Data-driven multiscale modelling and analysis of COVID-19 spatiotemporal evolution using explainable AI

Abstract

To quantificationally identify the optimal control measures for regulators to best minimize COVID-19’s growth (G-rate) and death (D-rate) rates in today's context, this paper develops a top-down multiscale engineering approach which encompasses a series of systematic analyses, namely: (global scale) predictive modelling of G-rate and D-rate due to COVID-19 globally, followed by determining the most effective control factors which can best minimize both parameters over time via explainable AI with SHAP (SHapley Additive exPlanations) method; (continental scale) same predictive forecasting of G-rate and D-rate in all continents, followed by performing explainable SHAP analysis to determine the most effective control factors for the respective continents; and (country scale) clustering the different countries (> 150 in total) into 3 main clusters to identify the universal set of effective control measures. By using the historical period between 2 May 2020 and 1 Oct 2021, the average MAPE scores for forecasting G-rate and D-rate are within 10%, or less on average, at the global and continental scales. Systematically, we have quantificationally demonstrated that the top 3 most effective control measures for regulators to best minimize G-rate universally are COVID-CONTACT-TRACING, PUBLIC-GATHERING-RULES, and COVID-STRINGENCY-INDEX, while the control factors relating to D-rate depend on the modelling scenario.

Effectiveness of testing, contact tracing and isolation interventions among the general population on reducing transmission of SARS-CoV-2: a systematic review

“…The overwhelming majority of these were transmission model-based studies, with only 25 empirical studies meeting our criteria of reporting the real-world impact of testing and/or contact tracing together with some adjustments for confounding factors such as changes to other control measures or population characteristics. Of these 25 studies, 11 adopted a broad statistical approach and attempted to link coarse classification of control measures in multiple countries to their epidemiological dynamics [56,57,60,63,64,66,[70][71][72]75,78]; five considered detailed contact tracing data from either England [65,74,76] or Colombia [59,73]; four considered strategies for isolation after testing or notification as contacts [61,[67][68][69]; two considered within-country stringency of TTI-type controls in China [58] and South Korea [62]; with other papers focusing on the impact of mass-testing [10,77] and weekly testing of people without symptoms [20].…”

“…(ii) Testing strategies (12 papers) Nine papers performed statistical analyses across multiple countries to assess the impact of changing patterns of control [57,63,64,66,[70][71][72]75,78], while three examined testing strategies in single countries [10,20,77]. Many of the cross-country studies used the OxCGRT [83] to inform the type and strength of epidemic controls in each country over time.…”

“…One included study was an individually randomized controlled trial [67]. Eleven studies used observational time-series designs [56,57,60,62,64,[70][71][72][73]75,78]. These compared variation across, for instance, geographical time units while variously adjusting for characteristics of the units, other interventions and features of the epidemic.…”

“…Chan et al [56] Nam et al [69] Malheiro et al [68] Love et al [67] Lopez et al [61] Zhang et al [77] Zamanzadeh & Cavoli [78] Yalaman et al [75] Ssentongo et al [72] Spiliopoulos [71] Pozo-Martin et al [70] Pavelka et al [10] Leffler et al [66] Hong et al [63] Haug et al [64] Gorji et al [20] Chew et al [57] Wymant et al [74] Vecino-Ortiz et al [73] Kendall et al [65] Gianino et al [60] Fetzer et al [76] Fernandez-Nino et al [59] Deng et al [58] contact tracing testing strategies isolation package…”

Artificial Intelligence-Based Machine Learning toward the Solution of Climate-Friendly Hydrogen Fuel Cell Electric Vehicles

“…Through learning the introduced data and improving the algorithms that are embedded in AI-based technologies, a fundamental transformation in the modelling and simulation mindset was reached. There have been various applications of AI used in different industries, such as energy [64][65][66][67][68][69][70], transportation [71], medicine [72][73][74][75], and various other natural sciences [76][77][78]. Furthermore, the use and implementation of traditional modelling methods have been enhanced by collaborating with AI-based machine learning tools [79][80][81].…”