Pith. sign in

REVIEW 4 major objections 5 minor 23 references

Hybrid Machine Learning and Mathematical Modeling for Tumor Dynamics Prediction: Comparing SPIONs against mNP-FDG

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Combining XGBoost with exponential and logistic growth models, the paper predicts that mNP-FDG contains tumor growth within days while SPIONs achieve complete elimination sooner, favoring sequenced combination therapy.

desk verdict Desk reject: the headline timing comparison is internally inconsistent, rests on non-exchangeable datasets, and the 'prediction' is an in-sample fit. read the letter →

arxiv 2505.21094 v1 pith:Q62MMR3V submitted 2025-05-27 q-bio.QM cond-mat.softphysics.med-ph

Amit K Chattopadhyay , Aimee Pascaline N Unkundiye , Gillian Pearce This is my paper
classification q-bio.QMcond-mat.softphysics.med-ph
keywords tumordynamicsSPIONsmNP-FDGXGBoostexponentialgrowthmodellogisticnanoparticletherapygraphicaluserinterface
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper tries to establish that two nanoparticle cancer treatments act on different timescales and can be ranked by a hybrid machine-learning plus mathematical-modeling pipeline. It claims that fluorodeoxyglucose-coated nanoparticles (mNP-FDG) contain tumor growth very quickly, within 2 to 4 days depending on the section, whereas superparamagnetic iron oxide nanoparticles (SPIONs) take roughly 18 to 23 days to contain the tumor but eliminate it by about 20 to 23 days, while mNP-FDG takes more than 40 days. If this is right, the best strategy is not either agent alone but sequential therapy: mNP-FDG first to rapidly tame the tumor, then SPIONs to eradicate it. The paper matters because it turns sparse published tumor-volume curves into an interactive prediction tool that a clinician or patient could query by treatment type and time.

What carries the argument

The engine is a hybrid fitting pipeline. XGBoost, a gradient-boosting ensemble, is the supervised learner that maps input features, including initial tumor volume, treatment type, and time, to subsequent tumor volumes; exponential and logistic growth equations serve as the mechanistic continuum models that the ML output is fitted against and that give the curves their biological shape. The load-bearing comparison rests on digitizing published tumor-volume trajectories with MATLAB's Grabit tool, so the extracted curves become the training data. A Tkinter GUI wraps the fitted model for real-time queries.

What would settle it

A head-to-head experiment using the same tumor cell line, the same animal model, and matched dosing would settle the claim: if mNP-FDG does not contain tumor growth within the predicted few days, or if SPIONs do not produce complete elimination near 20 to 23 days while mNP-FDG trails past 40 days, the central timescale ranking fails.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is a comparative timescale result: mNP-FDG delivers fast tumor containment while SPIONs deliver faster complete termination, so the two agents are complementary rather than simply better or worse. Using XGBoost trained on digitized longitudinal tumor-volume data and fitted alongside exponential and logistic growth models, the paper reports cross-validated predictions with R-squared above 0.95. From those predictions it reads that mNP-FDG reaches a saturation regime within days, whereas SPIONs eliminate the tumor in about 20 to 23 days against more than 40 days for mNP-FDG. The paper therefore concludes that combined therapy is optimum, using mNP-FDG for early control and SPIONs for eradication, and packages the framework in a Tkinter GUI for real-time prediction.

Load-bearing premise

The ranking of the two treatments depends on treating digitized tumor-volume curves from different published studies as comparable data, even though those studies may have used different tumor models, species, and experimental conditions.

