{"id":"f18a7644-8443-444a-8fe2-45c4985f51ea","arxiv_id":"1908.00687","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":3.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"In viral infection models, a nonmonotonic (hump-shaped) immune response creates a bistable interval between a post-treatment control threshold and an elite control threshold, while a monotonic response does not.","lead":"This paper analyzes virus-immune models with monotonic and nonmonotonic immune responses, showing that only nonmonotonic responses can produce two stable outcomes (bistability). It derives thresholds for immune intensity that separate viral rebound from viral control, which could help explain why some HIV patients control the virus after treatment.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No-bistability claims for the 3D systems rest only on equilibrium stability; stable limit cycles are never ruled out, so the categorical wording overstates the proof.","rationale":"The paper's central contribution is the contrast between monotone immune responses (no two stable equilibria) and Monod-Haldane responses (bistable interval between c2 and c**1). The equilibrium existence and stability analysis is mostly coherent and the thresholds are derived from the model, so the main mechanism is plausible. The weakest point is that the non-existence of bistability in the 3D systems is concluded from local equilibrium data; no Lyapunov function, Dulac-type argument, or Li-Muldowney criterion is given to rule out stable periodic attractors, and the paper nowhere defines bistability. This does not refute the claim under the standard two-stable-equilibria reading, but it makes the categorical phrasing unsupported. I also noted a separate concrete error: in Theorem 5.6 the transcritical bifurcation is said to be 'between E1 and E4-', whereas at c=c**2 the equilibrium that collides with E1 is E4+ (the larger y-root), since E4- has z>0 and does not meet the boundary; the Sotomayor conditions still support a transcritical bifurcation, so this is a correctable mislabeling rather than a fatal flaw. These issues justify a CONDITIONAL verdict, not a rejection.","tokens_in":20171,"tokens_out":27375,"duration_ms":271977,"concrete_test":"Apply the Li-Muldowney criterion (the 3D Bendixson-Dulac analogue) to system (1.1) with a generic monotone f: form the second additive compound matrix of the Jacobian and check whether its Lozinskiĭ measure is negative on the positive orthant. If the inequality holds, no simple closed orbits exist and Remark 2.1 is fully justified. If it fails, run long-time numerical integrations of (1.1) with f(y)=cy and f(y)=cy/(1+αy) over the parameter ranges used in §3.5 from many initial conditions, and test for periodicity via return maps or Floquet multipliers; a stable cycle coexisting with E1 or E* would refute the no-bistability claim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The categorical assertions that the 3D systems have no bistability 'in other cases' (Remark 2.1 for the monotonic system (1.1); the corresponding half of Remark 3.1 for the nonmonotonic system (1.3)) are inferred from the local stability analysis of equilibria alone. Theorems 2.1-2.3 and 3.2-3.4 show, at most, that no two stable equilibria coexist; they do not exclude stable periodic solutions. For the 2D systems this gap is closed by the Dulac arguments (Lemmas 4.2 and 5.5), but no 3D analogue is supplied. The Routh-Hurwitz checks at the positive equilibria only prevent Hopf bifurcation at those equilibria; they say nothing about limit cycles created by global bifurcations or about a stable cycle coexisting with a stable equilibrium. Thus the wording 'has no bistability appear' is stronger than what is proved. If the intended definition of bistability is specifically coexistence of two stable equilibria, the equilibrium analysis does support the results and the paper should state that definition explicitly; if periodic attractors are admitted, the claim is unproven.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper analyzes four ODE models of viral infection with immune response: a 3D model and a 2D model, each with a monotone immune response, and corresponding versions with a nonmonotone Monod-Haldane immune response. For the monotone models the authors prove local stability of the equilibria and assert that no bistability occurs. For the nonmonotone models they derive two thresholds, the post-treatment control threshold c2 and the elite control threshold c**1 (or c**2), and prove that for immune intensity between these thresholds the system has two stable equilibria. They also apply Sotomayor's theorem to establish saddle-node bifurcation at c2 and transcritical bifurcation at the elite control threshold, and they support the analytical results with numerical simulations.","tokens_in":20390,"tokens_out":12989,"duration_ms":119042,"significance":"The proposed contrast between monotone and nonmonotone immune responses as a mechanism for bistability is biologically relevant and gives a clean mechanistic interpretation of post-treatment control versus viral rebound. A notable strength is that the thresholds are derived from the model parameters rather than fitted to data, and the 2D no-bistability statements are backed by Dulac's criterion, which excludes limit