pith. sign in

arxiv: 2605.19886 · v1 · pith:IPG6TEJCnew · submitted 2026-05-19 · 🧮 math.DS

Constraint-Aware Physics-Informed Neural Networks for SEIR Reaction-Diffusion Epidemic Models with Vital Dynamics

Achraf Zinihi , Matthias Ehrhardt This is my paper

Pith reviewed 2026-05-20 01:39 UTC · model grok-4.3

classification 🧮 math.DS
keywords physics-informed neural networksSEIR modelreaction-diffusion equationsepidemic modelingparameter identificationconstraint-aware optimizationspatiotemporal dynamicsinverse problems
0
0 comments X

The pith

Constraint-aware physics-informed neural networks reconstruct spatiotemporal SEIR epidemic dynamics from sparse or noisy data.

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

This paper develops a physics-informed neural network approach for SEIR reaction-diffusion models that include vital dynamics and homogeneous Neumann boundaries. The method embeds the governing equations, available observations, boundary conditions, and explicit constraints on non-negative compartment sizes and admissible parameters into a single loss function for training. A reader would care because real epidemic data sets are usually incomplete and noisy yet forecasts must still respect basic biological limits such as never producing negative case counts. The authors generate reliable synthetic test data with structure-preserving numerical schemes and then demonstrate both forward prediction of spread and inverse recovery of rates in one- and two-dimensional domains. The central result is that adding these constraints allows the network to recover accurate dynamics and parameters even when measurements are limited or corrupted.

Core claim

The constraint-aware PINN integrates PDE residuals, observational data, boundary conditions, and epidemiological constraints within a unified optimization procedure. The loss function enforces non-negativity of compartment populations and admissibility of epidemiological parameters. Applied to forward simulation and inverse parameter estimation, the method accurately reconstructs spatiotemporal epidemic dynamics and identifies parameters reliably even with sparse or noisy data.

What carries the argument

The unified loss function that combines PDE residuals, data fit, boundary conditions, and explicit constraints enforcing non-negative populations and admissible parameters during neural-network training on the SEIR reaction-diffusion system.

If this is right

  • The framework enables accurate forward simulation of epidemic spread in one- and two-dimensional spatial domains.
  • It supports reliable inverse estimation of epidemiological parameters from limited observations.
  • The reconstruction and identification remain accurate when data are sparse or contain noise.
  • The approach supplies a data-driven methodology for spatial epidemic modeling that automatically respects biological non-negativity and parameter bounds.

Where Pith is reading between the lines

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

  • The same constraint-embedding strategy could be transferred to other reaction-diffusion systems that require positivity or conservation laws.
  • Once trained on synthetic data the networks might serve as fast surrogate simulators for exploring intervention scenarios in larger spatial domains.
  • The method opens a route to hybrid modeling in which sparse real outbreak reports are supplemented by physics constraints rather than by dense additional measurements.

Load-bearing premise

The synthetic benchmark data produced by the structure-preserving NSFD schemes faithfully represents the true solutions of the SEIR reaction-diffusion system and can therefore serve as valid ground truth.

What would settle it

Train the network on NSFD-generated trajectories and then compare its predictions and recovered parameters against an independent high-resolution numerical solution on the same parameters but with a different discretization scheme; large systematic discrepancies would indicate the claim does not hold.

Figures

Figures reproduced from arXiv: 2605.19886 by Achraf Zinihi, Matthias Ehrhardt.

