted stage functional are unweighted and triangularly weighted moving averages of incidence.

If this is right

  • The factor-two approximation holds accurately when epidemic waves are not sharply peaked.
  • Analytical error bounds follow directly from the curvature of incidence in the delay limit.
  • Curvature-based and parameter-based corrections substantially improve accuracy of the approximation.
  • Weighted-stage peaks can be used as reliable proxies for prevalence under Erlang scaling.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same moving-average representation could be used to estimate peak burden from partial stage data in ongoing outbreaks.
  • The delay limit connects naturally to other distributed-delay epidemic models that use non-exponential infectious periods.
  • Stochastic versions of the multistage model could be checked to see whether the deterministic convergence still guides the typical peak ratio.

Load-bearing premise

The analysis assumes deterministic differential equations together with either fixed or mean-preserving scalings of the stage-progression rate as the number of stages grows.

What would settle it

Direct numerical integration of the multistage model for successively larger stage counts k under Erlang scaling should show the ratio of prevalence peak to weighted-functional peak approaching two.

read the original abstract

Estimating peak prevalence is a central problem in epidemic modeling because it determines the period of greatest infectious burden and is closely linked to health-care demand. In multistage SIR models, however, peak prevalence is generally less tractable than in the classical model with exponentially distributed infectious periods. Motivated by the use of weighted infectious-stage aggregates as surrogates for prevalence, we investigate the relationship between the prevalence peak and the maximum of a weighted stage functional in deterministic SI$(k)$R epidemic models. We show that this relationship depends critically on how the stage-progression rate is scaled as the number of infectious stages increases. Under naive scaling, in which the progression rate remains fixed, the weighted peak is asymptotically equivalent to the prevalence peak and the commonly used factor-two approximation fails. Under Erlang scaling, which preserves the mean infectious period, the multistage model converges to a delay formulation in which prevalence and the weighted stage functional become unweighted and triangularly weighted moving averages of incidence. This limiting representation provides a theoretical basis for the factor-two approximation and identifies the regimes in which it is accurate. It also explains why this approximation deteriorates as epidemic waves become more sharply peaked. We derive analytical error bounds and develop curvature-based and parameter-based corrections that substantially improve accuracy. Numerical studies confirm these improvements across a broad range of epidemiological parameters. Overall, the results show when weighted-stage peaks can be used reliably as proxies for peak prevalence and how the resulting estimates can be refined when the standard approximation loses accuracy.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

0 major / 1 minor

Summary. The paper claims that in deterministic SI(k)R epidemic models, the relationship between the peak of prevalence and the maximum of a weighted infectious-stage functional depends on the scaling of the stage-progression rate with the number of stages k. Under naive scaling the weighted peak is asymptotically equivalent to the prevalence peak (so the factor-two approximation fails), while under Erlang scaling (which preserves mean infectious period) the system converges to a delay equation in which prevalence equals the unweighted moving average of incidence and the weighted functional equals the triangularly weighted moving average; this limit supplies a theoretical basis for the factor-two approximation together with explicit analytic error bounds, curvature-based corrections, and numerical confirmation across parameter regimes.

Significance. If the derivations hold, the work supplies a rigorous asymptotic foundation for a commonly used approximation in multistage epidemic models, directly linking the factor-two rule to moving-average representations of incidence. The combination of the ODE-to-DDE passage, closed-form error bounds, and parameter-based corrections, together with the numerical studies, strengthens the practical utility for estimating peak infectious burden. The explicit identification of regimes where the approximation deteriorates (sharply peaked waves) is a clear strength.

minor comments (1)
  1. The abstract refers to 'the weighted stage functional' without stating its precise definition; a one-sentence clarification would aid readers who encounter the paper via the abstract alone.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive and constructive report recommending acceptance. No major comments were raised in the review.

Circularity Check

0 steps flagged

No significant circularity; derivation is self-contained

full rationale

The central result is an asymptotic convergence proof: under explicit Erlang scaling (progression rate k/μ) the SI(k)R ODE system converges as k→∞ to a DDE in which prevalence equals the unweighted moving average of incidence and the weighted stage functional equals the triangular moving average. The paper supplies the ODE-to-DDE passage, the moving-average identities, analytic error bounds, and curvature corrections directly from the deterministic equations. No step reduces a claimed prediction to a fitted parameter from the same data, nor does any load-bearing premise rest on a self-citation chain. The factor-two approximation is justified by the derived limit rather than presupposed. This is the normal case of an internally consistent mathematical argument.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on standard assumptions of deterministic epidemic modeling and the mathematical convergence under Erlang scaling; no new entities or fitted parameters are introduced in the abstract.

axioms (2)
  • domain assumption Deterministic compartmental epidemic dynamics governed by systems of ODEs
    The paper uses deterministic SI(k)R models throughout.
  • domain assumption Stage-progression rates are scaled either naively (fixed) or via Erlang scaling (mean infectious period preserved) as k increases
    The abstract states that the relationship depends critically on this scaling choice.

pith-pipeline@v0.9.1-grok · 5809 in / 1534 out tokens · 49432 ms · 2026-07-02T01:32:23.174202+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

15 extracted references · 9 canonical work pages

  1. [1]

    Oxford University Press, Oxford (1991)

    Anderson, R.M., May, R.M.: Infectious Diseases of Humans: Dynamics and Control. Oxford University Press, Oxford (1991)

    1991

  2. [2]

    Princeton University Press, Princeton, NJ (2008)

