Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 46 of the 46 references it cites were found in a registry or an index; 2 of the 6 sources its claims cite were checked for support.

When the Optimal Is Not the Best: Parameter Estimation in Complex Biological Models

Open access
Fernández Slezak, Diego, Suárez, Cecilia, Cecchi, Guillermo A., Marshall, Guillermo, Stolovitzky, Gustavo · 2010 · PLoS ONE

Abstract

BACKGROUND: The vast computational resources that became available during the past decade enabled the development and simulation of increasingly complex mathematical models of cancer growth. These models typically involve many free parameters whose determination is a substantial obstacle to model development. Direct measurement of biochemical parameters in vivo is often difficult and sometimes impracticable, while fitting them under data-poor conditions may result in biologically implausible values. RESULTS: We discuss different methodological approaches to estimate parameters in complex biological models. We make use of the high computational power of the Blue Gene technology to perform an extensive study of the parameter space in a model of avascular tumor growth. We explicitly show that the landscape of the cost function used to optimize the model to the data has a very rugged surface in parameter space. This cost function has many local minima with unrealistic solutions, including the global minimum corresponding to the best fit. CONCLUSIONS: The case studied in this paper shows one example in which model parameters that optimally fit the data are not necessarily the best ones from a biological point of view. To avoid force-fitting a model to a dataset, we propose that the best model parameters should be found by choosing, among suboptimal parameters, those that match criteria other than the ones used to fit the model. We also conclude that the model, data and optimization approach form a new complex system and point to the need of a theory that addresses this problem more generally.

Overview

CoChat Summary

TL;DR

In a six-parameter avascular tumor growth model fitted to synthetic Gompertz growth data, the authors find that the global minimum of the least-squares cost function yields biologically implausible parameters. They argue that parameter estimation should select suboptimal but interpretable parameter sets using independent biological criteria rather than relying only on cost-function minimization.

Key findings

  • The cost-function landscape was highly rugged, with many local minima; the authors used more than 100,000 evaluations and found that the global minimum was not biologically meaningful.
  • Optimization performance differed by method: Levenberg-Marquardt found the smallest minima around 2, MIGRAD and parallel tempering found minima around 20, and downhill simplex found minima in the 100s.
  • Only one minimum centroid per method fell within biologically feasible values for parameters 3 and 4, and neither centroid corresponded to the cluster containing the minimum for LM or PT.
  • Sloppiness analysis showed the smallest eigenvalue corresponded essentially to the vector (0,0,0,0,1,0), making parameter 5 the most sloppy direction; LM’s 5th parameter was seven orders of magnitude smaller than PT’s.

Methods

  • The study used the Ward-King avascular tumor growth model with 6 free parameters bounded in [0,1], fitted to synthetic radius-versus-time data generated from a Gompertz curve with 5% noise over 100 time points.
  • Four optimization approaches were tested: Levenberg-Marquardt, MIGRAD/Fletcher-Davidon-Powell, downhill simplex, and parallel tempering, using a 6-dimensional mesh of 64 starting points and massive parallel runs on Blue Gene with MPI.
  • The authors analyzed minima using histograms, principal component analysis, hierarchical clustering, and sloppiness analysis around cluster centroids.

Limitations

  • The paper states that the model assumes perfect spherical symmetry, Michaelis-Menten growth behavior, ignores the discrete nature of cellular composition, and neglects mechanical stresses between cells.
  • The fitting used synthetic Gompertz data with added noise rather than direct experimental measurements, and the authors describe the global minimum as “what seems to be” the global minimum.
  • Evident from the text: MIGRAD runs were sometimes interrupted at a maximum iteration count, and the conclusions are based on one avascular tumor model and one synthetic dataset.

