CoChat
Sign Up
Cited by 15

Hybrid modeling for in silico optimization of a dynamic perfusion cell culture process

Agarwal, Piyush, McCready, Chris, Ng, Say Kong, Ng, Jake Chng, van de Laar, Jeroen, Pennings, Maarten, Zijlstra, Gerben·2025·Biotechnology Progress
DOI 10.1002/btpr.3503PMC PMC11831417
Open this paper in CoChat — free account

Abstract

The bio‐pharmaceutical industry heavily relies on mammalian cells for the production of bio‐therapeutic proteins. The complexity of implementing and high cost‐of‐goods of these processes are currently limiting more widespread patient access. This is driving efforts to enhance cell culture productivity and cost reduction. Upstream process intensification (PI), using perfusion approaches in the seed train and/or the main bioreactor, has shown substantial promise to enhance productivity. However, developing optimal process conditions for perfusion‐based processes remain challenging due to resource and time constraints. Model‐based optimization offers a solution by systematically screening process parameters like temperature, pH, and culture media to find the optimum conditions in silico. To our knowledge, this is the first experimentally validated model to explain the perfusion dynamics under different operating conditions and scales for process optimization. The hybrid model accurately describes Chinese hamster ovary (CHO) cell culture growth dynamics and a neural network model explains the production of mAb, allowing for optimization of media exchange rates. Results from six perfusion runs in Ambr® 250 demonstrated high accuracy, confirming the model's utility. Further, the implementation of dynamic media exchange rate schedule determined through model‐based optimization resulted in 50% increase in volumetric productivity. Additionally, two 5 L‐scale experiments validated the model's reliable extrapolation capabilities to large bioreactors. This approach could reduce the number of wet lab experiments needed for culture process optimization, offering a promising avenue for improving productivity, cost‐of‐goods in bio‐pharmaceutical manufacturing, in turn improving patient access to pivotal medicine.

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1002/btpr.3503

Cites

2 of 2 references verified0 citation errors · 0 not checkable

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • 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.”

    From Mechanistic model for HEK293 viral vector processes and its application in a digital shadow framework (2026) · cites this paper as [8]

    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

  • 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].”

    From Mechanistic model for HEK293 viral vector processes and its application in a digital shadow framework (2026) · cites this paper as [8]

  • 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)”

    From Mechanistic model for HEK293 viral vector processes and its application in a digital shadow framework (2026) · cites this paper as [8]

Self-check

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

The self-check needs the paper’s full text.

