CoChat
Sign Up
Cited by 0Viral Infectious Diseases and Gene Expression in InsectsVirus-based gene therapy researchFluid Dynamics and Mixing

Mechanistic model for HEK293 viral vector processes and its application in a digital shadow framework

Open access
Kuchemüller, Kim B., Austerjost, Jonas, Machleid, Rafael, Roßkamp, Alena, Niemann, Julia, McCready, Chris·2026·Bioprocess and Biosystems Engineering
DOI 10.1007/s00449-026-03362-2PMC PMC13582380
Open this paper in CoChat — free accountOpen PDF

Abstract

A mechanistic model describing HEK293 cell growth and viral vector production was developed and calibrated for adenoviral processes and subsequently transferred to adeno‑associated and lentiviral cultures. The model represents key cellular states-viable, dead, and lysed cells-linked through substrate consumption, metabolic inhibition, and virus‑specific toxicity terms. By adjusting a limited set of parameters, the same mechanistic structure, combined with vector specific parameter sets, could accurately reproduce distinct infection and transfection dynamics, demonstrating cross‑platform applicability. As a demonstrator for real‑time deployment, the mechanistic framework was combined with an orthogonal projections to latent structures model based on dielectric spectroscopy data to form a Digital Shadow. The soft‑sensor layer provided real‑time viable cell density values that continuously updated the mechanistic simulation, yielding online predictions of culture growth and infection timing. Validation with adenoviral perfusion datasets confirmed high agreement between simulated and measured viable cell concentrations. This study demonstrates that transferable mechanistic models can serve as a robust foundation for hybrid digital systems in bioprocessing. The proposed Digital Shadow exemplifies how such models can be connected to inline monitoring tools for enhanced process understanding and decision support, paving the way toward closed‑loop Digital Twin applications in viral vector manufacturing.

AI snapshot

Written from the paper’s full text.

TL;DR

A single mechanistic model structure was calibrated for HEK293 adenoviral vector production and then transferred to AAV and lentiviral processes by changing only parameter sets. Coupled with an OPLS soft sensor from capacitance spectroscopy, it formed a Digital Shadow that updated viable cell density predictions every 24 h and forecast infection timing.

Key findings

  • The same mechanistic model reproduced AdV, AAV, and LV processes without structural modification, using 40 cultivations: 13 AdV perfusion, 16 AAV fed-batch, and 11 LV batch runs.
  • Monte Carlo Latin Hypercube Sampling with ±5% parameter variation showed predicted viral titers remained within ±15% of the mean; growth rate, infection toxicity, and virus production were the most sensitive parameters.
  • OPLS soft-sensor performance for viable cell density: pre-infection model R²CV 0.98, RMSECV 3.4e6 cells/mL, RMSEP 4.2e6 cells/mL; post-infection R²CV 0.71, RMSECV 2.6e6, RMSEP 3.8e6; global R²CV 0.92, RMSECV 3.6e6, RMSEP 5.8e6.
  • In the Digital Shadow, real-time VCD estimates updated the mechanistic model every 24 h; validation with adenoviral perfusion datasets confirmed high agreement between simulated and measured viable cell concentrations, though early online cell counts were initially too low to immediately predict optimal infection timing.

Methods

  • Mechanistic ODE model tracking viable, dead, and lysed cells, with substrate consumption, metabolic inhibition, quadratic process effects, infection/transfection toxicity, virus production, and virus degradation; solved with RK45 and parameterized via Nelder-Mead in Python 3.12.10.
  • Calibration used AdV Run 1 (N = 12), AAV runs (N = 16), and LV runs (N = 11); AdV Run 2 (N = 1) was withheld as an external prediction dataset before bench-scale parameter refinement.
  • OPLS models were built from capacitance spectroscopy at 25 frequencies (50–20,000 kHz) using SIMCA® 18, then deployed via Node-RED and OPC UA into a SCADA system; the Digital Shadow updated the mechanistic simulation every 24 h.

