Nick G. Cogan
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Nick G. Cogan has a total of 35 co-authors. Here are the top co-authors by number of shared publications:
Barbara Szomolay
Mathematics Institute, University of Warwick, Coventry, CV4 7AL, UK
2 shared publications- Bhargav R. Karamched
Florida State University
2 shared publications - Manu Aggarwal
National Institutes of Health, Bethesda, MD
2 shared publications Shankararaman Chellam
2 shared publications- Zainab O. Dere
Department of Mathematics, Florida State University, Tallahassee, 32306, FL, USA
2 shared publications - Angela Jarrett
Institute for Computational Engineering and Sciences, University of Texas at Austin, Austin, USA
1 shared publication
Latest Publications
- Open Access
- 13 May 2026
- Published by The Royal Society in Journal of The Royal Society Interface
- Vol. 23 (238)
Order matters: how the sequence of sensitivity analysis, parameter identifiability and uncertainty quantification shapes model insights
Mathematical models are essential tools for understanding biological systems, but their predictive value depends on how parameter uncertainty is handled. Three core approaches—uncertainty quantification (UQ), sensitivity analysis (SA) and parameter identifiability (ID)—address distinct facets of the problem. However, they are usually applied in isolation. Here, we show that the workflow order in which these methods are implemented impacts parameter prioritization, model reliability and the interpretation of biological mechanisms. We compare three analytical sequences (SA→ID→UQ, ID→SA→UQ and UQ→ID→SA) and demonstrate that each suggests different parameters to focus on. Integrating these perspectives provides consistent insights that are not obtained from any single method. Our results establish a framework for combining UQ, SA and ID that links parameter analysis directly to biological questions and experimental design. We illustrate this in the context of bacterial persistence and wastewater filtration; however, the framework is general and applicable to a wide range of problems where reliable prediction from models is essential. Unlike prior studies that treat these analyses in isolation, we show that workflow order systematically changes parameter prioritization and experimental recommendations.
- 13 May 2026
- Published by The Royal Society in Journal of The Royal Society Interface
- Vol. 23 (238)
Open AccessOrder matters: how the sequence of sensitivity analysis, parameter identifiability and uncertainty quantification shapes model insights
Mathematical models are essential tools for understanding biological systems, but their predictive value depends on how parameter uncertainty is handled. Three core approaches—uncertainty quantification (UQ), sensitivity analysis (SA) and parameter identifiability (ID)—address distinct facets of the problem. However, they are usually applied in isolation. Here, we show that the workflow order in which these methods are implemented impacts parameter prioritization, model reliability and the interpretation of biological mechanisms. We compare three analytical sequences (SA→ID→UQ, ID→SA→UQ and UQ→ID→SA) and demonstrate that each suggests different parameters to focus on. Integrating these perspectives provides consistent insights that are not obtained from any single method. Our results establish a framework for combining UQ, SA and ID that links parameter analysis directly to biological questions and experimental design. We illustrate this in the context of bacterial persistence and wastewater filtration; however, the framework is general and applicable to a wide range of problems where reliable prediction from models is essential. Unlike prior studies that treat these analyses in isolation, we show that workflow order systematically changes parameter prioritization and experimental recommendations.
- Open Access
- 14 May 2025
- Published by Elsevier in Mathematical Biosciences
- Vol. 386, 109464
Optimal control strategies for mitigating antibiotic resistance: Integrating virus dynamics for enhanced intervention design
- 14 May 2025
- Published by Elsevier in Mathematical Biosciences
- Vol. 386, 109464
Open AccessOptimal control strategies for mitigating antibiotic resistance: Integrating virus dynamics for enhanced intervention design
- Open Access
Sensitivity based model agnostic scalable explanations of deep learning
Deep neural networks (DNNs) are powerful tools for data-driven predictive machine learning, but their complex architecture obscures mechanistic relations that they have learned from data. This information is critical to the scientific method of hypotheses development, experiment design, and model validation, especially when DNNs are used for biological and clinical predictions that affect human health. We design SensX, a model agnostic explainable AI (XAI) framework that outperformed current state-of-the-art XAI in accuracy (up to 52% higher) and computation time (up to 158 times faster), with higher consistency in all cases. It also determines an optimal subset of important input features, reducing dimensionality of further analyses. SensX scaled to explain vision transformer (ViT) models with more than 150, 000 features, which is computationally infeasible for current state-of-the-art XAI. SensX validated that ViT models learned justifiable features as important for different facial attributes of different human faces. SensX revealed biases inherent to the ViT architecture, an observation possible only when importance of each feature is explained. We trained DNNs to annotate biological cell types using single-cell RNA-seq data and SensX determined the sets of genes that the DNNs learned to be important to different cell types.