Cites

Flagged first, then by citations

Research works printed but not linked: 41

  1. [1]Yeo HC, Park S‐Y, Tan T, Ng SK, Lakshmanan M, Lee D‐Y. Combined multivariate statistical and flux balance analyses uncover media bottlenecks to the growth and productivity of Chinese hamster ovary cell cultures. Biotechnol Bioeng. 2022; 119(7): 1740‐ 1754. 35435243
  2. [2]Mazzoccoli JP, Feke DL, Baskaran H, Pintauro PN. Development of multilayered cell‐hydrogel composites using an acoustic focusing technique. Biotechnol Prog. 2010; 26(2): 600‐ 605. 19938043 PMC2854877
  3. [4]Rittershaus ES, Rehmann MS, Xu J, et al. N‐1 perfusion platform development using a capacitance probe for biomanufacturing. Bioengineering. 2022; 9(4): 128. 35447688 PMC9029935
  4. [5]Müller D, Klein L, Lemke J, et al. Process intensification in the biopharma industry: improving efficiency of protein manufacturing processes from development to production scale using synergistic approaches. Chem Eng Process‐Process Intensification. 2022; 171: 108727.
  5. [6]Xu J, Ou J, McHugh KP, Borys MC, Khetan A. Upstream cell culture process characterization and in‐process control strategy development at pandemic speed. MAbs. 2022; 14: 2060724. 35380922 PMC8986202
  6. [7]Lee AP, Kok YJ, Lakshmanan M, et al. Multi‐omics profiling of a CHO cell culture system unravels the effect of culture ph on cell growth, antibody titer, and product quality. Biotechnol Bioeng. 2021; 118(11): 4305‐ 4316. 34289087
  7. [8]McHugh KP, Xu J, Aron KL, Borys MC, Li ZJ. Effective temperature shift strategy development and scale confirmation for simultaneous optimization of protein productivity and quality in Chinese hamster ovary cells. Biotechnol Prog. 2020; 36(3): e2959. 31930722
  8. [9]Fox SR, Patel UA, Yap MG, Wang DI. Maximizing interferon‐gamma production by Chinese hamster ovary cells through temperature shift optimization: experimental and modeling. Biotechnol Bioeng. 2004; 85(2): 177‐ 184. 14705000
  9. [10]Xu W‐J, Lin Y, Mi C‐L, Pang J‐Y, Wang T‐Y. Progress in fed‐batch culture for recombinant protein production in CHO cells. Appl Microbiol Biotechnol. 2023; 107(4): 1063‐ 1075. 36648523 PMC9843118
  10. [11]Janoschek S, Schulze M, Zijlstra G, Greller G, Matuszczyk J. A protocol to transfer a fed‐batch platform process into semi‐perfusion mode: the benefit of automated small‐scale bioreactors compared to shake flasks as scale‐down model. Biotechnol Prog. 2019; 35(2): e2757. 30479066 PMC6667907
  11. [13]Sokolov M, von Stosch M, Narayanan H, Feidl F, Butté A. Hybrid modeling—a key enabler towards realizing digital twins in biopharma? Curr Opin Chem Eng. 2021; 34: 100715.
  12. [14]Pinto J, Ramos JR, Costa RS, Rossell S, Dumas P, Oliveira R. Hybrid deep modeling of a CHO‐K1 fed‐batch process: combining first‐principles with deep neural networks. Front Bioeng Biotechnol. 2023; 11: 1237963. 37744245 PMC10515724
  13. [15]Maria G. Model‐based optimization of a fed‐batch bioreactor for mab production using a hybridoma cell culture. Molecules. 2020; 25(23): 5648. 33266156 PMC7729860
  14. [16]Naderi S, Meshram M, Wei C, et al. Development of a mathematical model for evaluating the dynamics of normal and apoptotic Chinese hamster ovary cells. Biotechnol Prog. 2011; 27(5): 1197‐ 1205. 21618458
  15. [17]Nolan RP, Lee K. Dynamic model of CHO cell metabolism. Metab Eng. 2011; 13(1): 108‐ 124. 20933095
  16. [18]Selvarasu S, Ho YS, Chong WP, et al. Combined in silico modeling and metabolomics analysis to characterize fed‐batch CHO cell culture. Biotechnol Bioeng. 2012; 109(6): 1415‐ 1429. 22252269
  17. [19]Narayanan H, Luna M, Sokolov M, Butté A, Morbidelli M. Hybrid models based on machine learning and an increasing degree of process knowledge: application to cell culture processes. Ind Eng Chem Res. 2022; 61(25): 8658‐ 8672.
  18. [20]Richelle A, Corbett B, Agarwal P, Vernersson A, Trygg J, McCready C. Model‐based intensification of CHO cell cultures: one‐step strategy from fed‐batch to perfusion. Front Bioeng Biotechnol. 2022; 10: 948905. 36072286 PMC9443430
  19. [21]Ozturk S, Hu W‐S. Cell Culture Technology for Pharmaceutical and Cell‐Based Therapies. CRC Press; 2005.
  20. [22]Walsh I, Myint M, Nguyen‐Khuong T, Ho YS, Ng SK, Lakshmanan M. Harnessing the potential of machine learning for advancing “quality by design” in biomanufacturing. MAbs. 2022; 14: 2013593. 35000555 PMC8744891
  21. [23]Bayer B, Duerkop M, Pörtner R, Möller J. Comparison of mechanistic and hybrid modeling approaches for characterization of a CHO cultivation process: requirements, pitfalls and solution paths. Biotechnol J. 2023; 18(1): 2200381. 36382343
