Annu Int Conf IEEE Eng Med Biol Soc. 2021 Nov;2021:1674-1677. doi: 10.1109/EMBC46164.2021.9629653.
Nowadays, there is a growing need for the development of computationally efficient virtual population generators for large-scale in-silico clinical trials. In this work, we utilize the Gaussian Mixture Models (GMM) with variational Bayesian inference (BGMM) using robust estimations of Dirichlet concentration priors for the generation of virtual populations. The estimations were based on an exponential transformation of the number of Gaussian components. The proposed method was compared against state-of-the-art virtual data generators, such as, the Bayesian networks, the supervised tree ensembles (STE), the unsupervised tree ensembles (UTE), and the artificial neural networks (ANN) towards the generation of 20000 virtual patients with hypertrophic cardiomyopathy (HCM). Our results suggest that the proposed BGMM can yield virtual distributions with small inter- and intra-correlation difference (0.013 and 0.012), in lower execution time (4.321 sec) than STE which achieved the second-best performance.
如今，对于开发用于大规模计算机临床试验的计算效率高的虚拟人群生成器，需求日益增长。在这项工作中，我们利用具有变分贝叶斯推断（BGMM）的高斯混合模型（GMM），使用 Dirichlet 浓度先验的稳健估计来生成虚拟人群。这些估计是基于高斯成分数量的指数变换。将所提出的方法与最先进的虚拟数据生成器（例如贝叶斯网络、有监督树集成（STE）、无监督树集成（UTE）和人工神经网络（ANN））进行了比较，以生成 20000 名肥厚型心肌病（HCM）虚拟患者。我们的结果表明，所提出的 BGMM 可以生成具有较小的内部和内部相关性差异（0.013 和 0.012）的虚拟分布，执行时间（4.321 秒）比 STE 短，后者的性能排名第二。