Open AccessSensitivity based model agnostic scalable explanations of deep learning
Deep neural networks (DNNs) are powerful tools for data-driven predictive machine learning, but their complex architecture obscures mechanistic relations that they have learned from data. This information is critical to the scientific method of hypotheses development, experiment design, and model validation, especially when DNNs are used for biological and clinical predictions that affect human health. We design SensX, a model agnostic explainable AI (XAI) framework that outperformed current state-of-the-art XAI in accuracy (up to 52% higher) and computation time (up to 158 times faster), with higher consistency in all cases. It also determines an optimal subset of important input features, reducing dimensionality of further analyses. SensX scaled to explain vision transformer (ViT) models with more than 150, 000 features, which is computationally infeasible for current state-of-the-art XAI. SensX validated that ViT models learned justifiable features as important for different facial attributes of different human faces. SensX revealed biases inherent to the ViT architecture, an observation possible only when importance of each feature is explained. We trained DNNs to annotate biological cell types using single-cell RNA-seq data and SensX determined the sets of genes that the DNNs learned to be important to different cell types.
- Open Access
Optimal Control Strategies for Mitigating Antibiotic Resistance: Integrating Virus Dynamics for Enhanced Intervention Design
Given the global increase in antibiotic resistance, new effective strategies must be developed to treat bacteria that do not respond to first or second line antibiotics. One novel method uses bacterial phage therapy to control bacterial populations. Phage viruses replicate and infect bacterial cells and are regarded as the most prevalent biological agent on earth. This paper presents a comprehensive model capturing the dynamics of wild-type bacteria ( S ), antibiotic-resistant bacteria ( R ), and infective (I) strains, incorporating virus inclusion. Our model integrates biologically relevant parameters governing bacterial birth rates, death rates, and mutation probabilities and incorporates infection dynamics via contact with a virus. We employ an optimal control approach to study the influence of virus inclusion on bacterial population dynamics. Through numerical simulations, we establish insights into the stability of various system equilibria and bacterial population responses to varying infection rates. By examining the equilibria, we reveal the impact of virus inclusion on population trajectories, describe a medical intervention for antibiotic-resistant bacterial infections through the lense of optimal control theory, and discuss how to implement it in a clinical setting. Our findings underscore the necessity of considering virus inclusion in antibiotic resistance studies, shedding light on subtle yet influential dynamics in bacterial ecosystems.
Open AccessOptimal Control Strategies for Mitigating Antibiotic Resistance: Integrating Virus Dynamics for Enhanced Intervention Design
Given the global increase in antibiotic resistance, new effective strategies must be developed to treat bacteria that do not respond to first or second line antibiotics. One novel method uses bacterial phage therapy to control bacterial populations. Phage viruses replicate and infect bacterial cells and are regarded as the most prevalent biological agent on earth. This paper presents a comprehensive model capturing the dynamics of wild-type bacteria ( S ), antibiotic-resistant bacteria ( R ), and infective (I) strains, incorporating virus inclusion. Our model integrates biologically relevant parameters governing bacterial birth rates, death rates, and mutation probabilities and incorporates infection dynamics via contact with a virus. We employ an optimal control approach to study the influence of virus inclusion on bacterial population dynamics. Through numerical simulations, we establish insights into the stability of various system equilibria and bacterial population responses to varying infection rates. By examining the equilibria, we reveal the impact of virus inclusion on population trajectories, describe a medical intervention for antibiotic-resistant bacterial infections through the lense of optimal control theory, and discuss how to implement it in a clinical setting. Our findings underscore the necessity of considering virus inclusion in antibiotic resistance studies, shedding light on subtle yet influential dynamics in bacterial ecosystems.
