{"id":"cd71c28f-1cbe-47ab-84c5-078e2069c771","arxiv_id":"2508.19249","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":4.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":2,"one_line_summary":"Regularized least squares on parameter-linear dynamic models, named PIR, is claimed to estimate epidemic model parameters faster and more accurately than physics-informed neural networks.","lead":"This paper proposes a method for estimating parameters in dynamic models that are linear in their parameters, using regularized least squares. The authors name it Physics-Informed Regression and test it on epidemic models, claiming faster and more accurate results than physics-informed neural networks.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The superiority claim rests on derivative estimation from noisy time series; without details of that step, regularized OLS can be biased, so the PIR-vs-PINN comparison needs a noise robustness test.","rationale":"The reader correctly identified derivative estimation, noise handling, and identifiability as fragile and unstated. I focus on derivative/noise handling because it is the single step most likely to break the central empirical claim. The parameter-linear reformulation is a standard and generally valid modeling observation, and ordinary least squares is a well-understood estimator; the risk lies in the mapping from raw time series to the regression quantities. If that mapping is biased, regularized OLS will yield biased parameter estimates and the comparison against PINN becomes uninformative. The abstract does not show that this mapping is validated, and the full text provided is too corrupted to inspect for internal consistency. I am not claiming the method is wrong; I am saying the current evidence does not establish the superiority claim until a noise-robustness check is performed. Since the reader's verdict was UNVERDICTED and no additional verifiable material is available, the appropriate verdict remains unchanged: further evidence is needed before the central claim can be assessed.","tokens_in":14949,"tokens_out":3748,"duration_ms":48349,"concrete_test":"Re-run the epidemic model experiments with controlled observation noise: generate trajectories from known parameters, add Gaussian noise at several signal-to-noise levels (e.g., SNR 20, 5, and 1), sample at a frequency comparable to the Danish data, and estimate derivatives using exactly the method PIR prescribes (finite differences, splines, or whatever is stated in the paper). For each setting, report parameter relative errors and a bootstrap-based uncertainty interval for both PIR and PINN. If PIR's errors rise steeply with noise, or exceed PINN's errors at realistic noise levels, the superiority claim fails. Additionally, re-fit the real Danish COVID-19 data with two different derivative estimators; if the estimated parameters change materially, the method is not robust to this foundational step.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—that regularized OLS recovers parameters of parameter-linear dynamic models from time series and outperforms PINN—requires that the regression design matrix and target be reliably constructed from the available data. For ODE/PDE models the natural regression target is the state derivative, which is rarely observed directly. If derivatives are estimated by finite differences, noise is amplified and the OLS estimate becomes biased; if they are estimated by smoothing, a bias-variance trade-off is introduced and the bias is not controlled. The abstract reports synthetic data generated from known target parameters and real Danish COVID-19 data, but it says nothing about the derivative estimator, the observation noise model, the regularization choice, or identifiability. In epidemic compartment models, real data are often daily or weekly counts rather than continuous states, making derivative reconstruction especially unreliable. Because the headline experimental conclusion is 'PIR is superior to PINN,' a failure in the derivative-construction step would directly invalidate that conclusion, even if the linear-in-parameters reformulation is mathematically correct. This is a correctness risk in the empirical pipeline, not a disagreement with current consensus.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces Physics-Informed Regression (PIR), a parameter-estimation method for nonlinear dynamic models that are linear in the unknown parameters. The central claim is that, for such models, regularized ordinary least squares applied to a suitably constructed regression system can recover parameters from time-series data. The authors test PIR on ODE and PDE examples, compare it against physics-informed neural networks (PINN) on two epidemic compartment models using synthetic data with known targets and real Danish COVID-19 data, and report that PIR performs noticeably better, especially on the higher-complexity compartment model, while being computationally faster. The abstract also states that PIR can estimate time-varying parameters in a compartment model fitted to Danish COVID-19 data from 2020 to 2021. The supplied full text is unreadable because of encoding corruption, so the derivations, experimental details, error analysis, and comparison protocols cannot be audited from the body of the manuscript.","tokens_in":15280,"tokens_out":2526,"duration_ms":31610,"significance":"If the claims hold, PIR would be a useful and computationally lightweight alternative to PINN for parameter-linear ODE and PDE models, with potential practical value for real-time epidemic modeling. The use of synthetic data generated from known target parameters and the comparison on real Danish COVID-19 data are appropriate ways to support the central claim. In addition, the proposal to extend the method to time-varying parameters addresses a practically relevant modeling need. However, the significance cannot be fully assessed from the abstract alone, and the unreadable full text prevents verification of the mathematical derivations, the construction of the regression design matrix, the handling of observation noise, and the fairness of the PINN comparison. The paper does not, as supplied, allow a reader to confirm the claimed superiority or the computational advantage.","major_comments":[{"comment":"The full text supplied to me is encoded in a way that renders the entire body, including all equations, methods sections, results, and tables, unreadable. As a consequence, the