U.S. Department of Homeland Security
Hidden Signals Challenge

This is the third post in our Q&A series featuring the five Hidden Signals Challenge finalists. These solutions demonstrate the exciting potential for open data to help identify biothreats in real time.

Our thirdQ&A features finalists William Pilkington, Angi English, Merideth Bastiani, and Steven Polunsky.Their symptoms database, One Health Alert System (OHAS), analyzes the Daily Disease Report’s top ten symptoms as seen by 43 health care providers in North Carolina. The model flags disease outbreak using textual predictive analytics and accounts for seasonal rates of change.

“Although real-time analysis of potential biothreats has been actively explored for the past decade, very few models have proven to be successful as real-time data analytic models, and none have considered the intersection between human, animal and environmental factors in the way that the OHAS solution proposes to.”

– One Health Alert System

What inspired you to submit this concept?

The One Health Alert System team is comprised of four experts representing diverse disciplines who have a shared passion to tackle wicked problems in homeland security. As recipients of advanced degrees in the field of homeland security and other disciplines, and founding scholars of HSx- Advanced Thinking in Homeland Security at the Center for Homeland Defense and Security, we were naturally predisposed to evaluating the situation from a multidisciplinary perspective. From the onset, our group fundamentally understood that early identification of human vector biothreats offers significant benefits in terms of the opportunity for alerts and interventions, but also passionately believed, that a human-only approach would be insufficient to address the breadth of the problem. Empowered by the knowledge and data reported via the Daily Disease Report (DDR), which has productively influenced public health in counties in North Carolina, the team developed the OHAS concept. The OHAS concept is focused on broadening the list of critical factors to be considered when developing a more comprehensive assessment of nascent biothreats. These factors can be integrated into evidence-based notification and mitigation practices in the public health field.

How will your concept enable city-level operators to make critical and proactive decisions?

The OHAS will collect, compile and analyze data from key public and private sources to identify and confirm patterns that indicate risk in near real time, allowing city-level operators to implement protective measures earlier, and with a higher degree of confidence than any method currently available. By drawing information from a variety of sources, and compiling them on a user-friendly dashboard with analytic capabilities that will adjust over time, based on machine learning and user feedback, the system is designed to be used for steady-state monitoring with alert capabilities to bring immediate attention to outlying events that warrant review, in order to determine next steps in accordance with established procedures. The OHAS dashboard will include features such as text and email alerts, the ability for users to customize thresholds for notification, and the use of plain language for notification purposes. For entities with established management and warning systems, OHAS output would be available as a data feed to avoid duplication or navigation of multiple systems, or to run in parallel to those systems as a planned redundancy.

What sets your concept apart from existing solutions?

Although real-time analysis of potential biothreats has been actively explored for the past decade, very few models have proven to be successful as real-time data analytic models, and none have considered the intersection between human, animal and environmental factors in the way that the OHAS solution proposes to. In addition, the known success of and compliance with the DDR requirements in the pilot communities offers a premiere test environment through which to test the efficacy of applying machine learning when integrating environmental and animal vector information in to the model without the costs or challenges of relying on external partners to provide the primary dataset. Further, our model reflects the lessons learned from the latest research in detection via social media, in both health and non-health fields.

How did you see your concept evolving during Stage 2 of the Challenge?

The initial proposal for OHAS focused on use of the North Carolina Daily Disease Report (DDR) for patient level chief complaint data, and proposed replication of the DDR to implement the OHAS system in other regions. After evaluating cost and implementation challenges, the OHAS team has identified a series of alternative options for obtaining chief complaint information, including interfaces for gathering information directly from Electronic Medical Record (EMR) systems. Following analysis of the availability and value of open source data related to human, animal, and environmental signals, the OHAS approach will be refined to target the most valuable search information available in open source platforms, and to identify and integrate government systems that contain real-time data. Obtaining data at low, or no cost, will continue to be a priority for the OHAS model. However, when critical data sets are proprietary, and necessary public-private partnerships cannot be established to accomplish this goal, the OHAS approach would include a metric by which to evaluate the return on investment related to purchasing or subscribing to datasets.

What is the biggest insight you’ve uncovered through the Challenge thus far?

Good ideas and design components are the ones that can weather critical evaluation, but they may not necessarily present, or perform as initially expected; additionally, weak signals will extinguish themselves via the same process. This insight has heavily influenced that evaluation and integration of the lessons learned from attempts made by other entities to accomplish near real-time analysis of social media and other available data. Through this process, it has become clear that the success of OHAS hinges on having a solid foundation for human health information that does not rely on manual input. While the OHAS design will maintain human symptomatology at the center of system design, the team has determined that replication of the DDR is not a viable option for obtaining and maintaining this data set. Instead, OHAS must draw symptom information from existing, established data sources, rather than relying on creating new data elements or collection methods. In response, the team has shifted the design approach at a critical time in the process, and this insight has helped to establish a metric through which the team can evaluate the potential of other data sources, including those representing the animal and environmental vectors.

Learn more about other finalists’ solutions on the Challenge blog and subscribe to our newsletter to continue receiving updates. The winner(s), who will receive up to $200,000 in cash prizes, will be announced later this spring.