Limitations

  • The paper states that titers had only two data points per experiment, restricting full depiction of actual behavior; initial OPLS predictions were imprecise and online simulated cell counts were initially too low, hindering immediate infection-timing prediction.
  • The paper states no formal multi-start strategy was implemented, so additional local minima cannot be fully excluded; the Digital Shadow was unidirectional and did not allow real-time control, and moving to a Digital Twin would require larger datasets for validation.
  • The text makes evident that AdV Run 2 was N = 1 and required a subset of parameters to be refined for bench scale, and that broader uncertainty in biomass during late infection indicated missing process variables such as proteolytic degradation or dynamic pH effects.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1007/s00449-026-03362-2

Cites

34 of 34 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

10 of 40 cited sources checked · 2 supported · 0 flagged · accuracy not yet measured

  • Interpretationcitation

    The significant shift in glucose consumption following viral infection highlights profound metabolic changes induced in host cells.

    “The significant shift in glucose consumption following viral infection highlights the profound metabolic changes induced in host cells [31].”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [31] Sumbria D, Berber E, Mathayan M, Rouse BT (2020) Virus infections and host metabolism-can we manage the interactions? Front Immunol 11:594963. https://doi.org/10.3389/fimmu.2020.594963. (PMID: 33613518) source
  • Findingnone given

    In this study, a mechanistic model was established to unify the description of diverse viral vector processes in HEK293 cells within a single structural framework.

    “In this study, a mechanistic model was established to unify the description of diverse viral vector processes in HEK293 cells within a single structural framework.”
    Quote in the paperSubstantial claim
  • Findingown data

    The mechanistic framework was combined with an OPLS model based on dielectric spectroscopy data to form a Digital Shadow.

    “As a demonstrator for real‑time deployment, the mechanistic framework was combined with an orthogonal projections to latent structures model based on dielectric spectroscopy data to form a Digital Shadow.”
    Quote in the paperSubstantial claim
  • Findingown data

    Validation with adenoviral perfusion datasets confirmed high agreement between simulated and measured viable cell concentrations.

    “Validation with adenoviral perfusion datasets confirmed high agreement between simulated and measured viable cell concentrations.”
    Quote in the paperSubstantial claim
  • Findingnone given

    The model represents viable, dead, and lysed cell states linked through substrate consumption, metabolic inhibition, and virus-specific toxicity terms.

    “The model represents key cellular states—viable, dead, and lysed cells—linked through substrate consumption, metabolic inhibition, and virus‑specific toxicity terms.”
    Quote in the paperSubstantial claim
  • May combine two claims. 23
  • Classified as probably not a substantial claim. 26

  • May combine two claims.

    Premisecitation

    Most existing viral vector models are vector specific and typically require re-derivation or extensive re-parameterization when applied to other viral systems.

    “Most existing models are vector specific and typically require re-derivation or extensive re-parameterization when applied to other viral systems.”
    Quote in the paperSubstantial claim3 of 7 sources checked
    • cites [14] Gilbert P-A, Kamen A, Bernier A, Garnier A (2007) A simple macroscopic model for the diffusion and adsorption kinetics of r-adenovirus. Biotechnol Bioeng 98(1):239–251. https://doi.org/10.1002/bit.21340. (PMID: 17597100) source
    • cites [15] Dinh A-T, Theofanous T, Mitragotri S (2005) A model for intracellular trafficking of adenoviral vectors. Biophys J 89(3):1574–1588. https://doi.org/10.1529/biophysj.105.059477. (PMID: 15980163) source

      The Introduction and “Pharmacokinetic model versus current approach” section discuss limitations of earlier models, including simplified transport and fitted rate constants. The Conclusion says the authors’ approach could be applied to other viruses with appropriate modifications. None of these passages says that most existing models are vector specific or typically require re-derivation or extensive re-parameterization for other viral systems. (LLM, conf=0.72)

      Experimental — accuracy not yet measured
    • cites [5] Joiner J, Huang Z, McHugh K et al (2022) Process modeling of recombinant adeno-associated virus production in HEK293 cells. Curr Opin Chem Eng 36:100823. https://doi.org/10.1016/j.coche.2022.100823 source
    • cites [16] Nguyen TNT, Park D, Canova CT et al (2025) Perfusion-based production of rAAV via an intensified transient transfection process. Biotechnol Bioeng 122(6):1424–1440. https://doi.org/10.1002/bit.28967. (PMID: 40103325) source

