Oak Ridge Biosurveillance Toolkit: Scalable machine learning for public health surveillance

As the number of data sources for public health surveillance continues to grow both in volume and variety, there is a need to develop data-driven machine learning tools that can automate discovery and aid decision makers in obtaining quantifiable insights on emerging disease spread phenomena. In this talk, we present an overview of scalable machine learning tools that we have been developing as part of advancing this mission. In particular, our machine learning tools can automatically (a) detect multi-scale spatial and temporal break-out patterns of disease occurrence, (b) quantify multi-modal co-occurrence disease patterns to identify local-and national-level “hotspots” and (c) predict how patterns of co-occurrence correspond to `intervention' strategies.