Figure 1
Figure 1. Figure 1: Flow diagram of the SEIR reaction-diffusion model with vital dynamics. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Schematic diagram of the proposed PINN architecture for the reaction-diffusion SEIR model ( [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Space–time dynamics: Reference data vs PINN prediction; 1D case. [PITH_FULL_IMAGE:figures/full_fig_p013_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Temporal dynamics at fixed spatial locations: Reference data vs PINN prediction; 1D case. [PITH_FULL_IMAGE:figures/full_fig_p014_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: PINN training loss evolution over 8000 epochs; 1D case. [PITH_FULL_IMAGE:figures/full_fig_p014_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Convergence of estimated parameters during PINN training; 1D case. [PITH_FULL_IMAGE:figures/full_fig_p015_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Spatial snapshots: Reference data vs PINN prediction; 2D case. [PITH_FULL_IMAGE:figures/full_fig_p017_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Three-dimensional surface: Reference data vs PINN prediction; 2D case. [PITH_FULL_IMAGE:figures/full_fig_p019_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Pointwise error maps at five time snapshots; 2D case. [PITH_FULL_IMAGE:figures/full_fig_p020_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: PINN training loss evolution over 10000 epochs; 2D case. [PITH_FULL_IMAGE:figures/full_fig_p021_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Convergence of estimated parameters during PINN training; 2D case. [PITH_FULL_IMAGE:figures/full_fig_p021_11.png] view at source ↗
read the original abstract

Reaction-diffusion epidemic models with vital dynamics are an important framework for describing the spatial and temporal spread of infectious diseases. In this work, we present a constraint-aware, physics-informed neural network (PINN) approach to an SEIR reaction-diffusion system with homogeneous Neumann boundary conditions. Due to the scarcity of spatial epidemiological datasets, we generate synthetic benchmark data using structure-preserving implicit-explicit nonstandard finite difference (NSFD) schemes that ensure positivity, boundedness, and numerical stability. The PINN framework integrates PDE residuals, observational data, boundary conditions, and epidemiological constraints within a unified optimization procedure. Specifically, the loss function incorporates the non-negativity of compartment populations and the admissibility of epidemiological parameters. We apply the method to forward simulation and inverse parameter estimation in one- and two-dimensional settings. Numerical experiments demonstrate the framework's ability to accurately reconstruct spatiotemporal epidemic dynamics and reliably identify parameters, even when data is sparse or noisy. These results underscore the potential of constraint-aware PINNs as a robust, data-driven methodology for spatial epidemic modeling.

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

2 major / 0 minor

Summary. The manuscript presents a constraint-aware physics-informed neural network (PINN) for an SEIR reaction-diffusion system with vital dynamics and homogeneous Neumann boundary conditions. Synthetic benchmark data is generated using structure-preserving implicit-explicit nonstandard finite difference (NSFD) schemes that preserve positivity, boundedness, and stability. The PINN loss incorporates PDE residuals, observational data, boundary conditions, and constraints on non-negativity of populations and admissibility of epidemiological parameters. The work applies this to forward simulation and inverse parameter estimation in 1D and 2D, claiming accurate reconstruction of dynamics and reliable parameter identification even with sparse or noisy data.

Significance. If substantiated, the approach could advance data-driven modeling of spatial epidemics by enforcing physical and epidemiological constraints in PINNs, particularly useful given the scarcity of spatial data. The integration of NSFD-generated data is noted as a potential strength for providing structure-preserving benchmarks, but requires verification of fidelity to the continuous PDE solutions.

major comments (2)
  1. The abstract asserts 'accurate reconstruction' and 'reliable identification' of parameters but provides no quantitative metrics, error bars, implementation details, ablation studies, or comparisons with other methods. This lack of evidence is load-bearing for the central claim of the framework's ability to handle sparse or noisy data.
  2. The method relies on synthetic data from NSFD schemes as ground truth for training and validation. However, the abstract supplies no convergence rates, consistency proofs, or cross-comparisons with other discretizations (e.g., standard finite differences or finite elements) to confirm that NSFD solutions faithfully approximate the true weak solutions of the SEIR PDE system. This assumption is critical for the validity of the numerical experiments.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments on our manuscript. We respond to each major comment below and indicate the revisions we will make.

read point-by-point responses

  1. Referee: The abstract asserts 'accurate reconstruction' and 'reliable identification' of parameters but provides no quantitative metrics, error bars, implementation details, ablation studies, or comparisons with other methods. This lack of evidence is load-bearing for the central claim of the framework's ability to handle sparse or noisy data.