      The Discussion (page unavailable) says this rAAV model can be adapted to different serotypes and production methods through mathematical modifications and reparameterization. It does not establish that most existing models are vector-specific or typically require extensive reparameterization or re-derivation for other viral systems. (LLM, conf=0.82)

      Experimental — accuracy not yet measured
    • cites [17] Nguyen TNT, Sha S, Hong MS et al (2021) Mechanistic model for production of recombinant adeno-associated virus via triple transfection of HEK293 cells. Mol Ther Methods Clin Dev 21:642–655. https://doi.org/10.1016/j.omtm.2021.04.006. (PMID: 34095346) source

      In Materials and methods (Mathematical modeling; page unavailable), the paper says parameters from existing gene-delivery models did not fit its cell line or time course, so it re-estimated kinetic parameters. It does not say that most viral-vector models are vector specific or that applying them to other viral systems typically requires re-derivation or extensive re-parameterization. The Discussion (Conclusion; page unavailable) instead says its rAAV model can be adapted to other production methods and serotypes with small modifications. (LLM, conf=0.78)

      Experimental — accuracy not yet measured
    • cites [18] Roda WC, Li MY, Akinwumi MS et al (2017) Modeling brain lentiviral infections during antiretroviral therapy in AIDS. J Neurovirol 23(4):577–586. https://doi.org/10.1007/s13365-017-0530-3. (PMID: 28512685) source
    • cites [19] Cruz PE, Almeida JS, Murphy PN, Moreira JL, Carrondo MJ (2000) Modeling retrovirus production for gene therapy. 1. Determination of optimal bioreaction mode and harvest strategy. Biotechnol Prog 16(2):213–221. https://doi.org/10.1021/bp9901466 source

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Richelle et al. (2025) established a hybrid Digital Twin framework for CHO cell cultivation by integrating mechanistic and data driven modeling, achieving accurate predictions of viable cell density, product titer, and metabolite concentration.

    “Similarly, Richelle et al. (2025) established a hybrid DT framework for CHO cell cultivation by integrating mechanistic and data driven modeling, achieving accurate predictions of viable cell density, product titer, and metabolite concentration [12].”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [12] Richelle A, Andersson D, Antonakoudis A et al (2025). A hybrid modeling framework for predictive digital twins of CHO cell culture. https://doi.org/10.1101/2025.11.24.690194 source

      Split into 2 atoms — 1 confirmed, 1 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.

      Experimental — accuracy not yet measured

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Egan et al. (2025) reported a Digital Shadow for CAR T cell expansion in perfusion culture, enabling soft sensor estimation of cell concentration and optimal harvest timing with a mean relative error of 13%.

    “Egan et al. (2025) reported a DS for CAR T cell expansion in perfusion culture, enabling soft sensor estimation of cell concentration and optimal harvest timing with a mean relative error of 13% [11].”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [11] Egan JR, Marí-Buyé N, Benítez-Cano EV et al (2025) A digital shadow of CAR T cell expansion in a perfusion bioreactor: informing optimal harvest times for autologous cell therapy. Biotechnol Prog. https://doi.org/10.1002/btpr.70045 source

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    For AAV, several mechanistic frameworks describe triple transfection dynamics, cell growth, metabolism, genome replication, and capsid formation.

    “For AAV, several mechanistic frameworks describe triple transfection dynamics, cell growth, metabolism, genome replication, and capsid formation [5, 16, 17].”
    Quote in the paperNot a substantial claim2 of 3 sources checked
    • cites [5] Joiner J, Huang Z, McHugh K et al (2022) Process modeling of recombinant adeno-associated virus production in HEK293 cells. Curr Opin Chem Eng 36:100823. https://doi.org/10.1016/j.coche.2022.100823 source
    • cites [16] Nguyen TNT, Park D, Canova CT et al (2025) Perfusion-based production of rAAV via an intensified transient transfection process. Biotechnol Bioeng 122(6):1424–1440. https://doi.org/10.1002/bit.28967. (PMID: 40103325) source