Editorial extensions

If this is right

  • If the timescale ranking holds, treatment protocols should switch agents mid-course rather than pick one, using mNP-FDG for rapid containment and SPIONs for elimination.
  • The hybrid XGBoost plus growth-model pipeline can produce R-squared above 0.95 on sparse longitudinal tumor data, which supports its use as a quantitative treatment-planning tool.
  • The GUI makes the fitted predictions directly queryable by time, species, and treatment type, lowering the barrier to clinical self-assessment.
  • A sequenced mNP-FDG-then-SPION regimen would be the optimal joint strategy, rather than either agent alone, according to the paper's predictions.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Because the two treatment arms come from separate published datasets, the specific day counts should be read as exploratory; a matched experimental comparison could shift the crossover time between containment and elimination.
  • The same hybrid fitting approach could be ported to other nanoparticle-antibody pairs by digitizing their published tumor-volume curves, giving a cheap screening method for sequencing multiple therapies.
  • A natural next step the paper does not develop is to search over switch times: given the two fitted curves, one can compute an optimal day to transition from mNP-FDG to SPION so that total time to elimination is minimized.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The manuscript proposes a hybrid framework that combines XGBoost machine learning with exponential and logistic tumor-growth models to compare two nanoparticle treatments, SPIONs and mNP-FDG, using tumor-volume time series digitized from published figures. The central quantitative claims are that mNP-FDG controls tumor progression within about 2--4 days versus 17--23 days for SPIONs, while SPIONs achieve complete tumor elimination in about 20--23 days versus more than 40 days for mNP-FDG. The authors also present a Tkinter-based GUI intended for clinical users to input treatment type and time and receive predicted tumor volumes. The paper argues for a combined/sequenced therapy, with mNP-FDG for rapid containment and SPIONs for eventual eradication.

Significance. If the comparative ranking were rigorously established, the work would be of interest to the nanomedicine and computational oncology communities, and the GUI is a practical deliverable that could lower the barrier to using such models. However, the manuscript currently provides no raw digitized data, fitted parameter values, model equations, held-out validation, or uncertainty quantification, and the headline numerical claims are internally inconsistent. The paper also credits the use of ensemble learning and the choice of transparent growth models, but these strengths are not enough to support the central comparative conclusion without addressing the comparability and reproducibility problems described below.

major comments (4)
  1. [Abstract; Section III.C; Section V] The reported treatment-comparison times are mutually inconsistent. The abstract reports containment in 2 days (mNP-FDG) versus 18 days (SPIONs) and complete termination in 20 days (SPIONs) versus more than 40 days (mNP-FDG); Section III.C reports containment in 4 days versus about 23 days; and the Conclusion reports containment in 4 days versus 17--18 days and elimination in about 23 days versus more than 40 days. No equation, threshold definition, or algorithm is given for 'control' or 'complete termination,' so it is unclear whether these discrepancies are due to different threshold choices, curve-reading errors, or unstable model fits. This inconsistency undermines the paper's central quantitative claim.
  2. [Section II.A; Section III.A; Main Limitations] The comparative ranking requires that the two treatment arms be exchangeable, but the mNP-FDG data come from the authors' own prior mouse study [5] while the SPION data come from Russell et al. [2], which used different cells and combined SPIONs with radiation. Section III.A itself states that 'the real comparison that we want to make is against a mNP-FDG led treatment regime against a combinatorial drug outlined on radiation and SPIONs together.' The Main Limitations concedes that the sources 'may have been conducted under varying experimental conditions' and that 'different forms of cancer may have been addressed.' Under these conditions, the 2/18-day or 4/23-day difference cannot be attributed to the nanoparticle type rather than to differences in cancer model, treatment protocol, or measurement procedure.
  3. [Section II.B; Section III.B] The so-called predictions are in-sample restatements of the digitized curves. The XGBoost models are trained and evaluated on the same extracted tumor-volume time series, and the headline containment/elimination times are read off the fitted curves. No held-out data, cross-validation splits, or independent test set are described, and the reported R² values exceeding 0.95 are presented without confidence intervals or error bars. Consequently, the manuscript provides no evidence that the framework can predict tumor dynamics for unseen experiments, which is the central predictive claim.
  4. [Section III.C.2; Section IV] The mathematical modeling component is not specified enough to reproduce or assess. The exponential and logistic growth equations are described verbally, but the fitted parameter values (growth rate r, carrying capacity K), the threshold criteria for 'containment' and 'complete termination,' and the way XGBoost is combined with these models are not given. The Discussion refers to 'probabilistic fitting' and 'quantification of uncertainty,' but no uncertainty intervals, residual analyses, or model-selection criteria appear anywhere in the manuscript.
minor comments (5)
  1. [Abstract] The abstract opens with 'Outline:' rather than a standard summary sentence, and the spelling 'SPIONS' is used inconsistently with 'SPIONs' elsewhere; this should be corrected.
  2. [Throughout] There are numerous typos and grammatical errors, including 'hjighly,' 'uncleear,' 'trermed,' 'thee,' 'speciees,' and 'reast' in reference [17]; the manuscript needs careful proofreading.
  3. [Section II.A] The statement that 'Preprocessing was not required' is surprising for data extracted from figures with MATLAB Grabit, since digitization introduces measurement error; the authors should at least discuss how extraction error was handled.
  4. [Section III.D] The GUI is described as accepting 'speciees type (mice, pigs, humans),' but the manuscript gives no evidence that the underlying models were validated on pig or human data; this claim should be moderated or supported.
  5. [References] Reference [5] is cited as the source of the mNP-FDG data, but the reference title contains a typo ('Biostatisstics') and the relevant dataset is not provided in the manuscript; supplying the digitized data or a public repository link would improve reproducibility.