    Authors: We agree that the abstract, as a concise summary, does not include specific quantitative metrics or supporting details. The numerical experiments in the manuscript provide the evidence for these claims through error analysis and robustness tests under sparse and noisy conditions. To directly address the concern, we will revise the abstract to incorporate key quantitative highlights from the experiments. revision: yes

  2. Referee: The method relies on synthetic data from NSFD schemes as ground truth for training and validation. However, the abstract supplies no convergence rates, consistency proofs, or cross-comparisons with other discretizations (e.g., standard finite differences or finite elements) to confirm that NSFD solutions faithfully approximate the true weak solutions of the SEIR PDE system. This assumption is critical for the validity of the numerical experiments.

    Authors: The NSFD schemes are selected for their established structure-preserving properties. We acknowledge that the abstract does not detail convergence behavior or comparisons. The manuscript describes the generation of synthetic data, and we will add explicit convergence rates from refinement studies along with cross-comparisons to standard discretizations in the revised version to substantiate the fidelity of the benchmark data. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The abstract presents an independent workflow: NSFD schemes are used to generate synthetic benchmark data (ensuring positivity and stability), after which the constraint-aware PINN is trained by minimizing a loss that combines PDE residuals, observational data, boundary conditions, and epidemiological constraints. Numerical experiments then evaluate reconstruction accuracy and parameter recovery on this externally generated data, including under sparsity and noise. No equations, self-citations, or ansatzes are provided that would reduce the claimed reconstruction or parameter identification to a tautology or fitted input by construction. The PINN optimization and the NSFD data generation remain distinct, making the overall approach self-contained against the described benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

Assessment is limited to the abstract; the ledger therefore records only the high-level modeling assumptions visible in the summary. The central contribution is the addition of constraint terms rather than new physical entities or axioms.

free parameters (1)
  • Epidemiological rate parameters
    Transmission, progression, and recovery rates are recovered via the inverse problem and therefore fitted to the synthetic observations.
axioms (1)
  • domain assumption The SEIR reaction-diffusion system with vital dynamics and homogeneous Neumann boundary conditions is an appropriate model for spatial epidemic spread.
    This modeling choice is taken as given and underpins both the NSFD data generation and the PINN residual.

pith-pipeline@v0.9.0 · 5688 in / 1326 out tokens · 70401 ms · 2026-05-20T01:39:36.228769+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

  • IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear
    ?
    unclear

    Relation between the paper passage and the cited Recognition theorem.

    The total loss function is defined as a weighted sum of multiple components L_total = ω1 L_PDE + ω2 L_IC + ω3 L_BC + ω4 L_data + ω5 L_constraints ... Non-negativity ... max(0,-Û)^2 ... population balance ... (Ŝ+Ê+Î+Ř - Λ/μ)^2_+

  • IndisputableMonolith/Foundation/Atomicity.lean atomic_tick unclear
    ?
    unclear

    Relation between the paper passage and the cited Recognition theorem.

    Synthetic benchmark data ... structure-preserving implicit-explicit nonstandard finite difference (NSFD) schemes that ensure positivity, boundedness, and numerical stability.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

49 extracted references · 49 canonical work pages

  1. [1]

    W. O. Kermack and A. G. McKendrick. A contribution to the mathematical theory of epidemics.Proc. Royal Soc. A, 115(772):700–721, 1927

    work page 1927

  2. [2]

    H. W. Hethcote. The mathematics of infectious diseases.SIAM Review, 42(4):599–653, 2000. 23

    work page 2000

  3. [3]

    Zinihi, M

    A. Zinihi, M. R. Sidi Ammi, and D. F. M. Torres. Fractional differential equations of a reaction-diffusion SIR model involving the Caputo-fractional time-derivative and a nonlin- ear diffusion operator.Evolut. Eqs. Contr. Theo., 14(5):944–967, 2025

    work page 2025

  4. [4]

    W. Wang, Y. Cai, M. Wu, K. Wang, and Z. Li. Complex dynamics of a reaction-diffusion epidemic model.Nonlin. Anal. Real World Appl., 13(5):2240–2258, 2012

    work page 2012

  5. [5]