cycles. The bifurcation computations are explicit and checkable. The main limitation is that the corresponding no-bistability statements for the 3D systems are not global and therefore overstate the proved results if periodic attractors are admitted.","major_comments":[{"comment":"The categorical statements that the 3D monotone system (1.1) and the 3D nonmonotone system (1.3) have no bistability outside the bistable interval are supported only by local stability analysis of equilibria. Theorems 2.1–2.3 and 3.2–3.4 show, at most, that no two stable equilibria coexist; they do not rule out stable limit cycles in R3, so coexistence of a stable equilibrium with a stable periodic orbit is not excluded. If the authors intend bistability to mean coexistence of two stable equilibria, that definition should be stated explicitly and the remarks restricted accordingly; otherwise the claims overstate what is proved.","section":"Remarks 2.1 and 3.1; Theorems 2.1–2.3 and 3.2–3.4"},{"comment":"The proof of Theorem 5.3 contains a sign error: it states that the eigenvalue λ2 = cy1/(α+γy1+y1^2) − b satisfies λ2 > 0 for 0 < c < c**2, but this eigenvalue is in fact negative exactly when c < c**2. Since the theorem's conclusion that E(4)_1 is locally and globally stable requires λ2 < 0, the proof as written contradicts its own assertion and must be corrected.","section":"Theorem 5.3, Section 5.1"},{"comment":"The transcritical bifurcation proofs apply Sotomayor's theorem but do not verify the nondegeneracy conditions in the exceptional case R0 = Rc. At this parameter point y1 = √α, so α − y1^2 = 0, which makes the Sotomayor coefficient Γ3 (resp. Φ3) zero; in the 3D case c[tc] then coincides with c2 = c[sn], so the two bifurcations collide and the bifurcation is degenerate. The theorems state only R0 > 1 and c = c[tc], so this case is included; the authors should either exclude R0 = Rc or analyze the degenerate case separately.","section":"Theorems 3.6 and 5.6, transcritical bifurcation"}],"minor_comments":[{"comment":"The transcritical-bifurcation theorem is numbered Theorem 5.6 in both sections; the theorem in Section 3.4 should be renumbered Theorem 3.6.","section":"Section 3.4 and Section 5.3"},{"comment":"Figure 1 reports c2 = 0.2500 and c**1 ≈ 0.6505, whereas the text for the same parameter set (3.1) gives c2 = 0.3000 and c**1 ≈ 0.3837; the caption and text should be reconciled.","section":"Figure 1 caption and Section 3.5"},{"comment":"The text states that the bistable interval is (2.5000, 3.8333), but the computed value c**2 ≈ 3.5278 and the caption of Figure 5 give (2.5000, 3.5278); the numbers should be made consistent.","section":"Section 5.4"},{"comment":"The parameter list uses the symbol γ twice (γ = 6 and γ = 0.5); the two parameters should be given distinct names to avoid ambiguity.","section":"Section 5.4, parameter list"},{"comment":"The conclusion that E2+ (resp. E4+) is an unstable saddle should be justified by the full Routh-Hurwitz sign pattern rather than by the single inequality b3 < 0 (resp. b2 < 0), since that inequality alone does not identify the eigenvalue configuration.","section":"Proofs of Theorem 3.4(ii) and Theorem 5.4(ii)"},{"comment":"The sentence 'If c < c[sn], there is no positive equilibrium and there is two positive equilibria' should be split into the two cases: no positive equilibrium for c < c[sn] and two positive equilibria for c > c[sn].","section":"Section 3.3"}],"recommendation":"major_revision","confidential_remarks":"The manuscript has the core of a correct and useful analysis, but it needs a careful proofreading pass: theorem numbering, figure captions, and parameter values are inconsistent in several places, and the English phrasing throughout ('has no bistability appear', 'will undergoes') is below the standard expected for publication. The authors also cite their own submitted work [20] as a source for one of the models or thresholds; the editor may wish to confirm that unpublished material is not essential to the main results."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nQuick take: the genuinely new part is the 3D nonmonotonic model (1.3) with the Monod-Haldane immune response, and the central result there is real: in the parameter region R0 > Rc, there is a bistable interval (c2, c**1) where the immune-free equilibrium and the positive equilibrium E2-* are both stable. The explicit thresholds c2 = γb + 2b√α and c**1 = γb + bd(R0-1)/(β(1-ε)) + αβb(1-ε)/(d(R0-1)) are derived correctly, and the Sotomayor-based saddle-node and transcritical bifurcation computations check out. No data are fitted; the thresholds come from the model, so the circularity burden is low. The 2D sections, by contrast, largely overlap the authors' earlier paper [25].\n\nWhere I would push back:\n\n- The categorical claim that the monotonic 3D system \"has no bistability\" (Remark 2.1) is proved only for equilibria. The 2D systems get Dulac-style no-cycle arguments, but there is no 3D analogue. The Routh-Hurwitz check at the positive equilibrium rules out Hopf there, but it does not exclude stable limit cycles coexisting with a stable equilibrium. If bistability is meant as coexistence of two stable equilibria, then the claim is supported and should be stated as that definition explicitly. If periodic attractors count, the claim is unproven. The same gap affects the \"in other cases\" half of Remark 3.1.