Circularity Check

2 steps flagged · score 7.0 of 10

Central comparison is read off fitted curves trained on the same digitized data; the mNP-FDG arm derives from the authors' own prior paper.

  1. fitted input called prediction [Sections II.B, III.A, III.C; Figures 1-4]
    "Comparing Figures 1 and 2, it is visually obvious that mNP-FDG (Figure 1) display a saturation regime after 2 days that is not seen with SPIONs (Figure 2). ... Figures 3 and 4 present comparative analyzes of tumor volume reductions under mNP-FDG and SPION treatments, as predicted by XGBoost framework. Once again, mNP-FDG is shown to contain the tumor within 4 days compared to about 23 days for SPIONs."

    The XGBoost target is 'tumor volume at subsequent time intervals' from the same digitized curves (Section II.B), so the model is fit to the curves it is then said to predict. The headline containment times, whether 2 vs 18 days (Abstract), 4 vs 23 days (Section III.C), or 4 vs 17-18 days (Conclusion), are read off fitted curves trained on those very data. Thus the 'prediction' is an in-sample restatement of the input digitized curves, not an out-of-sample or independent forecast.

  2. self citation load bearing [Section II.A; Figure 1 caption; references [4,5]]
    "Tumor volume data were acquired from experimental and clinical studies involving SPIONs and mNP-FDG as therapeutic agents [5, 6]. ... As previously analyzed in [5], Figure 1 illustrates the temporal evolution of tumor volumes comparing the two forms of treatments."

    Reference [5] is the authors' own prior paper (Chattopadhyay, Pearce, Unkundiye, Russell), and the mNP-FDG arm used for the central comparison is digitized from it. The current study's claim that mNP-FDG contains tumors quickly is therefore inherited from the authors' self-cited dataset and re-fitted by XGBoost, rather than being supported by an independent mNP-FDG experiment. Without [5], the mNP-FDG arm and its rapid-saturation conclusion would not exist; the comparison reduces to re-analysis of the authors' own previous mouse data against an external SPION dataset.

full rationale

The paper's central comparative claim—mNP-FDG controls tumors faster but SPIONs achieve complete elimination sooner—is not an independent prediction. The XGBoost model is trained on the same digitized longitudinal tumor-volume curves from which the thresholds are then read off, so the reported containment/eradication times are in-sample interpolations of the training data. Compounding this, the mNP-FDG data come from the authors' own prior publication [5], making the ranking partly a self-citation-dependent re-fit. The paper itself concedes in 'Main Limitations' that the data sources 'may have been conducted under varying experimental conditions' and that 'different forms of cancer may have been addressed,' which further undermines the comparability of the two arms. The internal inconsistency of the headline numbers (2 vs 18 days in the Abstract, 4 vs 23 days in Section III.C, 4 vs 17-18 days and 23 vs 40+ days in the Conclusion) indicates the result is not a stable consequence of a single defined threshold or algorithm. No raw digitized data, code, or fitted parameter values are provided, so the 'predictions' cannot be independently reproduced. These features place the core finding at the boundary between a fitted-input-called-prediction and a self-citation-supported re-analysis; the external SPION data from Russell et al. prevent a score of 10, but the central ranking still reduces substantially to the training data and the authors' prior work.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The paper does not report the values of the growth model parameters, the XGBoost hyperparameters, or the operational definitions of 'control' and 'elimination.' The comparison therefore rests on unstated fitted parameters and threshold choices. No new material entities are introduced; the 'hybrid model' is a framing, not a concrete derived equation.

