Unlocking the Predictive Power of Heterogeneous Data to Build an Operational Dengue Forecasting System

Abstract

<jats:p>Predicting an infectious disease can help reduce its impact by advising public health interventions and personal preventive measures. While availability of heterogeneous data streams and sensors such as satellite imagery and the Internet have increased the opportunity to indirectly measure, understand, and predict global dynamics, the data may be prohibitively large and/or require intensive data management while also requiring subject matter experts to properly exploit the data sources (e.g., deriving features from fundamentally different data sets). Few efforts have quantitatively assessed the predictive benefit of novel data streams in comparison to more traditional data sources, especially at fine spatio-temporal resolutions. We have combined multiple traditional and non-traditional data streams (satellite imagery, Internet, weather, census, and clinical surveillance data) and assessed their combined ability to predict dengue in Brazil's 27 states on a weekly and yearly basis over seven years. For each state, we nowcast dengue based on several time series models, which vary in complexity and inclusion of exogenous data. We also predict yearly cumulative risk by municipality and state. The top-performing model and utility of predictive data varies by state, implying that forecasting and nowcasting efforts in the future may be made more robust by and benefit from the use of multiple data streams and models.</jats:p>

Using heterogeneous data to identify signatures of dengue outbreaks at fine spatio-temporal scales across Brazil

“…Our spatial clustering results can provide direction in how to incorporate current and historical spatial information in an informed and efficient way. Finally, our results also revealed that broad capturing and integration of disparate data streams are beneficial for explaining disease dynamics, and their utility in forecasting systems should continue to be explored [ 9 , 10 ]. Here, we found that an individual time series data stream, such as temperature, can be used in numerous ways, whether as a standard summary statistic (e.g., mean), or as more complex dynamic measurements (e.g., intensity).…”

“…Environmental data. From a previously published environmental data set for Brazil [9,10], we collected time series of five satellite remote sensing indices (Table 1: green Normalized Difference Water Index (green NDWI), short-wave infrared Normalized Difference Water Index (SWIR NDWI), Normalized Burn Ratio (NBR), Normalized Difference Vegetation Index (NDVI), and % cloudy pixels), and three climatic weather variables (Table 1: daily range in temperature, mean temperature, and relative humidity). Satellite remote sensing data from January 2010 to December 2016 were derived from the multispectral satellites Landsat 5, Landsat 7, Landsat 8, and Sentinel-2, and the source images were accessed via the Descartes Labs Platform [31].…”

“…To enhance public health preparedness, there have been several efforts to develop forecasting models of DENV in Brazil [7][8][9][10]. Forecasting models integrate single or multiple data streams into statistical or mechanistic frameworks to make short-term (e.g., present to multiple weeks ahead) or long-term (e.g., monthly and year-based) predictions of the number of DENV cases in a given location [11][12][13][14][15][16][17][18].…”