    N. Wang, L. Zhang, and Z. Teng. Dynamics in a reaction-diffusion epidemic model via environmental driven infection in heterogenous space.J. Biol. Dynam., 16(1):373–396, 2021

    work page 2021

  6. [6]

    Zinihi, M

    A. Zinihi, M. R. Sidi Ammi, and M. Ehrhardt. Optimal control of a diffusive epidemiolog- ical model involving the Caputo-Fabrizio fractional time-derivative.Part. Diff. Eqs. Appl. Math., 14:101188, 2025

    work page 2025

  7. [7]

    Chang, S

    L. Chang, S. Gao, and Z. Wang. Optimal control of pattern formations for an SIR reaction- diffusion epidemic model.J. Theor. Bio., 536:111003, 2022

    work page 2022

  8. [8]

    J. C. Strikwerda.Finite difference schemes and partial differential equations. Society for Industrial and Applied Mathematics, 2004

    work page 2004

  9. [9]

    Zinihi, M

    A. Zinihi, M. Ehrhardt, and M. R. Sidi Ammi. Spatiotemporal SEIQR epidemic modeling with optimal control for vaccination, treatment, and social measures, 2025. arXiv preprint

    work page 2025

  10. [10]

    H. T. Banks and K. Kunisch.Estimation techniques for distributed parameter systems. Birkhäuser Boston, 1989

    work page 1989

  11. [11]

    Raissi, P

    M. Raissi, P. Perdikaris, and G. E. Karniadakis. Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations.J. Comput. Phys., 378:686–707, 2019

    work page 2019

  12. [12]

    G. E. Karniadakis, I. G. Kevrekidis, L. Lu, P. Perdikaris, S. Wang, and L. Yang. Physics- informed machine learning.Nature Rev. Phys., 3(6):422–440, 2021

    work page 2021

  13. [13]

    Kissas, Y

    G. Kissas, Y. Yang, E. Hwuang, W. R. Witschey, J. A. Detre, and P. Perdikaris. Machine learning in cardiovascular flows modeling: Predicting arterial blood pressure from non- invasive 4D flow MRI data using physics-informed neural networks.Comput. Meth. Appl. Mech. Engrg., 358:112623, 2020

    work page 2020

  14. [14]

    Cuomo, V

    S. Cuomo, V. S. Di Cola, F. Giampaolo, G. Rozza, M. Raissi, and F. Piccialli. Scientific machine learning through physics–informed neural networks: Where we are and what’s next.J. Sci. Comput., 92(3), 2022

    work page 2022

  15. [15]

    J. D. Toscano, V. Oommen, A. J. Varghese, Z. Zou, N. Ahmadi Daryakenari, C. Wu, and G. E. Karniadakis. From PINNs to PIKANs: Recent advances in physics-informed machine learning.Machine Learn. Comput. Sci. Engrg., 1(1), 2025

    work page 2025

  16. [16]

    Zinihi, M

    A. Zinihi, M. R. Sidi Ammi, and A. Bachir. Multi-city modeling of epidemics using a topology-based SIR model: Neural network-enhanced SAIRD model.J. Comput. Sci., 92:102721, 2025

    work page 2025

  17. [17]

    Berkhahn and M

    S. Berkhahn and M. Ehrhardt. A physics-informed neural network to model COVID-19 infection and hospitalization scenarios.Adv. Contin. Discr. Mod., 2022(1), 2022. 24

    work page 2022

  18. [18]

    Millevoi, D

    C. Millevoi, D. Pasetto, and M. Ferronato. A physics-informed neural network approach for compartmental epidemiological models.PLOS Comput. Biol., 20(9):e1012387, 2024

    work page 2024

  19. [19]

    S. P. Nelson, R. Raja, P. Eswaran, J. Alzabut, and G. Rajchakit. Modeling the dynamics of COVID-19 in Japan: Employing data-driven deep learning approach.Int. J. Mach. Learn. Cybern., 2024

    work page 2024

  20. [20]