free parameters (5)
  • Exponential growth rate r
    Used in the exponential growth model fitted to digitized data, but the paper never reports the fitted value or the fitting procedure.
  • Logistic growth rate r and carrying capacity K
    The logistic model is mentioned in Section III.C.2, but the fitted parameters are not reported.
  • Containment threshold
    The paper never defines what 'control' or 'containment' means quantitatively; the reported day values must depend on an arbitrary threshold of tumor volume reduction.
  • Elimination threshold
    Similarly, 'complete termination' is not defined; the day values depend on an unstated endpoint criterion.
  • XGBoost hyperparameters
    No details are given for max depth, learning rate, number of estimators, or other settings that affect the fitted curves.
assumptions (3)
  • domain assumption Tumor growth follows exponential and logistic ODE models
    Invoked in Section III.C.2 as the continuum models 'machine learned against'; these are standard but their applicability to these specific therapies is not justified.
  • domain assumption Digitized trendlines from figures in [2,5] are accurate representations of the underlying tumor volume measurements
    Section II.A states data were extracted using MATLAB Grabit from published figures; no error analysis of the digitization is presented.
  • domain assumption Cross-study comparability of tumor volume data
    The abstract's Main Limitations admits the data come from varying experimental conditions and possibly different cancer types; the comparison of mNP-FDG data [5] with SPION+radiation data [2] assumes these arms are exchangeable.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Hybrid Machine Learning and Mathematical Modeling for Tumor Dynamics Prediction: Comparing SPIONs against mNP-FDG." pith.science (2026). https://pith.science/paper/Q62MMR3V

@misc{pith2026250521094,
  author       = {Pith},
  title        = {Pith review of: Hybrid Machine Learning and Mathematical Modeling for Tumor Dynamics Prediction: Comparing SPIONs against mNP-FDG},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Q62MMR3V}},
  note         = {Machine review of arXiv:2505.21094}
}
read the original abstract

This is a Machine Learning guided study towards zone-specific ray therapy. Combining Machine Learning (Extreme Gradient Boosting) with continuum modeling (exponential and logistic growth), we find that while fluorodeoxyglucose-coated (mNP-FDG) can control cancerous tumor progression within 2 days compared to 18 days by Superparamagnetic Iron Oxide Nanoparticles (SPIONs), for complete termination of the tumor, SPIONS (20 days) are superior compared to mNP-FDG (more than 40 days). We also provide an interactive graphical user interface (GUI) developed with Tkinter/Python that allows users to input relevant data, such as treatment type and time, to receive real-time tumor volume predictions. Our ML-guided prediction indicates joint therapy as the optimum choice, with mNP-FDG ideal for taming the tumor spread, followed by SPIONs for complete eradication, facilitating personalized cancer treatment in clinical practice.

Figures

Figures reproduced from arXiv: 2505.21094 by the authors.

Figure 1
Figure 1. FIG. 1: MATLAB-GRABIT extracted data from Chattopadhyay, et al: Time evolution of [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2: MATLAB-GRABIT extracted data from Russell, et al:Time evolution of tumor [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3: mNP-FDG prediction with XG Boost [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: FIG. 4: SPIONS prediction with XG Boost [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5: Tumor shrinkage under mNP-FDG [PITH_FULL_IMAGE:figures/full_fig_p012_5.png]
Figure 6
Figure 6. Figure 6: FIG. 6: Tumor shrinkage under SPIONS [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]
Figure 7
Figure 7. Figure 7: FIG. 7: Graphical User Interface [PITH_FULL_IMAGE:figures/full_fig_p013_7.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

