Analytics in Action: How analytics helps contain Ebola
By Douglas A. Samuelson and Brian Umana (l-r)
The outbreak of Ebola disease in West Africa, with some cases spreading elsewhere, has been a source of concern and frustration throughout the world. This is the story about how a small data science and analytics company found a way to help contain the spread of the virus.
In September 2014, Brian Umana, a consultant at Illumina Consulting Group (ICG), had a conversation with company CEO David Waldrop about the Ebola epidemic. They each expressed concern about the scale of the human suffering, but also about facets of security and public health that could limit the spread of the epidemic. Given the nature of their earlier work in real-time streaming data analytics, they eventually began speaking about various ways one might apply data science and analytics to the problem. They were not the first to do this, nor in any way did they take the most dramatic or significant action. However, within days of the conversation, ICG began to make a small contribution to safety in the U.S. response to the “maritime vector” of the Ebola virus.
Waldrop built a new-use case for LUX software, ICG’s principal product, a high-volume data streaming and pattern recognition tool. What Waldrop had noticed was that detection and prevention measures for potential maritime- and port-based spread of the Ebola virus appeared less rigorous than corresponding detection and prevention measures used in air travel. No one would dispute that it is essential to monitor airports where passengers are arriving from Ebola-stricken countries as a measure for containing the spread of the disease. West African seaports and merchant shipping, however, are another potential means for spreading the disease abroad, a topic that has drawn far less attention.
Waldrop used the software to draw an area of interest off the coast of West Africa, enclosed by a second area of interest encompassing all the water outside of the first area but still inside of a specific radius of the West African coast. Waldrop instructed LUX that any time a boat crossed from Area 1 (the port area) to Area 2 (farther from the coast) and then exited Area 2 to create a map-based alert of the boat’s location, satellite data, name, flag, stated destinations and other information. He then implemented more sophisticated versions of this model. The result was additional notice to ports, and to authorities outside of the specific ports in question, about boats that should receive specific attention. Waldrop’s experiment worked, and he promptly donated the results to a federal agency.
Owing to the highly contagious nature of Ebola and the close quarters onboard ships, there’s a high risk of an outbreak rapidly affecting an entire crew. Such a ship entering port would pose a grave health threat to persons in the port and surrounding area. U.S. authorities and shippers are aware of the Ebola Maritime Vector threat, but detection and prevention measures appear less rigorous than those in the air travel sector. The rules in effect for maritime traffic were and are that ships must file a notice of arrival 96 hours prior to U.S. landfall, and those notices must include declaring visits made to any Ebola risk ports during the ship’s last five port visits. In addition, ship captains or owners are required to self-report any communicable disease onboard during the 15-day period prior to entering port.
If such a report is filed, the appropriate U.S. authorities are notified and the ship may be boarded, inspected and possibly quarantined. This scenario created compelling incentives for non-compliance with disease reporting, as ship owners face major commercial revenue losses if their ships are quarantined, or if contract conditions are not met owing to delays, or if the ship is diverted and fails to reach the port specified in a bill of lading. Taking this situation into account, Waldrop’s experiment with LUX helped address some issues of public safety in ports.
Two federal agencies looked at the information. It brought the maritime aspect of the problem to their attention and demonstrated a relatively easy way to pay attention to it. They asked about the data sources and began monitoring them. The ships were monitored on the LUX watch dashboard. This procedure provided several days of advance warning about any ship that had visited West Africa and was headed to a U.S. port.
Ebola Maritime Vector Threat
To mitigate the Ebola threat, LUX uses two real-time data streams to perform analysis. The first is data streamed from internationally mandated  automatic identification system (AIS) reporting. AIS, installed on all merchant ships in the world, is especially relevant to monitoring ships traveling to and from West Africa ports. The second source is data generated by the Global Data on Events, Location and Tone (GDELT) project .
Figure 1: Illustration of the LUX user interface. The outer AOI provides alerts on arrivals and departures from the zone of concern. The inner AOI monitors activity in and around port areas.
The LUX Ebola Maritime Vector monitoring methodology works as follows:
- A user defines geographic areas of interest (AOI). In this case, two such areas are displayed on the map as shown in Figure 1.
- The user then writes rules instructing the software to generate alerts on any inbound or outbound ship crossing the boundaries of the AOIs.
- When an alert is first received on an inbound or outbound ship, AIS data is used to determine the ship’s name, its AIS unique identifier number and other data such as course, speed, flag, declared destination port and cargo embedded in the AIS reporting.
- Once the ship is identified, the user writes another rule containing the ship’s name and AIS identifier instructing the software to track and generate alerts on its location and activity.
- LUX tracks the ship to its next port (declared or not) and sends the user alerts based on rules and geographic areas of interest related to that port. A useful tool in this regard is the dynamic area of interest (DAOI). The DAOI is centered on the ship itself and moves with it. A DAOI of any radius may be established. For example, an alert could be generated anytime a ship is within 50 nautical miles of land, providing warning of a pending port visit.
- The user then instructs the software to generate alerts on information collected by GDELT that matches the user’s rule parameters such as reports of Ebola cases, quarantining of ships, illness among ship crew members, etc. This is a potential source for discovering Ebola’s spread through the maritime vector.
- LUX also provides the means to detect abnormal behavior, such as a ship deviating from course, diverting to a port other than its declared port destination, or, particularly, cessation of its AIS reporting stream.
- If a ship’s AIS reporting were to cease, the software’s forecasting analytics would still provide an estimated track and the ship’s progress along it.
- To facilitate monitoring of multiple ships (thousands), rules alerts can be sent to a user-established watch board. The watch board displays, as color-coded cells, aggregates of AIS alert reporting on as many individual ships as desired. It also monitors and displays GDELT alerts. In this way a user may be relieved of constantly monitoring activity for situational awareness and let LUX take up that task 24/7. When the number or type of alerts reaches a user specified threshold, the watch board changes the color of the appropriate watch item cells as a visual notification. An audible tone may be added as an additional notification aid.
What this story highlights is the vast amount of information available at the fingertips of everyone in the analytics community, the ability to act quickly on that information when it is properly applied and brought to the right people’s attention, and the potential benefit of such analysis and action.
Doug Samuelson (email@example.com) is president and chief scientist of InfoLogix, Inc., in Annandale, Va., and a senior operations research analyst with Group W, Inc., in Merrifield and Triangle, Va., supporting the Marine Corps Combat Development Command (MCCDC). He is a longtime member of INFORMS.
Brian Umana (firstname.lastname@example.org), a consultant at Illumina Consulting Group, has workedÂ in software analytics, political analytics and media.
- International Maritime Organization’s International Convention for the Safety of Life at Sea
- GDELT is the largest, most comprehensive and highest resolution open source database of human society ever created. GDELT monitors and analyzes the world’s news media from nearly every corner of every country in print, broadcast and web formats, in over 100 languages, every moment of every day.