      The Materials and Methods section describes a mechanistic triple-transfection model that includes cell growth, viral DNA replication, and capsid formation. The Results report metabolite measurements, but the modeling section uses dead-cell concentration as a surrogate for harmful species rather than describing a mechanistic model of metabolism. No page numbers are available in our copy. (LLM, conf=0.85)

      Experimental — accuracy not yet measured
    • cites [17] Nguyen TNT, Sha S, Hong MS et al (2021) Mechanistic model for production of recombinant adeno-associated virus via triple transfection of HEK293 cells. Mol Ther Methods Clin Dev 21:642–655. https://doi.org/10.1016/j.omtm.2021.04.006. (PMID: 34095346) source

      The Introduction and Materials and methods describe a mechanistic model of triple transfection, viral DNA replication, and capsid formation. Materials and methods also includes a cell-growth parameter for dilution during cell division, but the paper does not model metabolism. The sentence cites three sources collectively, so this verdict concerns only what this paper supports. (LLM, conf=0.87)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Premisecitation

    The terms used in this model (Eqs. 6-8) are adapted from structured CHO cell models developed in previous work of our group, where they have been shown to yield stable and biologically meaningful kinetics.

    “The terms used in this model (Eqs. 6–8) are adapted from structured CHO cell models developed in previous work of our group [8, 9], where they have been shown to yield stable and biologically meaningful kinetics.”
    Quote in the paperSubstantial claim2 of 2 sources checked
    • cites [8] Agarwal P, McCready C, Ng SK et al (2025) Hybrid modeling for in silico optimization of a dynamic perfusion cell culture process. Biotechnol Prog 41(1):e3503. https://doi.org/10.1002/btpr.3503. (PMID: 39291457) source

      The Growth model section (page unavailable) says Equations 6–8 use an approach from Richelle et al., and Model fitting reports that the predicted growth and death rates relate well to observations. The copy does not show that Equations 6–8 had been demonstrated to yield stable kinetics; its reference to stable simulations concerns a different modeling choice. (LLM, conf=0.85)

      Experimental — accuracy not yet measured
    • cites [9] Richelle A, Corbett B, Agarwal P, Vernersson A, Trygg J, McCready C (2022) Model-based intensification of CHO cell cultures: One-step strategy from fed-batch to perfusion. Front Bioeng Biotechnol 10:948905. https://doi.org/10.3389/fbioe.2022.948905. (PMID: 36072286) source

      Section 4.2 says the model shares a general cell-population formalism with existing models, and Section 4.1 reports that it captures viable-cell density well. Our copy does not say its terms were adapted from earlier structured CHO models by the authors’ group or that those earlier models yielded stable, biologically meaningful kinetics. (LLM, conf=0.78)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Findingown data

    By adjusting a limited set of parameters, the same mechanistic structure, combined with vector-specific parameter sets, could accurately reproduce distinct infection and transfection dynamics, demonstrating cross-platform applicability.

    “By adjusting a limited set of parameters, the same mechanistic structure, combined with vector specific parameter sets, could accurately reproduce distinct infection and transfection dynamics, demonstrating cross‑platform applicability.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    A mechanistic model describing HEK293 cell growth and viral vector production was developed and calibrated for adenoviral processes and subsequently transferred to adeno-associated and lentiviral cultures.

    “A mechanistic model describing HEK293 cell growth and viral vector production was developed and calibrated for adenoviral processes and subsequently transferred to adeno‑associated and lentiviral cultures.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingreasoning

    While the structural model formulation is identical across all vector systems, parameter sets differ between AdV, AAV, and LV to reflect their distinct biological replication mechanisms and metabolic burdens.

    “While the structural model formulation is identical across all vector systems, parameter sets differ between AdV, AAV, and LV to reflect their distinct biological replication mechanisms and metabolic burdens.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingnone given

    The model is designed to represent infection and transfection dynamics through the same structural framework, enabling flexible adaptation by parameter calibration rather than structural modification.

