Departments of Electrical and Computer Engineering, Center of Excellence on Soft Computing and Intelligent Information Processing (SCIIP), Ferdowsi University of Mashhad, Mashhad, Iran.
Department of Electrical and Computer Engineering, Semnan University, Semnan, Iran.
Comput Methods Programs Biomed. 2023 Apr;232:107443. doi: 10.1016/j.cmpb.2023.107443. Epub 2023 Feb 24.
Besides efforts on vaccine discovery, robust and intuitive government policies could also significantly influence the pandemic state. However, such policies require realistic virus spread models, and the major works on COVID-19 to date have been only case-specific and use deterministic models. Additionally, when a disease affects large portions of the population, countries develop extensive infrastructures to contain the condition that should adapt continuously and extend the healthcare system's capabilities. An accurate mathematical model that reasonably addresses these complex treatment/population dynamics and their corresponding environmental uncertainties is necessary for making appropriate and robust strategic decisions.
Here, we propose an interval type-2 fuzzy stochastic modeling and control strategy to deal with the realistic uncertainties of pandemics and manage the size of the infected population. For this purpose, we first modify a previously established COVID-19 model with definite parameters to a Stochastic SEIAR (SEIAR) approach with uncertain parameters and variables. Next, we propose to use normalized inputs, rather than the usual parameter settings in the previous case-specific studies, hence offering a more generalized control structure. Furthermore, we examine the proposed genetic algorithm-optimized fuzzy system in two scenarios. The first scenario aims to keep infected cases below a certain threshold, while the second addresses the changing healthcare capacities. Finally, we examine the proposed controller on stochasticity and disturbance in parameters, population sizes, social distance, and vaccination rate.
The results show the robustness and efficiency of the proposed method in the presence of up to 1% noise and 50% disturbance in tracking the desired size of the infected population. The proposed method is compared to Proportional Derivative (PD), Proportional Integral Derivative (PID), and type-1 fuzzy controllers. In the first scenario, both fuzzy controllers perform more smoothly despite PD and PID controllers reaching a lower mean squared error (MSE). Meanwhile, the proposed controller outperforms PD, PID, and the type-1 fuzzy controller for the MSE and decision policies for the second scenario.
The proposed approach explains how we should decide on social distancing and vaccination rate policies during pandemics against the prevalent uncertainties in disease detection and reporting.
除了疫苗研发工作外，强有力且直观的政府政策也可以显著影响大流行状态。然而，此类政策需要实际的病毒传播模型，而迄今为止针对 COVID-19 的主要工作仅针对特定病例，并使用确定性模型。此外，当疾病影响到大部分人群时，各国会开发广泛的基础设施来控制这种情况，这需要不断适应并扩展医疗保健系统的能力。需要一个准确的数学模型来合理处理这些复杂的治疗/人口动态及其相应的环境不确定性，以便做出适当且稳健的战略决策。
在这里，我们提出了一种区间型 2 模糊随机建模和控制策略，以应对大流行的现实不确定性并管理感染人群的规模。为此，我们首先将先前建立的具有确定参数的 COVID-19 模型修改为具有不确定参数和变量的随机 SEIAR（SEIAR）方法。接下来，我们提议使用归一化输入，而不是之前特定病例研究中的常用参数设置，从而提供更通用的控制结构。此外，我们在两种情况下检查了所提出的遗传算法优化模糊系统。第一种情况旨在将感染病例保持在一定阈值以下，而第二种情况则针对不断变化的医疗保健能力。最后，我们检查了所提出的控制器在参数、人口规模、社交距离和疫苗接种率的随机性和干扰方面的性能。
结果表明，在所提出的方法中，即使在 1%的噪声和 50%的干扰下，该方法也具有很强的鲁棒性和效率，可以跟踪感染人群的期望规模。将所提出的方法与比例微分（PD）、比例积分微分（PID）和 1 型模糊控制器进行了比较。在第一种情况下，尽管 PD 和 PID 控制器达到了较低的均方误差（MSE），但两个模糊控制器的性能都更加平滑。同时，在所提出的控制器在第二个场景的 MSE 和决策策略方面优于 PD、PID 和 1 型模糊控制器。
该方法解释了在疾病检测和报告中普遍存在不确定性的情况下，我们应该如何决定大流行期间的社交距离和疫苗接种率政策。