23 extracted references · 22 canonical work pages

  1. [5]

    Kirakli, E., Korkmaz, E., et al. (2018). Superparamagnetic Iron Oxide Nanoparticle (SPION) Mediated in Vitro Radiosensitization at Megavoltage Radiation Energies. Journal of Radioan- alytical and Nuclear Chemistry , 315 (3), 595–602. Available at: https://doi.org/10.1007/ s10967-018-5704-9 [Accessed 11 Feb. 2025]

    work page 2018

  2. [2]

    The exponential growth model, character- ized by a constant growth rate, is frequently used to describe the initial phases of tumor development when resources are abundant

    Mathematical Models of Tumor Growth Mathematical modeling of cancer growth often employs simplified models to capture the essential dynamics of tumor proliferation [18]. The exponential growth model, character- ized by a constant growth rate, is frequently used to describe the initial phases of tumor development when resources are abundant. However, as an...

    work page

  3. [1]

    In this study, we have used Machine Learning (ML) tools to predict tumor growth and for optimizing treatment regimens

    Machine Learning in Oncology Machine learning(ML) has emerged as a powerful tool, and even in oncology it can used for the analysis of datasets [5]. In this study, we have used Machine Learning (ML) tools to predict tumor growth and for optimizing treatment regimens. Of late, XGBoost has become a powerful tool in cancer modeling due to its high accuracy a...

    work page

  4. [3]

    WHO Report on Cancer: Setting Priorities, In- vesting Wisely and Providing Care for All

    World Health Organisation (2020). WHO Report on Cancer: Setting Priorities, In- vesting Wisely and Providing Care for All. [online] Google Books. Available at: https://books.google.co.uk/books?hl=en&lr=&id=anYOEQAAQBAJ&oi=fnd&pg=PA6&ots= N3E0sEHbKC&sig=05zbEYRXy3jFzsswrzlTQqDi_S0&redir_esc=y#v=onepage&q&f=false

    work page 2020

  5. [4]

    Russell, E., Ward, T., Hood, T., Howse, C., and Lin, F. (2021). Impact of Superparamagnetic Iron Oxide Nanoparticles on in Vitro and in Vivo Radiosensitisation of Cancer Cells.Radiation Oncology, 16 (1). Available at: https://doi.org/10.1186/s13014-021-01829-y [Accessed 25 Jan. 2025]

    work page doi:10.1186/s13014-021-01829-y 2021

  6. [6]

    Aras, O., Pearce, G., Watkins, A.J., Nurili, F., Medine, E.I., Kozgus Guldu, O., Tekin, V., Wong, J., Ma, X., Ting, R., Unak, P., & Akin, O. (2018). An in-vivo pilot study into the effects of FDG-mNP in cancer in mice. PLOS ONE , 13(8), e0202482. https://doi.org/10. 1371/journal.pone.0202482

    work page 2018

  7. [7]

    K., Pearce, G., Unkundiye P

    Chattopadhyay, A. K., Pearce, G., Unkundiye P. A. N., and Russell, S. T. (2024). Predicting the Progression of Cancerous Tumors in Mice: A Machine and Deep Learning Intuition.Annals of Biostatisstics and Biometrics Applications , 6 (2), DOI:10.33552/ABBA.2024.06.000632

    work page doi:10.33552/abba.2024.06.000632 2024

  8. [8]

    Vangijzegem, T., Stanicki, D., and Laurent, S. (2023). Superparamagnetic Iron Oxide Nanoparticles (SPION): From Fundamentals to State-of-The-Art Innovative Applications for Cancer Therapy. Pharmaceutics, 15 (1), 236. Available at: https://doi.org/10.3390/ pharmaceutics15010236 [Accessed 25 Jan. 2025]

    work page 2023

Show all 23 references
  1. [9]

    Gao, H., Cao, S., Yang, Z., Zhang, S., Zhang, Q., & Jiang, X. (2015). Preparation, Characterization and Anti-Glioma Effects of Docetaxel-Incorporated Albumin-Lipid 16 Nanoparticles. Journal of Biomedical Nanotechnology , 11(12), 2137–2147. https: //www.researchgate.net/publica...