    “The model is designed to represent infection and transfection dynamics through the same structural framework, enabling flexible adaptation by parameter calibration rather than structural modification.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingreasoning

    The soft-sensor layer provided real-time viable cell density values that continuously updated the mechanistic simulation, yielding online predictions of culture growth and infection timing.

    “The soft‑sensor layer provided real‑time viable cell density values that continuously updated the mechanistic simulation, yielding online predictions of culture growth and infection timing.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The global OPLS model had R2CV 0.92, RMSECV 3.6e6 cells/mL, and RMSEP 5.8e6 cells/mL on the external test set.

    “| Global | 1 + 3 | 0.92 | 3.6 | 5.8 |”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingnone given

    The model captures growth, death, and lysis kinetics, substrate and inhibitor dependencies, and the dynamics of infection or transfection through differential activation of process variables.

    “The model captures growth, death, and lysis kinetics, substrate and inhibitor dependencies, and the dynamics of infection or transfection through differential activation of process variables.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    The post-infection OPLS model had R2CV 0.71, RMSECV 2.6e6 cells/mL, and RMSEP 3.8e6 cells/mL on the external test set.

    “| Post-infection | 1 + 0 | 0.71 | 2.6 | 3.8 |”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingnone given

    kGlc was only relevant for AdV Run 1.

    “kGlc was only relevant for AdV Run 1”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    In total, 40 independent cultivations were available: 13 AdV perfusion runs, 16 AAV fed batch runs, and 11 LV batch runs.

    “In total, 40 independent cultivations were available: 13 AdV perfusion runs, 16 AAV fed batch runs, and 11 LV batch runs.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Findingnone given

    By adjusting only a limited number of parameters, distinct process types can be represented without changing the core structure.

    “By adjusting only a limited number of parameters, distinct process types can be represented without changing the core structure.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingreasoning

    The simulation based on Run 1 parameters systematically underestimated the growth rate and overpredicted metabolic inhibition when applied to Run 2.

    “The simulation based on these parameters systematically underestimated the growth rate and overpredicted metabolic inhibition”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Findingnone given

    The LV datasets derived from batch processes with controlled changes in dissolved oxygen, medium composition, pH, and agitation rate.

    “the LV datasets derived from batch processes with controlled changes in dissolved oxygen, medium composition, pH, and agitation rate.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data

    The process model describes varying initial cultivation conditions.

    “the process model describes varying initial cultivation conditions (Fig. 3, A)”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Findingnone given

    The parameter infection/transfection toxicity (ki/t, toxic) indicates the degree of toxicity that the biomaterial exerts on the death rate.

    “The parameter infection/transfection toxicity (ki/t, toxic) indicates the degree of toxicity that the biomaterial exerts on the death rate.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingnone given

    The model considers metabolic byproducts via a biomaterial variable (Φbio) which impacts growth.

    “Additionally, the model considers metabolic byproducts via a biomaterial variable (Φbio, Eq. 4), which impacts growth.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    The pre-infection OPLS model had R2CV 0.98, RMSECV 3.4e6 cells/mL, and RMSEP 4.2e6 cells/mL on the external test set.

    “| Pre-infection | 1 + 0 | 0.98 | 3.4 | 4.2 |”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Interpretationreasoning

    The proposed Digital Shadow exemplifies how such models can be connected to inline monitoring tools for enhanced process understanding and decision support.

    “The proposed Digital Shadow exemplifies how such models can be connected to inline monitoring tools for enhanced process understanding and decision support”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Such mechanistic models have successfully supported monoclonal antibody production and scale-up processes in mammalian systems.

    “Such models have successfully supported monoclonal antibody production and scale up processes in mammalian systems [8, 9].”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [8] Agarwal P, McCready C, Ng SK et al (2025) Hybrid modeling for in silico optimization of a dynamic perfusion cell culture process. Biotechnol Prog 41(1):e3503. https://doi.org/10.1002/btpr.3503. (PMID: 39291457) source

      Checked before — its earlier verdict is held back until a new check runs.

