{"id":"c4cc7c90-73c3-4d16-9374-eed814e62c38","arxiv_id":"2504.13953","paper_version":1,"verdict":"REJECT","confidence":"HIGH","novelty_score":2.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"A curve-fitting comparison of cellular automaton and ODE versions of SI, SIR, and SEIR models finds power-law versus exponential early growth and proposes a hyperbolic tangent fit for both.","lead":"This paper compares three compartmental epidemic models (SI, SIR, SEIR) in cellular automaton and differential equation forms, reporting power-law early growth in the automaton and exponential in the equations. It also fits a hyperbolic tangent to every early growth curve and reports high R-squared values.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The paper's own SEIR ODE fit (R²=0.7509) contradicts the abstract's claim that ODE growth is 'well-represented by an exponential function'; the central CA-power-law/ODE-exponential dichotomy is not supported by the presented evidence.","rationale":"I agree with the reader's REJECT verdict, but I locate the decisive weakness differently from the reader's stated weakest_assumption. The strongest problem is not primarily that numerically equal CA/ODE parameters fail to represent equivalent transmission rates; it is that the paper's own SEIR ODE fit contradicts the abstract's central claim. The authors report R²=0.7509 for Eq. (6) in the SEIR case and then state that exponential and power-law forms do not describe that curve well, while the abstract says ODE growth is 'well-represented by an exponential function.' This internal contradiction is directly load-bearing: it does not depend on any choice about parameter comparability. The tanh claim is also not established by the analysis as presented, because three-parameter in-sample fits to a short simulated segment can achieve high R² for many flexible functional forms; the mechanistic explanation offered is too generic to distinguish tanh from alternatives. A single reproducible check—refitting the SEIR ODE curve with Eq. (6) and reporting R²—would settle the main issue. I therefore recommend REJECT: the current manuscript overstates its central result, and the universal exponential-growth claim would need to be removed or heavily qualified before the paper could be considered. The parameter-equivalence issue flagged by the reader is real but secondary; if the SEIR contradiction were fixed by restricting the claim to SI and SIR, the paper would still need to address parameter comparability to support statements about relative growth speed.","tokens_in":12313,"tokens_out":8529,"duration_ms":86721,"concrete_test":"Recompute the SEIR ODE trajectory with the stated parameters (β=0.25, γ=0.1, ω=0.2, E(0)=I(0)=0.01, h=0.01, RK4) and refit I(t)=K·exp(Bt) over the same fitting window (until I=0.15N) using the same least-squares procedure. If R² remains near 0.7509, the abstract's claim that ODE growth is well-represented by an exponential is false for the SEIR model. As a stricter check, fit the same pre-peak data with Eq. (7) and with a logistic curve, then compare out-of-sample prediction error on the second half of the pre-peak interval; if tanh does not outperform alternatives out of sample, the 'substantial contribution' reduces to in-sample curve fitting with extra parameters.","verdict_should_be":"REJECT","load_bearing_attack":"The most load-bearing problem is internal to the paper rather than a matter of parameter convention. The abstract claims that 'the ODE growth rate is well-represented by an exponential function' for the considered models, but the SEIR ODE fit in Section 5 (Fig. 9(d), Eq. (6)) has R²=0.7509, and the text itself states that 'the power-law and exponential do not describe very well the i curve for SEIR.' Since SEIR is one of the three models the abstract explicitly covers, the universal claim fails on the authors' own numbers. The high-R² hyperbolic-tangent fits (Eq. (7)) do not repair this: they use three free parameters fitted in-sample to the same simulated curve, with no out-of-sample validation and no comparison against other sigmoid families; the mechanistic rationale (tanh has odd powers in its Taylor expansion and exponentials in its definition) applies to any sufficiently flexible sigmoid and therefore cannot establish the claimed universal growth law. Separately, setting β=0.1 in both CA and ODE does not make the transmission rates equivalent—CA β is a per-neighbor infection probability per time step while ODE β is a per-capita contact rate per unit time—so the reported difference in growth speed is confounded. But even setting that issue aside, the SEIR exponential fit alone invalidates the abstract's headline statement.","agreement_with_reader":"partial"},"referee_report":null,"author_rebuttal":null,"desk_editor":null,"rs_alignment":null,"lean_confirmation":null,"pith_extraction":null,"created_at":"2026-08-16T12:38:42.751070+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":null,"supporting_citations":[],"review_version":1}