\n\n- The abstract and conclusion overstate the threshold picture. \"Below c2 the virus rebounds, above c**1 it is controlled, between them bistable\" is only true when R0 > Rc. For 1 < R0 < Rc the thresholds order differently and there is no bistable interval. The paper's Table 2 gets this right, but the abstract does not and should carry the condition.\n\n- Editorial quality is below what I would want to see. Figure 1's caption gives different threshold values from the text (0.2500/0.6505 versus 0.3000/0.3837), the transcritical theorem in Section 3 is numbered \"Theorem 5.6\", and there are many typos. Minor but worth saying: the saddle conclusion for E2+* is inferred from the sign of b3; it is correct because the remaining two eigenvalues then have negative real parts, but that step is not written out.\n\nNet: I did not find a fatal error in the core computations. The paper deserves a serious referee, not a desk reject, but my recommendation would be major revision: define bistability, close or explicitly caveat the limit-cycle gap, qualify the abstract, and clean up the figures and numbering. I would not cite it in its current form.","headline":"Solid but rough: the Monod-Haldane bistable interval is a real, explicitly computed result, yet the no-bistability wording outruns the proof and the manuscript needs revision before I would cite it.","tokens_in":20926,"tokens_out":9058,"would_cite":false,"duration_ms":90558,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35B35","35B40","92D25"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper claims that nonmonotonic immune responses can make viral rebound and viral control both stable for the same parameters, while monotonic immune responses cannot.","keywords":["monotonic immune response","nonmonotonic immune response","Monod-Haldane function","bistability","post-treatment control threshold","elite control threshold","saddle-node bifurcation","transcritical bifurcation"],"falsifier":"A concrete check: scan system (1.1) with a monotonic $f$ over the $b$–$p$ plane for two simultaneously stable equilibria, or continue periodic orbits numerically; finding a stable limit cycle coexisting with a stable equilibrium would show that the equilibrium-only version of Remark 2.1 leaves out an attractor, while finding two stable equilibria would refute it outright.","tokens_in":19951,"feed_emoji":"🦠","tokens_out":15857,"duration_ms":142518,"temperature":0.7,"pith_summary":"Viral infection models in which immune-cell growth follows a nonmonotonic Monod-Haldane curve can settle into either of two stable states for the same parameters: an immune-free state in which the virus rebounds and an immune-present state in which the virus is controlled. The paper fixes the interval of immune intensity $c$ in which this happens, bounded below by the post-treatment control threshold $c_2=\\gamma b+2b\\sqrt{\\alpha}$ and above by the elite control threshold $c_1^{**}$. For models with a monotonic immune response, the paper finds the equilibria are stable in mutually exclusive parameter regions and concludes that no such bistability appears. The distinction matters because it identifies the shape of the immune-stimulation curve, not just its strength, as a possible mechanism behind post-treatment control versus viral rebound.","feed_headline":"A nonmonotonic immune response opens a bistable interval","feed_subtitle":"Below the post-treatment threshold the virus rebounds; above elite control it is suppressed; between them, both outcomes are stable.","key_machinery":"The engine is the hump-shaped Monod-Haldane immune response $f(y)=cy/(\\alpha+\\gamma y+y^2)$. Because $f(y)$ rises to a peak and then falls, the balance equation $f(y)=b$ can have two positive roots $y_*^{2-}$ and $y_*^{2+}$ once $c>c_2$, and the sign of $\\alpha-y^2$ decides which of these equilibria is stable and which is a saddle. The same nonmonotonic shape places the exchange of stability between the immune-free and immune-present equilibria at the transcritical threshold $c_1^{**}$ and the merging of the two positive equilibria at the saddle-node threshold $c_2$. Characteristic-equation stability analysis and the transversality conditions for bifurcations certify these transitions.","core_discovery":"Consider system (1.3), where uninfected cells $x$, infected cells $y$, and immune cells $z$ evolve with immune growth $f(y)=cy/(\\alpha+\\gamma y+y^2)$. Under $\\gamma>2\\sqrt{\\alpha}$ and $R_0^{(2)}>R_c^{(1)}>1$, the paper shows that for $c_2<c<c_1^{**}$ the immune-free equilibrium $E_1^{(2)}$ and the positive equilibrium $E_2^{*-}$ are simultaneously locally asymptotically stable, while $E_2^{*+}$ is a saddle: this is the bistable interval. Here $c_2=\\gamma b+2b\\sqrt{\\alpha}$ is the post-treatment control threshold below which the immune-free state is the only stable outcome, and $c_1^{**}$ is the elite control threshold above which the positive equilibrium is the only stable outcome. The 2D analog (5.1) has the same threshold structure with its own elite threshold. For the monotonic systems (1.1) and (4.1), where $f'(y)>0$, the equilibria are stable in mutually exclusive regimes, so the paper concludes that bistability does not occur.","pith_inferences":["If the threshold