    • cites [9] Richelle A, Corbett B, Agarwal P, Vernersson A, Trygg J, McCready C (2022) Model-based intensification of CHO cell cultures: One-step strategy from fed-batch to perfusion. Front Bioeng Biotechnol 10:948905. https://doi.org/10.3389/fbioe.2022.948905. (PMID: 36072286) source

      Checked before — its earlier verdict is held back until a new check runs.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Recent work on machine learning supported model predictive control illustrates how hybrid frameworks can improve bioreactor operation and real time optimization in mammalian cell processes.

    “Recent work on machine learning supported model predictive control further illustrates how hybrid frameworks can improve bioreactor operation and real time optimization in mammalian cell processes [13].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [13] Dan A, Liu B, Patil U et al (2025) Machine learning model-based design and model predictive control of a bioreactor for the improved production of mammalian cell-based bio-therapeutics. Control Eng Pract 156:106198. https://doi.org/10.1016/j.conengprac.2024.106198 source

  • Classified as probably not a substantial claim.

    Premisecitation

    Mechanistic models form the foundational layer of both Digital Twins and Digital Shadows, bridging data analytics and biological interpretation.

    “Mechanistic models form the foundational layer of both concepts, bridging data analytics and biological interpretation (Fig. 2) [7].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [7] Schmidt A, Lütge J, Uhl A, Köster D, Strube J (2025) Business cases for digital twins in biopharmaceutical manufacturing—market overview, stakeholders, technologies in 2025 and beyond. Processes 13(5):1498. https://doi.org/10.3390/pr13051498 source

  • May combine two claims.

    Interpretationreasoning

    The study establishes a unified mechanistic model for HEK293-based viral vector production and demonstrates its practical deployment within a digital shadow environment for real-time monitoring.

    “the study establishes a unified mechanistic model for HEK293-based viral vector production and demonstrates its practical deployment within a DS environment for real-time monitoring.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Premisecitation

    Lentiviral systems have been investigated through dynamic studies quantifying vector secretion, production kinetics, and metabolic coupling in HEK293T producer lines, providing the basis for kinetic and process level modeling.

    “Lentiviral systems have been investigated through dynamic studies quantifying vector secretion, production kinetics, and metabolic coupling in HEK293T producer lines, providing the basis for kinetic and process level modeling [18, 19].”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [18] Roda WC, Li MY, Akinwumi MS et al (2017) Modeling brain lentiviral infections during antiretroviral therapy in AIDS. J Neurovirol 23(4):577–586. https://doi.org/10.1007/s13365-017-0530-3. (PMID: 28512685) source
    • cites [19] Cruz PE, Almeida JS, Murphy PN, Moreira JL, Carrondo MJ (2000) Modeling retrovirus production for gene therapy. 1. Determination of optimal bioreaction mode and harvest strategy. Biotechnol Prog 16(2):213–221. https://doi.org/10.1021/bp9901466 source

  • Classified as probably not a substantial claim.

    Premisecitation

    The mathematical model, described in detail in [8, 9], explains cell growth in three phases: viable, dead, and lysed cells.

    “The mathematical model, described in detail in [8, 9], explains the cell growth in three phases: viable (xVCD, Eq. 1), dead (xDCD, Eq. 2), and lysed cells (xLCD, Eq. 3)”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [8] Agarwal P, McCready C, Ng SK et al (2025) Hybrid modeling for in silico optimization of a dynamic perfusion cell culture process. Biotechnol Prog 41(1):e3503. https://doi.org/10.1002/btpr.3503. (PMID: 39291457) source

      Checked before — its earlier verdict is held back until a new check runs.

    • cites [9] Richelle A, Corbett B, Agarwal P, Vernersson A, Trygg J, McCready C (2022) Model-based intensification of CHO cell cultures: One-step strategy from fed-batch to perfusion. Front Bioeng Biotechnol 10:948905. https://doi.org/10.3389/fbioe.2022.948905. (PMID: 36072286) source

      Checked before — its earlier verdict is held back until a new check runs.

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing). May combine two claims.

    Interpretationreasoning

    This configuration exemplifies a stepping stone toward Digital Twin implementations, combining the explanatory power of mechanistic modeling with the responsiveness of data driven monitoring tools.