  22. [24]Doyle K, Tsopanoglou A, Fejér A, Glennon B, del Val IJ. Automated assembly of hybrid dynamic models for CHO cell culture processes. Biochem Eng J. 2023; 191: 108763.
  23. [25]Kotidis P, Jedrzejewski P, Sou SN, et al. Model‐based optimization of antibody galactosylation in CHO cell culture. Biotechnol Bioeng. 2019; 116(7): 1612‐ 1626. 30802295
  24. [26]Kontoravdi C, Pistikopoulos EN, Mantalaris A. Systematic development of predictive mathematical models for animal cell cultures. Comput Chem Eng. 2010; 34(8): 1192‐ 1198.
  25. [27]Xu J, Tang P, Yongky A, et al. Systematic development of temperature shift strategies for Chinese hamster ovary cells based on short duration cultures and kinetic modeling. MAbs. 2019; 11(1): 191‐ 204. 30230966 PMC6343780
  26. [28]Bylesjö M, Rantalainen M, Cloarec O, Nicholson JK, Holmes E, Trygg J. Opls discriminant analysis: combining the strengths of pls‐da and simca classification. J Chemom. 2006; 20(8–10): 341‐ 351.
  27. [29]Agarwal P. Application of Deep Learning in Chemical Processes: Explainability, Monitoring and Observability (PhD dissertation). University of Waterloo; 2021.
  28. [30]Nagrath D, Messac A, Bequette BW, Cramer S. A hybrid model framework for the optimization of preparative chromatographic processes. Biotechnol Prog. 2004; 20(1): 162‐ 178. 14763840
  29. [31]Mehrian M, Guyot Y, Papantoniou I, et al. Maximizing neotissue growth kinetics in a perfusion bioreactor: an in silico strategy using model reduction and bayesian optimization. Biotechnol Bioeng. 2018; 115(3): 617‐ 629. 29205280
  30. [32]Khuat TT, Bassett R, Otte E, Grevis‐James A, Gabrys B. Applications of machine learning in antibody discovery, process development, manufacturing and formulation: current trends, challenges, and opportunities. Comput Chem Eng. 2024; 182: 108585.
  31. [33]Saleh D, Wang G, Rischawy F, Kluters S, Studts J, Hubbuch J. In silico process characterization for biopharmaceutical development following the quality by design concept. Biotechnol Prog. 2021; 37(6): e3196. 34309240
  32. [34]Wang Z, Wang C, Chen G. Kinetic modeling: a tool for temperature shift and feeding optimization in cell culture process development. Protein Expr Purif. 2022; 198: 106130. 35691496
  33. [35]Kroll P, Eilers K, Fricke J, Herwig C. Impact of cell lysis on the description of cell growth and death in cell culture. Eng Life Sci. 2017; 17: 440– 447. 32624789 PMC6999402
  34. [36]Strober W. Trypan blue exclusion test of viability. Curr Protoc Immunol. 2015; 111: A3.B.1‐ A3.B.3 PMC6716531 26529666
  35. [37]Agarwal P, Tamer M, Sahraei MH, Budman H. Deep learning for classification of profit‐based operating regions in industrial processes. Ind Eng Chem Res. 2019; 59(6): 2378‐ 2395.
  36. [38]Agarwal P, Budman H. Classification of profit‐based operating regions for the tennessee eastman process using deep learning methods. IFAC‐PapersOnLine. 2019; 52(1): 556‐ 561.
  37. [39]Goodfellow I, Bengio Y, Courville A, Bengio Y. Deep Learning. Vol 1. MIT Press; 2016.
  38. [40]Adams TA II, Seider WD. Practical optimization of complex chemical processes with tight constraints. Comput Chem Eng. 2008; 32(9): 2099‐ 2112.
  39. [42]Virtanen P, Gommers R, Oliphant TE, et al. SciPy 1.0 contributors, “SciPy 1.0: fundamental algorithms for scientific computing in python”. Nat Methods. 2020; 17: 261‐ 272. 32015543 PMC7056644
  40. [43]Agarwal P, Tamer M, Budman H. Explainability: relevance based dynamic deep learning algorithm for fault detection and diagnosis in chemical processes. Comput Chem Eng. 2021; 154: 107467.
  41. [44]Lundberg SM, Lee S‐I. A unified approach to interpreting model predictions. Adv Neural Inf Proces Syst. 2017; arXiv:1705.07874. · arXiv

Cited by

Sorted by citations

Authors
Agarwal, P., McCready, C., Ng, S. K., Ng, J. C., van de Laar, J., Pennings, M., & Zijlstra, G.
Affiliations
  • Sartorius Corporate Research Oakville Ontario Canada
  • Bioprocessing Technology Institute (BTI) A*STAR Biopolis Way Singapore
  • BiosanaPharma Leiden Netherlands
  • FMBT Marketing, Sartorius Netherlands Amersfoort Netherlands
ORCID
Agarwal, Piyush
Venue
Biotechnology Progress
Publisher
Wiley
Published
2024-09-18
Volume
41
Issue
1
ISSN
  • 8756-7938
  • 1520-6033
DOI
10.1002/btpr.3503
PMC
PMC11831417
Cited by
15
References
2
Language
en
Licence
creativecommons.org/licenses/by/4.0/ (vor)
Funders
Agency for Science, Technology and Research (I1901E0047)

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.

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