    2015

  2. [10]

    Qi, M., Xu, D., Wang, S., Liang, J., Liu, B., Guan, Q., & Liu, S. (2022). In Vivo Metabolic Analysis of the Anticancer Effects of Plasma-Activated Saline in Three Tumor Animal Models. Biomedicines, 10 (3), 528–528. https://doi.org/10.3390/biomedicines10030528

    2022 doi

  3. [11]

    et al (2016)

    Subramanian, M., Pearce, G. et al (2016). A pilot study into the use of FDG-mNP as an alternative approach in neuroblastoma cell hyperthermia. IEEE Trans Nanobioscience , 15 (6):517–525. doi: 10.1109/TNB.2016.2584543

    2016

  4. [12]

    et al (2021)

    Nosrati, R. et al (2021). Targeted SPION siderophore conjugate loaded with doxorubicin as a theranostic agent for imaging and treatment of colon carcinoma.Scientific Reports, 11 : 13065

    2021

  5. [13]

    S., Leong, D

    Muthu, M. S., Leong, D. T., Mei, L., & Feng, S. S. (2014). Nanotheranostics—application and further development of nanomedicine strategies for advanced theranostics. Theranostics, 4 (6), 660-677

    2014

  6. [14]

    Unak, P., Unak, T., & Yildirim, A. (2017). A pilot study into the use of FDG-mNP as an alternative approach in neuroblastoma cell hyperthermia. Journal of Cancer Research and Therapeutics, 13 (2), 252-256

    2017

  7. [15]

    Kumar, R., & Panda, D. (2018). Mechanisms underlying 18F-fluorodeoxyglucose accumulation in colorectal cancer. World Journal of Nuclear Medicine , 17 (1), 3-6

    2018

  8. [16]

    et al (2022)

    Rezaeian, M. et al (2022). Mathematical Modeling of Targeted Drug Delivery Using Magnetic Nanoparticles during Intraperitoneal Chemotherapy. Pharmaceutics, 14 (2), 324; https:// doi.org/10.3390/pharmaceutics14020324

    2022 doi

  9. [17]

    Mathematical modeling in cancer nanomedicine: a review

    Dogra, et al. Mathematical modeling in cancer nanomedicine: a review. Biomed Microdevices, 4 ;21(2):40. doi: 10.1007/s10544-019-0380-2

    doi

  10. [18]

    et al (2023)

    Guan, X. et al (2023). Construction of the XGBoost model for early lung cancer predic- tion based on metabolic indices. BMC Med Inform Decis Mak , 13 ;23:107. doi: 10.1186/ s12911-023-02171-x

    2023

  11. [19]

    Mathew, T. N. reast Cancer Classification Using an Extreme Gradient Boosting Model with F-Score Feature Selection Technique.Journal of Advances in Information Technology, Vol. 14, No. 2. 17

  12. [20]

    Murphy H., Jaafari H., Dobrovolny, H. M. (2016). Differences in predictions of ODE mod- els of tumor growth: a cautionary example. BMC Cancer. , 26 (16):163. doi: 10.1186/ s12885-016-2164-x

    2016

  13. [21]

    E., et al (2019)

    Johnson, K. E., et al (2019). Cancer cell population growth kinetics at low densities deviate from the exponential growth model and suggest an Allee effect. Plos Biology : https://doi. org/10.1371/journal.pbio.3000399

    2019 doi

  14. [22]

    V., Clinton, S

    Jain, H. V., Clinton, S. K., Bhinder, A., and Avner Friedman (2011). Mathematical modeling of prostate cancer progression in response to androgen ablation therapy. Proc. Natl. Acad. Sci., 108 (49) 19701-19706. https://doi.org/10.1073/pnas.11157501

    2011 doi

  15. [23]

    et al (2020)

    Vaghi, C. et al (2020). Population modeling of tumor growth curves and the reduced Gom- pertz model improve prediction of the age of experimental tumors. PLoS Comput Biol. , 25 ;16(2):e1007178. doi: 10.1371/journal.pcbi.1007178. 18

    2020 doi

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.