Emulator-based Bayesian calibration of the CISNET colorectal cancer models.
Emulator-based Bayesian calibration of the CISNET colorectal cancer models.
复制标题
基于模拟器的 CISNET 结直肠癌模型贝叶斯校准。
DOI：
10.1101/2023.02.27.23286525
复制
发表时间：
2024
期刊：
影响因子：
--
通讯作者：
Alarid-Escudero,Fernando
中科院分区：
文献类型：
--
作者：
Pineda-Antunez,Carlos;Seguin,Claudia;vanDuuren,LuukA;Knudsen,AmyB;Davidi,Barak;deLima,PedroNascimento;Rutter,Carolyn;Kuntz,KarenM;Lansdorp-Vogelaar,Iris;Collier,Nicholson;Ozik,Jonathan;Alarid-Escudero,Fernando
目的利用基于仿真器的贝叶斯算法对中国肿瘤干预与监测模型网S的自然病史结直肠癌的SimCRC、MISCAN-Colon和CRC-SPIN模型进行校正，并对模型预测结果与校正目标进行内部验证。方法采用拉丁超立方体抽样方法，对每个模型最多50,000个参数集进行采样并生成相应的输出。我们使用每个CISNET-CRC模型的输入和输出样本来训练多层感知器人工神经网络(ANN)作为仿真器。我们选择了具有相应超参数(即隐含层数、节点数、激活函数、历元和优化器)的ANN结构，以最小化验证样本上的预测均方误差。我们用概率编程语言实现了神经网络仿真器，并用基于哈密顿蒙特卡罗的算法对输入参数进行了校准，得到了CISNET-CRC模型参数的联合后验分布。结果SimCRC的最优ANN有4个隐含层和360个隐含节点，MISCAN-COLON有4个隐含层和114个隐含节点，CRC-SPIN有1个隐含层和140个隐含节点。对于SimCRC、MISCAN-COLON和CRC-SPIN，训练和校准仿真器的总时间分别为7.30、4.0和0.66 h。模型预测输出的均值落在校准目标的95%可信区间内，SimCRC为110个，MISCAN为93个，CRC-SPIN为31个。结论使用人工神经网络仿真器是一种实用的解决方案，可以降低用于政策分析的单级仿真模型的贝叶斯校正的计算负担和复杂性。在这项工作中，我们提供了一步一步的指南来构建仿真器来使用贝叶斯方法来校准3个真实的CRC个体级模型。亮点我们使用人工神经网络(ANN)来构建仿真器来替代复杂的基于个体的模型来减少贝叶斯校准过程中的计算负担。ANN在仿真CISNET-CRC微观仿真模型方面表现出了良好的性能，尽管有许多输入参数和输出，但使用人工神经网络仿真器是一种实用的解决方案，可以降低用于政策分析的个体级仿真模型的贝叶斯校准的计算负担和复杂性。本工作旨在支持健康决策科学家在贝叶斯框架下量化计算密集型仿真模型校准参数的不确定性。
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 (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.ResultsThe 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.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, 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.HighlightsWe 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.