central mathematical derivation, the error analysis, the implementation details, and the comparison protocol cannot be audited. This is a load-bearing issue because the paper's claim that PIR outperforms PINN depends on exactly those details, and the manuscript in its current supplied form does not permit an independent evaluation.","section":"Full text"},{"comment":"The abstract reports that PIR uses regularized ordinary least squares to estimate parameters from time-series data, but it does not state how the regression design matrix and the target are constructed, in particular how state derivatives are obtained when they are not directly observed. Since finite-difference or smoothing-based derivative estimates in noisy epidemic count data can introduce bias, and since the headline conclusion is that PIR is superior to PINN, the absence of any description of the derivative-estimation and noise-handling steps leaves the empirical claim unsupported at the level of detail provided.","section":"Abstract"},{"comment":"The abstract claims that PIR can estimate time-varying parameters of a compartment model, but it does not specify how the time variation is parameterized, what basis or window is used, or how the regularization strength is chosen. These choices determine whether the time-varying estimates are identifiable and whether the reported fit is meaningful, so this extension cannot be evaluated from the information provided.","section":"Abstract"}],"minor_comments":[{"comment":"The first sentence contains an unnecessary comma: 'based on the idea, that if nonlinear...' should read 'based on the idea that if nonlinear...'.","section":"Abstract"},{"comment":"The abstract alternates between 'ordinary least squares' and 'regularized ordinary least squares'; the manuscript should state clearly which estimator is proposed and when the regularization is active.","section":"Abstract"},{"comment":"The term 'physics-informed neural networks (PINN)' uses a singular acronym for what is presumably a family of methods; consider using 'PINNs' or defining the acronym more precisely.","section":"Abstract"},{"comment":"The phrases 'noticeably better' and 'higher complexity' are qualitative; quantitative results and a definition of model complexity should be provided in the main text, though the current encoding prevents verification.","section":"Abstract"}],"recommendation":"uncertain","confidential_remarks":"The full text supplied to me is not readable due to encoding corruption, which prevents any meaningful audit of the manuscript's technical content. If this is a pipeline artifact, I would need a readable version before making a soundness judgment; as it stands, I cannot recommend acceptance or major revision based on evidence."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: the algorithmic core here is not new. Regularized least squares for models linear in parameters has been around for decades in system identification and SINDy-style sparse regression. The name \"Physics-Informed Regression\" is packaging, not a new method. What is potentially useful is the benchmark: ODE and PDE examples, two epidemic compartment models of different complexity, synthetic data with known parameters, real Danish COVID-19 data, and time-varying parameter estimation. That is a concrete contribution worth a serious look.\n\nWhat the abstract does well: it states the setting clearly enough to see the method, it tests on real data rather than only synthetic simulations, it includes a harder compartment model where PINN reportedly struggles, and it reports computational speed, which matters for real-time epidemic monitoring. The claim that PIR is \"superior\" to PINN is too strong for two models and with no uncertainty quantification, but as a provisional finding it is reasonable.\n\nThe soft spot is exactly where the stress-test note points: derivative estimation. For ODE and PDE models, regularized OLS needs a design matrix built from state derivatives or smoothed states. If those derivatives are not observed directly, finite differences amplify noise and can bias the estimates; smoothing introduces an uncontrolled bias-variance trade-off. The abstract says nothing about the observation noise model, the derivative estimator, the regularization choice, or identifiability. That is the load-bearing piece of the empirical pipeline. The claim of superiority rests on it, and a noise robustness test is the right thing to ask for. I could not verify whether the full paper addresses this: the text I received is corrupted mojibake, so I am reviewing from the abstract only. That is not the authors' fault, but it means the decision needs the actual manuscript.\n\nMinor point: the method's novelty is low, but that alone is not a reason to reject. A reproducible, well-tested application of an old technique can be a good paper if it makes the comparison honest and the limitations clear.\n\nThis paper is for people fitting compartmental epidemic models and anyone looking for a cheap alternative to PINN for parameter-linear dynamics. If the full version includes a derivative-robustness study, reproducible code and data, and fair comparison settings, it deserves peer review. I would not desk reject based on the abstract. Send it to a referee who can check the noise handling and the PINN tuning protocol.","headline":"Useful benchmark package wrapped around a well-known trick; the real question is whether the derivative-estimation step makes the PIR-vs-PINN comparison fair.","tokens_in":15679,"tokens_out":2669,"would_cite":false,"duration_ms":34680,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"For dynamic models that are linear in their parameters, regularized least squares can estimate the parameters from time series data and beats physics-informed neural networks on epidemic models at far lower cost.","keywords":["physics-informed regression","parameter estimation","regularized least squares","nonlinear dynamic models","epidemic compartment models","physics-informed neural networks","COVID-19","ODE and PDE models"],"falsifier":"Take a parameter-linear epidemic model with known coefficients, generate synthetic time series with progressively sparser sampling and higher noise, and run PIR; if the estimates become biased or unstable at noise and sampling levels where a conventional nonlinear least-squares fitter still recovers the parameters, the paper's central claim fails.","tokens_in":14788,"feed_emoji":"📊","tokens_out":7557,"duration_ms":82651,"temperature":0.7,"pith_summary":"The