    “This configuration exemplifies a stepping stone toward DT implementations, combining the explanatory power of mechanistic modeling with the responsiveness of data driven monitoring tools.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Premisenone given

    While hybrid frameworks have emerged successfully for CHO and CAR T systems, their application to viral vector manufacturing remains limited.

    “While such hybrid frameworks have emerged successfully for CHO and CAR T systems, their application to viral vector manufacturing remains limited.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationnone given

    The higher Φbio value suggests weaker inhibition once a threshold of 50 was reached, whereas under other conditions, toxic effects occurred even at lower values.

    “The higher Φbio value suggests weaker inhibition once a threshold of 50 was reached, whereas under other conditions, toxic effects occurred even at lower values.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Study design and conductreasoning

    These adaptations were limited to the scaling context and did not alter the underlying mechanistic model structure.

    “These adaptations were limited to the scaling context and did not alter the underlying mechanistic model structure.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Premisenone given

    LHS is a statistical method that ensures uniform sampling across the range of each parameter, offering efficient coverage compared to simple random sampling.

    “LHS is a statistical method that ensures uniform sampling across the range of each parameter, offering efficient coverage compared to simple random sampling.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    The ability to reuse the same structural model substantially reduces model-development time and facilitates comparative analyses between viral production modes.

    “The ability to reuse the same structural model substantially reduces model‑development time and facilitates comparative analyses between viral production modes.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Study design and conductown data

    Although no formal multi-start strategy was implemented, parameter optimization was repeated with perturbed initial values, yielding consistently similar solutions, indicating that the results were robust within the explored parameter space, although the possibility of additional local minima cannot be fully excluded.

    “Although no formal multi-start strategy was implemented, parameter optimization was repeated with perturbed initial values, yielding consistently similar solutions. This indicates that the results were robust within the explored parameter space, although the possibility of additional local minima cannot be fully excluded.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Metabolic flux analysis has provided important insights into HEK293 metabolic limitations during viral vector production.

    “Additionally, metabolic flux analysis has provided important insights into HEK293 metabolic limitations during viral vector production [20].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [20] Gurazada SGR, Kennedy HM, Braatz RD, Mehrman SJ, Polson SW, Rombel IT (2025) HEK-omics: The promise of omics to optimize HEK293 for recombinant adeno-associated virus (rAAV) gene therapy manufacturing. Biotechnol Adv 79:108506. https://doi.org/10.1016/j.biotechadv.2024.108506. (PMID: 39708987) source

      Checked before — its earlier verdict is held back until a new check runs.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductown data

    Limitations exist due to having only two data points per experiment, which restricts the full depiction of actual behavior.

    “though limitations exist due to having only two data points per experiment, which restricts the full depiction of actual behavior.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    The distribution of the model simulations encompasses process knowledge, whereby broad or overlapping distributions may indicate uncharacterized effects, while narrower distributions usually represent well-understood phenomena.

    “The distribution of the model simulations encompasses process knowledge, whereby broad or overlapping distributions may indicate uncharacterized effects, while narrower distributions usually represent well-understood phenomena [24].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [24] Kuchemüller KB, Pörtner R, Möller J (2023) Design of cell expansion processes for adherent-growing cells with mDoE-workflow. Eng Life Sci 23(5):e2200059. https://doi.org/10.1002/elsc.202200059. (PMID: 37153028) source

      Checked before — its earlier verdict is held back until a new check runs.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Interpretationnone given

    The model is equipped to handle simulations involving boluses or dilutions.

    “It is also equipped to handle simulations involving boluses or dilutions”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Study design and conductnone given

    AdV Run 2 (N = 1) was initially withheld from calibration and used as an external prediction dataset to evaluate scale transferability.

    “AdV Run 2 (N = 1) was initially withheld from calibration and used as an external prediction dataset to evaluate scale transferability.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    A comparable integration concept using Node-RED and OPC UA for real-time soft-sensor deployment in an AAV production process has previously been demonstrated by Machleid et al. (2025), establishing the feasibility of this framework for data-driven model execution in bioprocessing environments.