- 22 May 2024
- Published by Springer Nature in Bulletin of Mathematical Biology
- Vol. 86 (7) , 1-24
A Model of Gastric Mucosal pH Regulation: Extending Sensitivity Analysis Using Sobol’ Indices to Understand Higher Moments
Several recent theoretical studies have indicated that a relatively simple secretion control mechanism in the epithelial cells lining the stomach may be responsible for maintaining a neutral (healthy) pH adjacent to the stomach wall, even in the face of enormous electrodiffusive acid transport from the interior of the stomach. Subsequent work used Sobol’ Indices (SIs) to quantify the degree to which this secretion mechanism is “self-regulating” i.e. the degree to which the wall pH is held neutral as mathematical parameters vary. However, questions remain regarding the nature of the control that specific parameters exert over the maintenance of a healthy stomach wall pH. Studying the sensitivity of higher moments of the statistical distribution of a model output can provide useful information, for example, how one parameter may skew the distribution towards or away from a physiologically advantageous regime. In this work, we prove a relationship between SIs and the higher moments and show how it can potentially reduce the cost of computing sensitivity of said moments. We define \(\gamma \)-indices to quantify sensitivity of variance, skewness, and kurtosis to the choice of value of a parameter, and we propose an efficient strategy that uses both SIs and \(\gamma \)-indices for a more comprehensive sensitivity analysis. Our analysis uncovers a control parameter which governs the “tightness of control” that the secretion mechanism exerts on wall pH. Finally, we discuss how uncertainty in this parameter can be reduced using expert information about higher moments, and speculate about the physiological advantage conferred by this control mechanism.- 22 May 2024
- Published by Springer Nature in Bulletin of Mathematical Biology
- Vol. 86 (7) , 1-24
A Model of Gastric Mucosal pH Regulation: Extending Sensitivity Analysis Using Sobol’ Indices to Understand Higher Moments
Several recent theoretical studies have indicated that a relatively simple secretion control mechanism in the epithelial cells lining the stomach may be responsible for maintaining a neutral (healthy) pH adjacent to the stomach wall, even in the face of enormous electrodiffusive acid transport from the interior of the stomach. Subsequent work used Sobol’ Indices (SIs) to quantify the degree to which this secretion mechanism is “self-regulating” i.e. the degree to which the wall pH is held neutral as mathematical parameters vary. However, questions remain regarding the nature of the control that specific parameters exert over the maintenance of a healthy stomach wall pH. Studying the sensitivity of higher moments of the statistical distribution of a model output can provide useful information, for example, how one parameter may skew the distribution towards or away from a physiologically advantageous regime. In this work, we prove a relationship between SIs and the higher moments and show how it can potentially reduce the cost of computing sensitivity of said moments. We define \(\gamma \)-indices to quantify sensitivity of variance, skewness, and kurtosis to the choice of value of a parameter, and we propose an efficient strategy that uses both SIs and \(\gamma \)-indices for a more comprehensive sensitivity analysis. Our analysis uncovers a control parameter which governs the “tightness of control” that the secretion mechanism exerts on wall pH. Finally, we discuss how uncertainty in this parameter can be reduced using expert information about higher moments, and speculate about the physiological advantage conferred by this control mechanism.- Open Access
- 10 June 2021
- Published by Oxford University Press (OUP) in Mathematical Medicine and Biology: A Journal of the IMA
- Vol. 38 (3) , 314-332
Data assimilation of synthetic data as a novel strategy for predicting disease progression in alopecia areata
The goal of patient-specific treatment of diseases requires a connection between clinical observations with models that are able to accurately predict the disease progression. Even when realistic models are available, it is very difficult to parameterize them and often parameter estimates that are made using early time course data prove to be highly inaccurate. Inaccuracies can cause different predictions, especially when the progression depends sensitively on the parameters. In this study, we apply a Bayesian data assimilation method, where the data are incorporated sequentially, to a model of the autoimmune disease alopecia areata that is characterized by distinct spatial patterns of hair loss. Using synthetic data as simulated clinical observations, we show that our method is relatively robust with respect to variations in parameter estimates. Moreover, we compare convergence rates for parameters with different sensitivities, varying observational times and varying levels of noise. We find that this method works better for sparse observations, sensitive parameters and noisy observations. Taken together, we find that our data assimilation, in conjunction with our biologically inspired model, provides directions for individualized diagnosis and treatments.
