Characterization and Valuation of the Uncertainty of Calibrated Parameters in Microsimulation Decision Models

Alarid‐Escudero, Fernando; Knudsen, Amy B.; Ozik, Jonathan; Collier, Nicholson; Kuntz, Karen M.

doi:10.3389/fphys.2022.780917

Cited by 5 publications

(4 citation statements)

References 59 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Highly correlated parameters are often found in these types of simulation models with a high number of parameters. 2…”

Section: Resultsmentioning

confidence: 99%

“…For many of these models, data to directly inform the parameters are lacking and often estimated through calibration. Ideally, the uncertainty in model parameters should be accurately quantified during the calibration process so that the ultimate model predictions also reflect this uncertainty (2). Therefore, the calibration process aims to obtain a joint posterior distribution of the calibrated parameters that produce model outputs consistent with the calibration targets (observed data arising from a clinical or epidemiological system) and their uncertainty (3).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Emulator-Based Bayesian Calibration of the CISNET Colorectal Cancer Models

Pineda-Antunez,

Seguin,

van Duuren

et al. 2024

Med Decis Making

Self Cite

View full text Add to dashboard Cite

Purpose To calibrate Cancer Intervention and Surveillance Modeling Network (CISNET)’s SimCRC, MISCAN-Colon, and CRC-SPIN simulation models of the natural history colorectal cancer (CRC) with an emulator-based Bayesian algorithm and internally validate the model-predicted outcomes to calibration targets. Methods We used Latin hypercube sampling to sample up to 50,000 parameter sets for each CISNET-CRC model and generated the corresponding outputs. We trained multilayer perceptron artificial neural networks (ANNs) as emulators using the input and output samples for each CISNET-CRC model. We selected ANN structures with corresponding hyperparameters (i.e., number of hidden layers, nodes, activation functions, epochs, and optimizer) that minimize the predicted mean square error on the validation sample. We implemented the ANN emulators in a probabilistic programming language and calibrated the input parameters with Hamiltonian Monte Carlo–based algorithms to obtain the joint posterior distributions of the CISNET-CRC models’ parameters. We internally validated each calibrated emulator by comparing the model-predicted posterior outputs against the calibration targets. Results The optimal ANN for SimCRC had 4 hidden layers and 360 hidden nodes, MISCAN-Colon had 4 hidden layers and 114 hidden nodes, and CRC-SPIN had 1 hidden layer and 140 hidden nodes. The total time for training and calibrating the emulators was 7.3, 4.0, and 0.66 h for SimCRC, MISCAN-Colon, and CRC-SPIN, respectively. The mean of the model-predicted outputs fell within the 95% confidence intervals of the calibration targets in 98 of 110 for SimCRC, 65 of 93 for MISCAN, and 31 of 41 targets for CRC-SPIN. Conclusions Using ANN emulators is a practical solution to reduce the computational burden and complexity for Bayesian calibration of individual-level simulation models used for policy analysis, such as the CISNET CRC models. In this work, we present a step-by-step guide to constructing emulators for calibrating 3 realistic CRC individual-level models using a Bayesian approach. Highlights We use artificial neural networks (ANNs) to build emulators that surrogate complex individual-based models to reduce the computational burden in the Bayesian calibration process. ANNs showed good performance in emulating the CISNET-CRC microsimulation models, despite having many input parameters and outputs. Using ANN emulators is a practical solution to reduce the computational burden and complexity for Bayesian calibration of individual-level simulation models used for policy analysis. This work aims to support health decision scientists who want to quantify the uncertainty of calibrated parameters of computationally intensive simulation models under a Bayesian framework.

show abstract

“…Highly correlated parameters are often found in these types of simulation models with a high number of parameters. 2…”

Section: Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Emulator-Based Bayesian Calibration of the CISNET Colorectal Cancer Models

Pineda-Antunez,

Seguin,

van Duuren

et al. 2024

Med Decis Making

Self Cite

View full text Add to dashboard Cite

show abstract

“…The joint posterior distributions of all pairwise parameters for the three CISNET-CRC models are shown in Figure S4 in the supplementary material. Highly correlated parameters are often found in these types of simulation models with a high number of parameters (2)…”