    “A comparable integration concept using Node‑RED and OPC UA for real‑time soft‑sensor deployment in an AAV production process has previously been demonstrated by Machleid et al. (2025) establishing the feasibility of this framework for data‑driven model execution in bioprocessing environments [30].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [30] Machleid R, Nunna S, George A, Austerjost J, Tomala M, Surowiec I (2025) Real-time VCC monitoring and forecasting in HEK-cell-based rAAV vector production using capacitance spectroscopy. Eng Life Sci 25(2):e70004. https://doi.org/10.1002/elsc.70004. (PMID: 39927211) source

      Checked before — its earlier verdict is held back until a new check runs.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Interpretationreasoning

    The higher ki/t, toxic in AdV Run 2 may indicate stronger toxic effects or interactions during this run, possibly influenced by variations in toxic biomaterial and production rates.

    “which may indicate stronger toxic effects or interactions during this run, possibly influenced by the variations in toxic biomaterial and production rates.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Interpretationreasoning

    The increase in μmax is likely due to improved mixing and mass-transfer conditions at larger scale.

    “likely due to improved mixing and mass‑transfer conditions at larger scale.”
    Quote in the paperNot a substantial claim
44 of 44

  • The authors state this as an assumption.

    Before infection/transfection, the typical growth conditions as shown above can be assumed.

    “Before infection/transfection, the typical growth conditions as shown above can be assumed.”

  • The paper relies on this without support.

    In HEK infection/transfection processes, viral vectors are produced inside the cells and are harvested 48–72 h post-infection/transfection.

    “In HEK infection/transfection processes, viral vectors are produced inside the cells and are harvested 48–72-h post-infection/transfection.”

  • The paper relies on this without support.

    Infection itself triggers an inhibition effect on cell growth.

    “Here, infection itself triggers an inhibition effect on cell growth, which means it is implemented as a variable into Eq. 7.”
3 of 3

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Figure 4 calls percentile curves standard-deviation curves

    Minor

    The caption identifies the plotted curves as a mean and standard deviations, then identifies them as a mean and the 10th and 90th percentiles. Percentile curves are not standard-deviation curves.

1 of 1

Cites

Flagged first, then by citations

Research works printed but not linked: 1

  1. Howell DC (2013) Statistical methods for psychology, 8th edn. Cengage, Boston

Cited by

Still being checked.

Related

Authors
Kuchemüller, K. B., Austerjost, J., Machleid, R., Roßkamp, A., Niemann, J., & McCready, C.
Affiliations
Sartorius (Germany) (DE)
Corresponding author
Kuchemüller, Kim B.
ORCID
Venue
Bioprocess and Biosystems Engineering
Publisher
Springer Science and Business Media LLC
Published
2026-06-16
Volume
49
Issue
9
Pages
2459-2470
ISSN
  • 1615-7591 (ISSN-L)
  • 1615-7605
DOI
10.1007/s00449-026-03362-2
PMC
PMC13582380
OpenAlex
W7164885723
Cited by
0
References
31
FWCI
0.00
Type
journal article
Language
en
Open access
Licence
Locations
  • Bioprocess and Biosystems Engineering (publishedVersion, cc-by-nc-nd)
  • PubMed (publishedVersion)
  • PubMed Central (publishedVersion, cc-by-nc-nd)
  • GoeScholar The Publication Server of the Georg-August-Universität Göttingen (Georg-August-Universität Göttingen) (submittedVersion, cc-by-nc-nd)
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Viral Infectious Diseases and Gene Expression in Insects
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Genetics › Virus-based gene therapy research
  • Physical Sciences › Engineering › Biomedical Engineering › Fluid Dynamics and Mixing
Keywords
HEK293 cells, viral vector production, mechanistic modeling, adenoviral vector, adeno-associated virus vectors, lentiviral vectors, digital shadow, dielectric spectroscopy, soft sensors, cell lysis, infection dynamics, digital twin, perfusion bioreactor, bioprocessing
MeSH terms
Adenoviridae, Genetic Vectors, Humans, Models, Biological, Lentivirus, HEK293 Cells

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-11-06.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-11-06.

Cited sources are checked by gpt-6-sol.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review