Insights about Health and Well-Being from Large-Scale Logs

Monday, 17 February 2014
Grand Ballroom A (Hyatt Regency Chicago)
Eric Horvitz , Microsoft Research, Redmond, WA
Large-scale anonymized logs of human activities are becoming available via search engines, cellular communication systems, social media, and databases of news events.  I will discuss several efforts on using machine learning and inference to transform data drawn from logs of behavioral data and other data sources into sensor networks for public health. I will discuss several recent projects, including performing pharmacovigilance from search logs, predicting postpartum changes in affect for new mothers from social media, tracking the influence of earthquakes from data drawn from wireless communications, and predicting cholera epidemics from news corpora.