Machine learning and dengue forecasting: Comparing random forests and artificial neural networks for predicting dengue burdens at the national sub-national scale in Colombia

Abstract

29 30 31 32 33 34 35 36 2 37 Abstract:38 The robust estimate and forecast capability of random forests (RF) has been widely recognized, 39 however this ensemble machine learning method has not been widely used in mosquito-borne 40 disease forecasting. In this study, two sets of RF models were developed for the national and 41 departmental levels in Colombia to predict weekly dengue cases at 12-weeks ahead. A national 42 model based on artificial neural networks (ANN) was also developed and used as a comparator 43 to the RF models. The various predictors included historic dengue cases, satellite-derived 44 estimates for vegetation, precipitation, and air temperature, population counts, income inequality, 45 and education. Our RF model trained on the national data was more accurate for department-46 specific weekly dengue cases estimation compared to a local model trained only on the 47 department's data. Additionally, the forecast errors of the national RF model were smaller to 48 those of the national ANN model and were increased with the forecast horizon increasing from 49 one-week ahead (mean absolute error, MAE: 5.80; root mean squared error, RMSE: 11.10) to 50 12-weeks ahead (MAE: 13.38; RMSE: 26.82). There was considerable variation in the relative 51 importance of predictors dependent on forecast horizon. The environmental and meteorological 52 predictors were relatively important for short-term dengue forecast horizons while socio-53 demographic predictors were relevant for longer-term forecast horizons. This study showed the 54 potential of RF in dengue forecasting with also demonstrating the feasibility of using a national 55 model to forecast at finer spatial scales. Furthermore, sociodemographic predictors are important 56 to include to capture longer-term trends in dengue. 57 58 59 3 60 Author summary:61 Dengue virus has the highest disease burden of all mosquito-borne viral diseases, infecting 390 62 million people annually in 128 countries. Forecasting is an important warning mechanism that 63 can help with proactive planning and response for clinical and public health services. In this 64 study, we compare two different machine learning approaches to dengue forecasting: random 65 forest (RF) and neural networks (NN). National and local (departmental-level) models were 66 compared and used to predict dengue cases in the future. The results showed that the counts of 67 future dengue cases were more accurately estimated by RF than by NN. It was also shown that 68 environmental and meteorological predictors were more important for forecast accuracy for 69 shorter-term forecasts while socio-demographic predictors were more important for longer-term 70 forecasts. Finally, the national model applied to local data was more accurate in dengue 71 forecasting compared to the local model. This research contributes to the field of disease 72 forecasting and highlights different considerations for future forecasting studies. 73 74 75 76 77 78 79 80 81 4 82 Introduction83 Dengue virus is most prevalent of the mos...

LARIISA: soluções digitais inteligentes para apoio à tomada de decisão na gestão da Estratégia de Saúde da Família

“…Em um evento de potencial epidêmico observa-se o espalhamento viral em quatro fases bem definidas: (I) crescimento exponencial do número de infectados, (II) redução rápida da taxa de novos casos, (III) inversão na taxa de novos casos e (IV) redução exponencial do número de novos casos. Por se tratar de processo estocástico, naturalmente relacionamos diversos fatores correlacionados às epidemias de arboviroses, em especial medidas meteorológicas que influenciam diretamente à proliferação do vetor da doença: o mosquito 31,32 . Medidas populacionais, meteorológicas e de infecção agregadas são subsídios para que um conjunto de Redes Neurais Artificiais (RNA) predigam a quantidade de infectados que serão observados nas semanas subsequentes, o que possibilita recomendar e adotar medidas de prevenção e controle das doenças ou agravos relativos às arboviroses.…”