paper introduces Physics-Informed Regression (PIR), a hybrid method for estimating parameters of nonlinear dynamic models that are linear in the parameters. The central claim is that once a model is written so that unknown parameters multiply known functions of the state, regularized ordinary least squares can recover those parameters directly from time series data, with no neural-network training needed. The authors demonstrate this on ODE and PDE examples, then compare PIR with physics-informed neural networks (PINN) on two epidemic compartment models, using both synthetic data with known target parameters and real Danish COVID-19 data from 2020 to 2021. PIR estimates the target parameters at least as well as PINN, noticeably better on the higher-complexity compartment model, and far faster, which the paper argues makes it superior for the models considered. The same approach also tracks time-varying parameters from real data, pointing toward fast, possibly real-time inference.","feed_headline":"Least squares beats neural networks on epidemic model fitting","feed_subtitle":"Physics-informed regression estimates nonlinear model parameters faster and more accurately on Danish COVID-19 data.","key_machinery":"The engine is the parameter-linear reformulation plus regularized ordinary least squares. For an ODE written as $\\dot{x}(t)=\\sum_{i=1}^{p} \\theta_i \\phi_i(x(t))$, the parameters $\\theta_i$ appear linearly; substituting observed states and estimated derivatives at multiple times produces a linear system $\\Phi \\theta \\approx b$, and the estimate is $\\hat{\\theta} = \\arg\\min_\\theta \\|\\Phi\\theta - b\\|_2^2 + \\lambda R(\\theta)$. The design matrix $\\Phi$ is built from the known functions $\\phi_i$ evaluated on the data, so the physics enters through the model structure and the data enter through the measurements. For PDEs the same idea works with spatial derivatives discretized. This object does the work: it converts a nonlinear inverse problem into a convex linear algebra problem with a one-shot solution, which is why the paper can claim speed and reliability.","core_discovery":"The core discovery is that parameter estimation in a class of nonlinear dynamic models can be reduced to solving a linear least-squares problem. If the dynamics are parameter-linear, meaning the model can be written as a sum of unknown parameters times known functions of the state, then the observed trajectory and its derivatives can be assembled into a design matrix and a target vector, and regularized ordinary least squares yields the parameter estimates. The paper verifies this on nonlinear ODE and PDE models, then applies it to two epidemic compartment models. Against physics-informed neural networks, PIR recovers the same target parameters from synthetic data, performs noticeably better on the higher-complexity compartment model, and runs much faster. On real Danish COVID-19 surveillance data from 2020 to 2021, PIR also estimates time-varying parameters, demonstrating the method on data that were not generated by the model.","pith_inferences":["Beyond the paper, the parameter-linear regression trick should transfer to other inverse problems, such as chemical kinetics, population biology, or biophysical systems, wherever unknown rates multiply known state functions, making fast online calibration feasible if derivative estimation is reliable.","A test the paper leaves undone is to compare PIR with a classical nonlinear least-squares curve fitter on the same data; if PIR matches that baseline, its value stands independently of the PINN comparison.","The condition number of the PIR design matrix offers a cheap identifiability diagnostic: near-dependent columns mean the data cannot separate certain parameter combinations, a limitation that is hidden in neural-network training but explicit in the regression.","The real-data demonstration uses historical Danish data; a sliding-window or streaming variant of PIR would be the direct route to the real-time estimation the authors suggest."],"forward_implications":["Parameter estimation reduces to a one-shot least-squares solve, so a fitted model can be produced without any neural-network training loop.","On the two epidemic compartment models tested, PIR recovered the known synthetic target parameters and matched or beat PINN, with the largest accuracy gap on the higher-complexity compartment model.","On real Danish COVID-19 data from 2020 to 2021, PIR estimated time-varying parameters, indicating the method works outside synthetic benchmarks.","Because the regression formulation handles both ODE and PDE examples, the method is not tied to epidemic modeling alone."],"supporting_citations":[],"fun_headline_variants":["Least squares outperforms neural nets for epidemic parameter fitting","Physics-informed regression: least squares for nonlinear dynamic models","Least-squares trick turns nonlinear models into linear estimation","Faster than neural nets: least squares fits COVID models well"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the model can be rewritten linearly in its parameters and that the observed time series are sampled densely and cleanly enough to build a trustworthy design matrix, including any state derivatives.","fun_headline_variants_meta":{"raw":{"variants":["Least squares outperforms neural nets for epidemic parameter fitting","Physics-informed regression: least squares for nonlinear dynamic models","Least-squares trick turns nonlinear models into linear estimation","Faster than neural nets: least squares fits COVID models well"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000841,"raw_usage":{"total_tokens":3702,"prompt_tokens":1019,"completion_tokens":2683,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":635,"completion_tokens_details":{"reasoning_tokens":2618}},"tokens_in":635,"tokens_out":2683,"duration_ms":21659,"temperature":1.0,"reasoning_tokens":2618,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T14:31:02.246255+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a parameter-linear epidemic model with known coefficients, generate synthetic time series with progressively sparser sampling and higher noise, and run PIR; if the estimates become biased or unstable at noise and sampling levels where a conventional nonlinear least-squares fitter still recovers the parameters, the paper's central claim fails.","supporting_citations":[],"review_version":1}