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

Integrity signals

Registry

Crossref · 10.1371/journal.pone.0013283

Cites

46 of 46 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Figure 1 Data generated using a Gompertz model that fitted previously reported experimental data. Gompertzian fit to the measured volumetric data with a of noise added. The in-box axis corresponds to the same data, where the non-dimensionalized radius as a function of non-dimensionalized time is illustrated, used as the data in the remaining of this paper.
Figure 1 Data generated using a Gompertz model that fitted previously reported experimental data. Gompertzian fit to the measured volumetric data with a of noise added. The in-box axis corresponds to the same data, where the non-dimensionalized radius as a function of non-dimensionalized time is illustrated, used as the data in the remaining of this paper.CC BY · PubMed Central
Figure 2 Histograms of the minima of the cost function obtained using different optimization methods. In all cases, runs started from a fixed 6-dimensional mesh of 64 points, in which each parameter was evaluated at 2 values. The optimization methods were launched and run until convergence, and the minimum obtained in each run was recorded to create these histograms.
Figure 2 Histograms of the minima of the cost function obtained using different optimization methods. In all cases, runs started from a fixed 6-dimensional mesh of 64 points, in which each parameter was evaluated at 2 values. The optimization methods were launched and run until convergence, and the minimum obtained in each run was recorded to create these histograms.CC BY · PubMed Central
Figure 3 6th () vs. 4th () parameters runs for all methods. Grey points represent individual runs. Colored points are the best cost function values obtained in for each of the 64 initial conditions. The value of the attained minimum for each run is color-coded in the color scale at the right of each subfigure. Note that the color bars have different scales.
Figure 3 6th () vs. 4th () parameters runs for all methods. Grey points represent individual runs. Colored points are the best cost function values obtained in for each of the 64 initial conditions. The value of the attained minimum for each run is color-coded in the color scale at the right of each subfigure. Note that the color bars have different scales.CC BY · PubMed Central
Figure 4 The landscape of the cost function as a function of the 6th () and 4th () parameters. The cost function value has been averaged over the values that correspond to the same values of and , but for which the other coordinates differed. This data was taken from all runs available of all methods.
Figure 4 The landscape of the cost function as a function of the 6th () and 4th () parameters. The cost function value has been averaged over the values that correspond to the same values of and , but for which the other coordinates differed. This data was taken from all runs available of all methods.CC BY · PubMed Central
Figure 5 All optimization runs as seen from the cut of the parameter space through the 5th () and 1st () parameters, for the LM, PT and MIGRAD methods. Gray points represent individual runs. Color points are the best cost function value obtained in each compute node. The color bars coding the value of the cost function at the minima are in different scales for the different methods.
Figure 5 All optimization runs as seen from the cut of the parameter space through the 5th () and 1st () parameters, for the LM, PT and MIGRAD methods. Gray points represent individual runs. Color points are the best cost function value obtained in each compute node. The color bars coding the value of the cost function at the minima are in different scales for the different methods.CC BY · PubMed Central
Figure 6 vs points of all minima found (dots) and cluster centroids (crosses). Blue and green represents the minima obtained using the LM and PT methods respectively. The gray box shows the biologically feasible parameter space.
Figure 6 vs points of all minima found (dots) and cluster centroids (crosses). Blue and green represents the minima obtained using the LM and PT methods respectively. The gray box shows the biologically feasible parameter space.CC BY · PubMed Central

Verification

Claims the paper makes

1 Valid · 1 Not valid · 3 Not checked · 67 Own results and interpretations

40 more claims

Create a free account to see all 72 claims and every source verdict.

See all claims — free account
  • It may state two separate claims as one. 18
  • Probably not a substantial claim. 11

Create a free account to see these claims.

Citations

Cites

Show all 44 references

Research works printed but not linked: 1

  1. D Fernández Slezak, 2007, Nonlinear pde system as model of avascular tumor growth., Mecánica Computacional, XXVI, 1788

Cited by

See all 58

Metadata

Authors
Fernández Slezak, D., Suárez, C., Cecchi, G. A., Marshall, G., & Stolovitzky, G.
Affiliations
  • Universidad de Buenos Aires (AR)
  • IBM (United States) (US)
Corresponding author
Stolovitzky, Gustavo
Venue
PLoS ONE
Published
2010-10-25
Volume
5
Issue
10
Pages
e13283
ISSN
1932-6203 (ISSN-L)
OpenAlex
W2071629571
Cited by
57
References
44
Citations by year
2026: 1 · 2024: 2 · 2023: 2 · 2022: 5 · 2021: 2 · 2020: 5 · 2019: 3 · 2018: 3 · 2017: 3 · 2016: 4 · 2015: 9 · 2014: 6 · 2013: 3 · 2012: 4
FWCI
4.10
Type
journal article
Language
en
Open access
Locations
  • PLoS ONE (publishedVersion, cc-by)
  • LA Referencia (Red Federada de Repositorios Institucionales de Publicaciones Científicas) (publishedVersion, other-oa)
  • Europe PMC (PubMed Central) (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • Repositorio Digital Institucional de la Universidad de Buenos Aires (Universidad de Buenos Aires) (publishedVersion, other-oa)
  • arXiv (Cornell University) (submittedVersion)
  • RePEc: Research Papers in Economics (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • arXiv (Cornell University)
  • Greater South Information System
Funders
  • Consejo Nacional de Investigaciones Científicas y Técnicas
  • Universidad de Buenos Aires (UBACYT M415, CONICET)
  • Agencia Nacional de Promoción Científica y Tecnológica (CONICET)
  • Secretaría de Ciencia y Técnica, Universidad de Buenos Aires
Topics
  • Physical Sciences › Mathematics › Modeling and Simulation › Mathematical Biology Tumor Growth
  • Physical Sciences › Mathematics › Applied Mathematics › Mathematical and Theoretical Epidemiology and Ecology Models
  • Life Sciences › Agricultural and Biological Sciences › Agronomy and Crop Science › Agriculture, Soil, Plant Science
Keywords
parameter estimation, avascular tumor growth, tumor growth modeling, local minima, biological plausibility, model fitting, parameter space exploration
MeSH terms
Cell Division, Humans, Models, Biological, Neoplasms

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.

A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (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

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: When the Optimal Is Not the Best: Parameter Estimation in Complex Biological Models. https://app.cochat.ai/papers/10.1371/journal.pone.0013283
The paper
Fernández Slezak, D., Suárez, C., Cecchi, G. A., Marshall, G., & Stolovitzky, G. (2010). When the Optimal Is Not the Best: Parameter Estimation in Complex Biological Models. PLoS ONE. https://doi.org/10.1371/journal.pone.0013283

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.