picture transfers to therapy, a patient whose immune intensity lies in the bistable interval should be pushed above $c_1^{**}$ by a temporary boost to immune stimulation; the paper does not model treatment, so this is an extrapolation.","Because the 3D monotonic analysis proves local stability and one global result but does not rule out stable limit cycles, a broader definition of bistability that counts oscillatory attractors might still allow bistability in monotonic systems; that question is not settled here.","The hump-shaped mechanism depends only on the shape of $f(y)$, so the same threshold structure may appear in other within-host infections whenever immune stimulation falls off at high viral load; the paper itself only analyzes the specific model families."],"forward_implications":["In the nonmonotonic model, immune intensity below $c_2$ leaves only the immune-free equilibrium stable, so the virus rebounds after treatment.","Above $c_1^{**}$ only the immune-present equilibrium is stable, so the virus stays under control.","In the bistable interval $(c_2,c_1^{**})$, two stable outcomes coexist, so the same host parameters can produce either rebound or control depending on the initial infection load.","The saddle-node at $c_2$ and the transcritical at $c_1^{**}$ are the organizing events through which the stable and unstable equilibria appear and disappear as immune intensity $c$ varies.","For the monotonic systems, no two stable equilibria coexist, so each parameter regime has a single eventual outcome."],"supporting_citations":[{"why":"Supplies the three-dimensional infection-immune system that models (1.1) and (1.3) are built on.","marker":"[12]"},{"why":"Introduces the Monod-Haldane growth function whose role in the model is the nonmonotonic immune response.","marker":"[13]"},{"why":"Proposes the simplified Monod-Haldane form used here, giving the hump-shaped response that creates two positive equilibria.","marker":"[14]"},{"why":"Provides an HIV infection model in which the immune function is a Monod-Haldane function, motivating the choice of $f(y)$ in system (1.3).","marker":"[20]"},{"why":"Defines the post-treatment control and elite control thresholds, which the paper identifies with $c_2$ and $c_1^{**}$.","marker":"[21]"},{"why":"Supplies the invariance-principle argument used to prove global stability of equilibria in several theorems.","marker":"[22, 23]"},{"why":"Supplies the bifurcation theorem whose transversality conditions certify the saddle-node and transcritical bifurcations.","marker":"[26, 27, 28]"}],"fun_headline_variants":["Bistability emerges only in nonmonotonic immune response","Nonmonotonic immune response enables bistable viral control","Immune intensity bounds define bistable interval in infection","Viral control bistable only for nonmonotonic immune response","Two immune thresholds separate bistable from stable outcomes"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The no-bistability conclusion for the monotonic systems assumes that 'bistability' means two stable steady states; if repeating cycles count as states, the 3D monotonic case is not settled, because Theorems 2.1–2.3 do not rule out stable oscillations.","fun_headline_variants_meta":{"raw":{"variants":["Bistability emerges only in nonmonotonic immune response","Nonmonotonic immune response enables bistable viral control","Immune intensity bounds define bistable interval in infection","Viral control bistable only for nonmonotonic immune response","Two immune thresholds separate bistable from stable outcomes"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000605,"raw_usage":{"total_tokens":2825,"prompt_tokens":954,"completion_tokens":1871,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":570,"completion_tokens_details":{"reasoning_tokens":1790}},"tokens_in":570,"tokens_out":1871,"duration_ms":16770,"temperature":1.0,"reasoning_tokens":1790,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T15:39:41.896611+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A concrete check: scan system (1.1) with a monotonic $f$ over the $b$–$p$ plane for two simultaneously stable equilibria, or continue periodic orbits numerically; finding a stable limit cycle coexisting with a stable equilibrium would show that the equilibrium-only version of Remark 2.1 leaves out an attractor, while finding two stable equilibria would refute it outright.","supporting_citations":[{"cited_title":"Nowak, C.R","cited_arxiv_id":null,"evidence_quote":"Supplies the three-dimensional infection-immune system that models (1.1) and (1.3) are built on."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the Monod-Haldane growth function whose role in the model is the nonmonotonic immune response."},{"cited_title":"Sokol, J.A","cited_arxiv_id":null,"evidence_quote":"Proposes the simplified Monod-Haldane form used here, giving the hump-shaped response that creates two positive equilibria."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides an HIV infection model in which the immune function is a Monod-Haldane function, motivating the choice of $f(y)$ in system (1.3)."},{"cited_title":"Conway, A.S","cited_arxiv_id":null,"evidence_quote":"Defines the post-treatment control and elite control thresholds, which the paper identifies with $c_2$ and $c_1^{**}$."}],"review_version":1}