- 10 June 2021
- Published by Oxford University Press (OUP) in Mathematical Medicine and Biology: A Journal of the IMA
- Vol. 38 (3) , 314-332
Open AccessData assimilation of synthetic data as a novel strategy for predicting disease progression in alopecia areata
The goal of patient-specific treatment of diseases requires a connection between clinical observations with models that are able to accurately predict the disease progression. Even when realistic models are available, it is very difficult to parameterize them and often parameter estimates that are made using early time course data prove to be highly inaccurate. Inaccuracies can cause different predictions, especially when the progression depends sensitively on the parameters. In this study, we apply a Bayesian data assimilation method, where the data are incorporated sequentially, to a model of the autoimmune disease alopecia areata that is characterized by distinct spatial patterns of hair loss. Using synthetic data as simulated clinical observations, we show that our method is relatively robust with respect to variations in parameter estimates. Moreover, we compare convergence rates for parameters with different sensitivities, varying observational times and varying levels of noise. We find that this method works better for sparse observations, sensitive parameters and noisy observations. Taken together, we find that our data assimilation, in conjunction with our biologically inspired model, provides directions for individualized diagnosis and treatments.
- 1 July 2019
- Published by Elsevier in Mathematical Biosciences
- Vol. 313, 12-32
Enhanced disinfection of bacterial populations by nutrient and antibiotic challenge timing
- 1 July 2019
- Published by Elsevier in Mathematical Biosciences
- Vol. 313, 12-32
Enhanced disinfection of bacterial populations by nutrient and antibiotic challenge timing
- 15 January 2018
- Published by Elsevier in Journal of Membrane Science
- Vol. 546, 215-224
Uncertainty propagation in a model of dead-end bacterial microfiltration using fuzzy interval analysis
- 15 January 2018
- Published by Elsevier in Journal of Membrane Science
- Vol. 546, 215-224
Uncertainty propagation in a model of dead-end bacterial microfiltration using fuzzy interval analysis
- Open Access
- 2 December 2016
- Published by Oxford University Press (OUP) in FEMS Microbiology Letters
- Vol. 363 (23)
Theoretical and experimental evidence for eliminating persister bacteria by manipulating killing timing
Formation of a transient sub-population of bacteria, referred to as persisters, is one of the most important and least understood mechanisms that bacteria employ to evade elimination. Persister cells appear to be slow-growing bacteria that are broadly protected from a wide range of antibiotics. Using both theoretical and experimental methods, we show that alternating the application and withdrawal of antibiotics can be an effective treatment—as long as the timing of the protocol is estimated with precision. More specifically, we demonstrate that timing the alternating treatment based on theoretical predictions is confirmed using experimental observations. These results support a large class of theoretical studies that show that, even without complete understanding of the biological mechanisms, these models can provide insight into properties of the system.
- 2 December 2016
- Published by Oxford University Press (OUP) in FEMS Microbiology Letters
- Vol. 363 (23)
Open AccessTheoretical and experimental evidence for eliminating persister bacteria by manipulating killing timing
Formation of a transient sub-population of bacteria, referred to as persisters, is one of the most important and least understood mechanisms that bacteria employ to evade elimination. Persister cells appear to be slow-growing bacteria that are broadly protected from a wide range of antibiotics. Using both theoretical and experimental methods, we show that alternating the application and withdrawal of antibiotics can be an effective treatment—as long as the timing of the protocol is estimated with precision. More specifically, we demonstrate that timing the alternating treatment based on theoretical predictions is confirmed using experimental observations. These results support a large class of theoretical studies that show that, even without complete understanding of the biological mechanisms, these models can provide insight into properties of the system.