    Shamsara, F

    E. Shamsara, F. König, and N. Pfeifer. An informed deep learning model of the Omicron wave and the impact of vaccination.Comput. Biol. Med., 191:109968, 2025

    work page 2025

  21. [21]

    A. Zinihi. Identifying memory effects in epidemics via a fractional SEIRD model and physics-informed neural networks, 2025. arXiv preprint

    work page 2025

  22. [22]

    C. P. Igiri and S. Shikaa. Monkeypox transmission dynamics using fractional disease informed neural network: A global and continental analysis.IEEE Access, 13:77611–77641, 2025

    work page 2025

  23. [23]

    Heldmann, S

    F. Heldmann, S. Berkhahn, M. Ehrhardt, and K. Klamroth. PINN training using biob- jective optimization: The trade-off between data loss and residual loss.J. Comput. Phys., 488:112211, 2023

    work page 2023

  24. [24]

    Cheng, Y

    H. Cheng, Y. Mao, and X. Jia. A framework based on physics-informed graph neural ODE: For continuous spatial-temporal pandemic prediction.Appl. Intell., 54(24):12661–12675, 2024

    work page 2024

  25. [25]

    Liu and A

    D. Liu and A. Sopasakis. A combined neural ODE-Bayesian optimization approach to resolve dynamics and estimate parameters for a modified SIR model with immune memory. Heliyon, 10(19):e38276, 2024

    work page 2024

  26. [26]

    B. Shan, Y. Li, and S.-J. Huang. VI-PINNs: Variance-involved physics-informed neural networks for fast and accurate prediction of partial differential equations.Neurocomputing, 623:129360, 2025

    work page 2025

  27. [27]

    W. G. Madden, W. Jin, B. Lopman, A. Zufle, B. Dalziel, C. Jessica E. Metcalf, B. T. Gren- fell, and M. S. Y. Lau. Deep neural networks for endemic measles dynamics: Comparative analysis and integration with mechanistic models.PLOS Comput. Biol., 20(11):e1012616, 2024

    work page 2024

  28. [28]

    X. Chen, F. Li, H. Lian, and P. Wang. A deep learning framework for predicting the spread of diffusion diseases.Electronic Research Archive, 33(4):2475–2502, 2025

    work page 2025

  29. [29]

    Pazy.Semigroups of linear operators and applications to partial differential equations

    A. Pazy.Semigroups of linear operators and applications to partial differential equations. Springer Science & Business Media, 2012

    work page 2012

  30. [30]

    M. H. Maamar, M. Ehrhardt, and L. Tabharit. A nonstandard finite difference scheme for a time-fractional model of Zika virus transmission.Math. Biosci. Engrg., 21(1):924–962, 2024

    work page 2024

  31. [31]

    R. E. Mickens.Nonstandard finite difference models of differential equations. World Sci- entific, 1993

    work page 1993

  32. [32]

    R. E. Mickens.Nonstandard finite difference schemes: Methodology and applications. World Scientific Publishing Company, 2020. 25

    work page 2020

  33. [33]

    Zinihi, M

    A. Zinihi, M. Ehrhardt, and M. R. Sidi Ammi. A nonstandard finite difference scheme for an SEIQR epidemiological PDE model.Appl. Math. Comput., 520:129953, 2026

    work page 2026

  34. [34]

    B. M. Chen-Charpentier and H. V. Kojouharov. An unconditionally positivity pre- serving scheme for advection-diffusion reaction equations.Math. Comput. Model., 57(9–10):2177–2185, 2013

    work page 2013

  35. [35]

    S. A. Pasha, Y. Nawaz, and M. S. Arif. On the nonstandard finite difference method for reaction–diffusion models.Chaos, Solitons & Fractals, 166:112929, 2023

    work page 2023

  36. [36]

    Ehrhardt and R

    M. Ehrhardt and R. E. Mickens. A nonstandard finite difference scheme for convection- diffusion equations having constant coefficients.Appl. Math. Comput., 219(12):6591–6604, 2013

    work page 2013

  37. [37]

