The Science and Technology Directorate and the National Biosurveillance Integration Center, both offices within the U.S. Department of Homeland Security, unveiled five finalists for the first stage of their Hidden Signals Challenge Feb. 14.
The Hidden Signals Challenge invited IT developers and researchers to develop early warning systems that use existing datasets to identify emerging biothreats. A biothreat represents a harmful pathogen — released either naturally or deliberately — that poses a risk to national security and public health.
A panel of judges with backgrounds in bioinformatics, biological defense, epidemiology and emergency management helped the Science and Technology Directorate and the National Biosurveillance Integration Center select the finalists for stage one of the challenge.
Here are the five finalists and the early warning systems they created.
- Commuter Pattern Analysis for Early Biothreat Detection (Tacoma, Wash.-based Readiness Acceleration & Innovation Network). This system detects possible disease outbreaks by recognizing commuter absenteeism. To detect low levels of commuters, the system cross-references de-identified traffic information with municipal health data and internet keyword searches.
- Monitoring emergency department wait times to detect emergent influenza pandemics (Emeryville, Calif.-based Vituity). This model alerts authorities to spikes in ER wait times attributed to emergent flu pandemics. It uses real-time data from 142 hospitals in 19 states to identify these spikes.
- One Health Alert System (Cabarrus County, N.C.-based team led by William Pilkington). This model flags emergent disease outbreaks based on an analysis of a daily report of top disease symptoms in North Carolina.
- Pandemic Pulse (Computational Epidemiology Lab at Boston Children's Hospital). This tool detects potential biothreats by pinpointing anomalies in six integrated data streams: Twitter, Google Search, transportation, news, HealthMap and Flu Near You.
- Pre-syndromic Surveillance (Pittsburgh-based team of Daniel B. Neill and Mallory Nobles). This machine learning system detects emerging clusters of rare diseases by analyzing real-time ER chief complaint data alongside social media and news data.
Each of the five finalists will receive $20,000 and advance to stage two of the challenge. During the second stage, the finalists will develop these early warning concepts into system designs with the help of a team of mentors.
The panel of judges will select a winner, who will receive a $200,000 grand prize, in spring 2018.