Section: Resultsmentioning

confidence: 99%

Emulator-based Bayesian calibration of the CISNET colorectal cancer models

Pineda-Antúnez¹,

Seguin

Duuren

et al. 2023

Preprint

Self Cite

View full text Add to dashboard Cite

PurposeTo calibrate Cancer Intervention and Surveillance Modeling Network (CISNET) ‘s SimCRC, MISCAN-Colon, and CRC-SPIN simulation models of the natural history colorectal cancer (CRC) with an emulator-based Bayesian algorithm and internally validate the model-predicted outcomes to calibration targets.MethodsWe used Latin hypercube sampling to sample up to 50,000 parameter sets for each CISNET-CRC model and generated the corresponding outputs. We trained multilayer perceptron artificial neural networks (ANN) as emulators using the input and output samples for each CISNET-CRC model. We selected ANN structures with corresponding hyperparameters (i.e., number of hidden layers, nodes, activation functions, epochs, and optimizer) that minimize the predicted mean square error on the validation sample. We implemented the ANN emulators in a probabilistic programming language and calibrated the input parameters with Hamiltonian Monte Carlo-based algorithms to obtain the joint posterior distributions of the CISNET-CRC models’ parameters. We internally validated each calibrated emulator by comparing the model-predicted posterior outputs against the calibration targets.ResultsThe optimal ANN for SimCRC had four hidden layers and 360 hidden nodes, MISCAN-Colon had 4 hidden layers and 114 hidden nodes, and CRC-SPIN had one hidden layer and 140 hidden nodes. The total time for training and calibrating the emulators was 7.3, 4.0, and 0.66 hours for SimCRC, MISCAN-Colon, and CRC-SPIN, respectively. The mean of the model-predicted outputs fell within the 95% confidence intervals of the calibration targets in 98 of 110 for SimCRC, 65 of 93 for MISCAN, and 31 of 41 targets for CRC-SPIN.ConclusionsUsing ANN emulators is a practical solution to reduce the computational burden and complexity for Bayesian calibration of individual-level simulation models used for policy analysis, like the CISNET CRC models.

show abstract

“…The IMIS algorithm has previously been used to calibrate cancer health policy models. (22,26) The parameters concerned HPV transmission and cervical cancer natural history, such as progression and regression rates (Supplementary Material). To generate projections for all primary outcomes for all scenarios, we sampled 1,000 parameter sets from the posterior distribution.…”

Section: Methodsmentioning

confidence: 99%

State-level disparities in cervical cancer prevention and outcomes in the U.S.: A modeling study

Alarid-Escudero,

Gracia,

Wolf

et al. 2024

Preprint

Self Cite

View full text Add to dashboard Cite

BackgroundDespite the availability of HPV vaccines for over a decade, coverage across the United States (US) is varied. While some states have made concerted efforts to increase HPV vaccination coverage, most model-based analyses have estimated vaccine impact on the US. We estimated the impact of hypothetical changes in HPV vaccination coverage at the state level for three states with varying levels of HPV vaccination coverage and cervical cancer incidence (California, New York, Texas) using a mathematical model.MethodsWe developed a new mathematical model of HPV transmission and cervical cancer tailored to state-level cancer incidence and mortality. We quantified the public health impact of increasing HPV vaccination coverage to 80% by 2025 or 2030 and the effect on time to elimination in the three states.ResultsIncreasing vaccination coverage to 80% in Texas in 10 years could reduce cervical cancer incidence by 50.9% (95%-CrI: 46.6-56.1%) by 2100. In New York and California, achieving the same coverage could reduce incidence by 27.3% (95%-CrI: 23.9-31.5%) and 24.4% (95%-CrI: 20.0-30.0%), respectively. Achieving 80% coverage in 5 years will slightly increase the reduction. If 2019 vaccination coverage continues, cervical cancer elimination would be reached in the US by 2051 (95%-Crl: 2034-2064). However, the timeline by which individual states reach elimination could vary by decades.ConclusionAchieving an HPV vaccination coverage target of 80% by 2030 will benefit states with low vaccination coverage and high cervical cancer incidence the most. Our results highlight the value of more geographically focused analyses to inform priorities.

show abstract

Characterization and Valuation of the Uncertainty of Calibrated Parameters in Microsimulation Decision Models

Cited by 5 publications

References 59 publications

Emulator-Based Bayesian Calibration of the CISNET Colorectal Cancer Models

Emulator-Based Bayesian Calibration of the CISNET Colorectal Cancer Models

Emulator-based Bayesian calibration of the CISNET colorectal cancer models

State-level disparities in cervical cancer prevention and outcomes in the U.S.: A modeling study

Contact Info

Product

Resources

About