    S. Wang, Y. Teng, and P. Perdikaris. Understanding and mitigating gradient pathologies in physics-informed neural networks.SIAM J. Sci. Comput., 43(5):A3055–A3081, 2021

    work page 2021

  38. [38]

    Tancik, P

    M. Tancik, P. P. Srinivasan, B. Mildenhall, S. Fridovich-Keil, N. Raghavan, U. Singhal, R. Ramamoorthi, J. T. Barron, and R. Ng. Fourier features let networks learn high frequency functions in low dimensional domains. InAdvances in Neural Information Pro- cessing Systems (NeurIPS), volume 33, pages 7537–7547, 2020

    work page 2020

  39. [39]

    C. Wang, S. Li, D. He, and L. Wang. IsL2 physics informed loss always suitable for training physics informed neural network?Adv. Neural Inform. Process. Syst., 35:8278–8290, 2022

    work page 2022

  40. [40]

    Glorot and Y

    X. Glorot and Y. Bengio. Understanding the difficulty of training deep feedforward neural networks. InProceedings of the 13th international conference on artificial intelligence and statistics, pages 249–256. JMLR Workshop and Conference Proceedings, 2010

    work page 2010

  41. [41]

    D. C. Liu and J. Nocedal. On the limited memory BFGS method for large scale optimiza- tion.Math. Programm., 45(1-3):503–528, 1989

    work page 1989

  42. [42]

    Liu and H

    P. Liu and H. X. Li. Global behavior of a multi-group SEIR epidemic model with spatial diffusion in a heterogeneous environment.International Journal of Applied Mathematics and Computer Science, 32(2), 2022

    work page 2022

  43. [43]

    CDC COVID data tracker; https://www.cdc

    Centers for Disease Control and Prevention. CDC COVID data tracker; https://www.cdc. gov/covid-data-tracker/index.html, Accessed on: May 11, 2026

    work page 2026

  44. [44]

    Seasonal influenza (Flu), https://www.cdc

    Centers for Disease Control and Prevention. Seasonal influenza (Flu), https://www.cdc. gov/flu/index.htm, Accessed on: May 11, 2026

    work page 2026

  45. [45]

    Zinihi, M

    A. Zinihi, M. Ehrhardt, and M. R. Sidi Ammi. A Koopman operator framework for non- linear epidemic dynamics: Application to an SIRSD model.Nonlinear Science, 6:100095, April 2026

    work page 2026

  46. [46]

    Epidemiology of COVID-19 in England: January 2020 to December 2024; https://www

    UK Health Security Agency. Epidemiology of COVID-19 in England: January 2020 to December 2024; https://www. gov.uk/government/publications/epidemiology-of-covid-19-in-england/ epidemiology-of-covid-19-in-england-january-2020-to-december-2024, Accessed on: May 10, 2026. 26

    work page 2020

  47. [47]

    COVID-19 epidemiological up- date, 24 December 2024; https://www.who.int/publications/m/item/ covid-19-epidemiological-update---24-december-2024, Accessed on: May 10, 2026

    World Health Organization. COVID-19 epidemiological up- date, 24 December 2024; https://www.who.int/publications/m/item/ covid-19-epidemiological-update---24-december-2024, Accessed on: May 10, 2026

    work page 2024

  48. [48]

    COVID-19 dashboard; https://coronavirus.jhu.edu/data, Ac- cessed on: May 11, 2026 (2020)

    Johns Hopkins University. COVID-19 dashboard; https://coronavirus.jhu.edu/data, Ac- cessed on: May 11, 2026 (2020)

    work page 2026

  49. [49]

    Seasonal influenza data; https: //www.ecdc.europa.eu/en/seasonal-influenza, Accessed on: May 11, 2026

    European Centre for Disease Prevention and Control. Seasonal influenza data; https: //www.ecdc.europa.eu/en/seasonal-influenza, Accessed on: May 11, 2026. 27

    work page 2026