Forecasting COVID-19 spreading through an ensemble of classical and machine learning models: Spain’s case study

Abstract

In this work the applicability of an ensemble of population and machine learning models to predict the evolution of the COVID-19 pandemic in Spain is evaluated, relying solely on public datasets. Firstly, using only incidence data, we trained machine learning models and adjusted classical ODE-based population models, especially suited to capture long term trends. As a novel approach, we then made an ensemble of these two families of models in order to obtain a more robust and accurate prediction. We then proceed to improve machine learning models by adding more input features: vaccination, human mobility and weather conditions. However, these improvements did not translate to the overall ensemble, as the different model families had also different prediction patterns. Additionally, machine learning models degraded when new COVID variants appeared after training. We finally used Shapley Additive Explanation values to discern the relative importance of the different input features for the machine learning models’ predictions. The conclusion of this work is that the ensemble of machine learning models and population models can be a promising alternative to SEIR-like compartmental models, especially given that the former do not need data from recovered patients, which are hard to collect and generally unavailable.

A dynamic ensemble model for short-term forecasting in pandemic situations

“…Our best model - evaluated on the French COVID-19 hospitalization dataset - achieved a mean MAPE of around 17% and around 20% to 23% (for the meta-model test windows or all test windows) on country-level and regional aggregated country-level data, respectively. Heredia Cacha et al [10] forecasted COVID-19 cases in Spain using different ensemble methods (mean, median, weighted average) and documented a mean MAPE of around 30% for a 14-day forecasting horizon. We achieved a MAPE of around 17% to 25% for forecasting the number of COVID-19 cases in Germany and France.…”

“…Since each modeling technique unavoidably comes along with its own assumptions and limitations, ensemble models have been proposed for forecasting the spread of infectious diseases like Influenza [6–8] or Ebola [9] and later for COVID-19 [10,11]. In principle, ensemble models can be understood as a collection of rather simplistic base models, which all produce an output based on each model’s assumption plus an algorithm or meta-model that combines them into one ensemble output.…”

“…The advantage of such an ensemble approach is that the bias of the individual models is reduced, making the final output more robust [12]. In the literature, such ensemble methods often use the mean (e.g., [11]) or median (e.g., [10]) of the base model outputs. However, pandemics like COVID-19 are dynamic, there are times when the number of cases barely changes, there is exponential growth and decay, and there are turning points of waves, which can all depend on external factors like interventions [13,14], people’s behavior [15], seasonality [16,17], or variants of concern [18,19].…”

An Overview of Forecast Analysis with ARIMA Models during the COVID-19 Pandemic: Methodology and Case Study in Brazil

“…Mathematically-based models such as the modified SEIR and long short-term memory structures were considered in [37,38]. Moreover, in [39], a fusion of ML and mathematical models was assessed to enhance the precision and robustness of near-future COVID-19 pandemic predictions. This fusion involved integrating random forest, gradient boosting, k-nearest neighbors, and kernel ridge regression ML models with Bertalanffy, Gompertz, logistic, and Richards models.…”

Inference Based on the Stochastic Expectation Maximization Algorithm in a Kumaraswamy Model with an Application to COVID-19 Cases in Chile

“…The impact of the pandemic extended beyond public health, affecting aspects such as the economy and gross domestic product, especially in Latin America [41]. Different models have proven to be relevant in forecasting epidemic diseases, and various methodologies have been employed to gain insights into diverse aspects of epidemics around the world [42][43][44][45][46][47][48][49][50][51][52][53][54][55][56][57][58][59][60].…”