    Keeling, M.J., Rohani, P.: Modeling Infectious Diseases in Humans and Animals. Princeton University Press, Princeton, NJ (2008)

    2008

  3. [3]

    Imperial College COVID-19 Response Team Report 9 (2020) https: //doi.org/10.25561/77482

    Ferguson, N.M., Laydon, D., Nedjati-Gilani, G.,et al.: Impact of non- pharmaceutical interventions (npis) to reduce covid-19 mortality and healthcare demand. Imperial College COVID-19 Response Team Report 9 (2020) https: //doi.org/10.25561/77482

    work page doi:10.25561/77482 2020

  4. [4]

    Proceedings of the Royal Society of London

    Kermack, W.O., McKendrick, A.G.: A contribution to the mathematical theory of epidemics. Proceedings of the Royal Society of London. Series A115(772), 700–721 (1927) https://doi.org/10.1098/rspa.1927.0118

    work page doi:10.1098/rspa.1927.0118 1927

  5. [5]

    Bulletin of Mathematical Biology53(1-2), 33–55 (1991) https: //doi.org/10.1007/BF02464423

    Kermack, W.O., McKendrick, A.G.: Contributions to the mathematical theory of epidemics–i. Bulletin of Mathematical Biology53(1-2), 33–55 (1991) https: //doi.org/10.1007/BF02464423

    work page doi:10.1007/bf02464423 1991

  6. [6]

    SIAM Review42(4), 599–653 (2000) https://doi.org/10.1137/S0036144500371907

    Hethcote, H.W.: The mathematics of infectious diseases. SIAM Review42(4), 599–653 (2000) https://doi.org/10.1137/S0036144500371907

    work page doi:10.1137/s0036144500371907 2000

  7. [7]

    Lecture Notes in Biomathemat- ics, vol

    MacDonald, N.: Time Lags in Biological Models. Lecture Notes in Biomathemat- ics, vol. 27. Springer, Berlin (1978). https://doi.org/10.1007/978-3-642-93107-9

    work page doi:10.1007/978-3-642-93107-9 1978

  8. [8]

    An Intro- duction to Delay Differential Equations with Applications to the Life Sciences, 119–130 (2011) https://doi.org/10.1007/978-1-4419-7646-8 7

    Smith, H.L.: Distributed delay equations and the linear chain trick. An Intro- duction to Delay Differential Equations with Applications to the Life Sciences, 119–130 (2011) https://doi.org/10.1007/978-1-4419-7646-8 7

    work page doi:10.1007/978-1-4419-7646-8 2011

  9. [9]

    Journal of Mathematical Biology79, 1831–1883 (2019) https://doi.org/10.1007/ s00285-019-01412-w

    Hurtado, P.J., Kirosingh, A.S.: Generalizations of the ’linear chain trick’: Incor- porating more flexible dwell time distributions into mean field ODE models. Journal of Mathematical Biology79, 1831–1883 (2019) https://doi.org/10.1007/ s00285-019-01412-w

    2019

  10. [10]

    SIAM Journal on Applied Mathematics78(6), 3258–3278 (2018) https://doi.org/10.1137/18M1186411

    Champredon, D., Dushoff, J., Earn, D.J.D.: Equivalence of the Erlang-distributed SEIR epidemic model and the renewal equation. SIAM Journal on Applied Mathematics78(6), 3258–3278 (2018) https://doi.org/10.1137/18M1186411

    work page doi:10.1137/18m1186411 2018

  11. [11]

    Princeton University Press, Princeton, NJ (2013)

    Diekmann, O., Heesterbeek, H., Britton, T.: Mathematical Tools for Under- standing Infectious Disease Dynamics. Princeton University Press, Princeton, NJ (2013)

    2013

  12. [12]

    predictive reliability: A case study of the 1978 influenza outbreak

    Tverskoi, D., Rempala, G.A.: Model fit vs. predictive reliability: A case study of the 1978 influenza outbreak. Scientific Reports15, 42110 (2025) https://doi.org/ 10.1038/s41598-025-26072-3 24

    work page doi:10.1038/s41598-025-26072-3 1978

  13. [13]

    ISPRS Journal of Photogrammetry and Remote Sensing (P&RS)118, 83–100 (2016).https://doi.org/10.1016/j

    I¸ slier, Z.G., G¨ ull¨ u, R., H¨ ormann, W.: An exact and implementable computa- tion of the final outbreak size distribution under erlang distributed infectious period. Mathematical Biosciences325, 108363 (2020) https://doi.org/10.1016/j. mbs.2020.108363

    work page doi:10.1016/j 2020

  14. [14]

    Academic Press, Amsterdam (2014)

    Ross, S.M.: Introduction to Probability Models, 11th edn. Academic Press, Amsterdam (2014)

    2014

  15. [15]

    Dover Publications, New York (1981) 25 Appendix A1 Point mass limit in Theorem 2 To complete the proof of Theorem 2, it remains to justify the weak limit of the initial cohort

    Bruijn, N.G.: Asymptotic Methods in Analysis, 3rd edn. Dover Publications, New York (1981) 25 Appendix A1 Point mass limit in Theorem 2 To complete the proof of Theorem 2, it remains to justify the weak limit of the initial cohort. For fixedt >0, consider the measure µt k(dx) :=εkp k,⌊kx⌋(t)dx. For any bounded continuous test functionφ, Z 1 0 φ(x)µ t k(dx...

    1981