Noblis Team - Multiviz Data Fusion

VAST 2010 Challenge
Grand Challenge: Arms Dealing and Pandemic

Authors and Affiliations:

Catherine Campbell, PhD, Noblis, Catherine.Campbell@noblis.org [PRIMARY contact]
Seth Blanchard, Noblis, Seth.Blanchard@noblis.org
Shin Chin, PhD, Noblis, Shin.Chin@noblis.org
Casey Henderson, Centrifuge Systems, chenderson@centrifugesystems.com
Mitchell Holland, Noblis, Mitchell.Holland@noblis.org
Katharine Jennings, PhD, Noblis, Katharine.Jennings@noblis.org
Peter Kuehl, MD, PhD MedStar, peter.kuehl@medstar.net
Daniel Lucey, MD, Georgetown University, drl23@georgetown.edu
Matthew McCoy, Noblis, Matthew.McCoy@noblis.org
Jill McCracken, PhD, Noblis, Jill.McCracken@noblis.org
Benjamin Pecheux, Noblis, Benjamin.Pecheux@noblis.org
Mik Pietrzak, MD, MedStar, mik.pietrzak@gmail.com
David Roberts, PhD, MPH, Noblis, droberts@noblis.org
Mark Sanders, Future Point Systems, msanders@futurepointystems.com
Virginia Sielen, Noblis, Virginia.sielen@noblis.org
Karen Taylor, Noblis, Karen.Taylor@noblis.org
Daniel Whitford, Noblis, Daniel.Whitford@noblis.org
Harry Cummins, Noblis, graphic artist, hcummins@noblis.org
Richard P. DiMassimo, Noblis, video producer, rdimassimo@noblis.org
Austin Blanton, Noblis Intern, austin.blanton@noblis.org
   
Noblis VAST Webpage: http://www.noblis.org/VAST

Tool(s):

1) Centrifuge, a partner in this challenge (http://www.centrifugesystems.com/), was used to analyze visual relationships from structured data for both intelligence and healthcare data.
2) Starlight from Future Point Systems (http://www.futurepointsystems.com/), another partner in this project, was used as a comprehensive visual analytics platform for analysis of intelligence data. We used Starlight XEE for unstructured data preprocessing.
3) SNUFER (http://www.bioinformation.net/003/001300032008.htm) was used to generate SNP tables.
4) Clustal W (http://www.clustal.org/) was used to align sequences and generate phylogenetic trees.
5) Perl (http://www.perl.org/) was used to develop scripts to organize tables.
6) R (http://www.r-project.org/), a statistical package, was used to calculate the significance of SNPs.
7) Processing (http://processing.org/) was used to develop interactive SNP plots. Processing is a open source design language started by Ben Fry and Casey Reas in 2001.
8) Protovis (http://vis.stanford.edu/protovis/) was used to develop interactive pedigree trees and communications networks. 
9) Copernic Desktop Search (http://www.copernic.com) allowed for indexed search of files, e-mails (and attachments).
10)  IN-SPIRE (http://in-spire.pnl.gov/), was used to integrate information visualization with query to analyze unstructured text documents.
11)  Sentinal Visualizer (http://www.fmsasg.com/) is an automated link analysis software for visualizing, analyzing and managing complex data.
12)  Google Earth (http://earth.google.com/), offered maps and satellite imagery to explore geographical content and render user data.
13)  We also used the open source database PostgreSQL as the underlying database, phpPGAdmin/SQL for database administration and data manipulation, and Microsoft Access/Excel/SQL for ad-hoc queries and data visualization.

Video:
 
Noblis_Fusion_Grand.mp4


ANSWERS:

GC: Investigate any possible linkage between the illegal arms dealing and the pandemic outbreak.  
In particular, you need to address the following:
1. Briefly describe your hypothesized linkage between the arms dealing activity and the pandemic outbreak.  
2. Where did the disease originate (as far as you can tell with the data)?  Provide a time-based tracing of its spread among countries.  If you have established linkages between arms dealers in certain countries and the pandemic, please indicate this.  
3. We had countries with arms dealers identified in MC 1 that did not suffer pandemic outbreaks in MC 2.  Provide a hypothesis as to why some countries that may have been involved with arms dealers did not suffer an outbreak?  

In the spring of 2009 a pandemic outbreak of Drafa Fever occurred spanning at least nine countries across three continents. This outbreak began nearly simultaneously in all affected countries and there was concern that the outbreak was the result of an intentional release of a biothreat agent.

With moderate confidence, the Drafa Fever pandemic of 2009 is assessed to have not been the result of an intentional release of a biothreat agent.
* Intelligence evidence and disease timelines from Health records suggest that the outbreak was likely the consequence of a series of meetings between illegal arms dealing networks that occurred in Dubai in late April.
* Genetic analysis of outbreak strains indicates that the origin of the outbreak was almost certainly Nigeria.
* Intelligence evidence indicates that arms dealers probably traveled from Nigeria, where an ongoing outbreak of Drafa was occurring, to the Dubai meeting.

Linkage Between the Arms Dealing Activity and the Pandemic Outbreak:

With moderate confidence there is evidence to link the Drafa outbreak to a series of meetings between international arms dealers which took place at the Burj Al Arab hotel in Dubai from April 15 to April 23, 2009.

The intelligence dataset provided insight into the structure and organization of a Ukrainian arms dealing network. At the center of the network lie the apparent suppliers, Mikhail Dombrovski and Nicolai Kuryakin, who took over the network from the deceased Leonid Minsky. There are buyer networks that interact indirectly with these suppliers from a variety of countries in Africa, Asia, the Middle East, and South America. Intelligence reports describe a meeting between two or three individuals from Nigeria and a highly connected arms dealer on April 15, 2009. Intelligence evidence suggests that this was likely the first meeting of many that Dombrovski attended during his stay in Dubai. Intelligence reporting also indicates that all of the countries that experiences outbreaks were associated with the Dubai meetings. All of the outbreaks in the countries begin within weeks of the arms dealers returning home and the epidemiological evidence from hospital admission and death records in those countries correlates the timing of the disease outbreak to the arms dealers’ arrivals in Dubai.

From the provided hospital records, intelligence data and genomics data we assess with moderate confidence that Drafa fever is a contagious disease and not part of an intentional bio-event. Based on plots of hospital admission rates by date, there is no evidence of a sudden spike in disease outbreak that might indicate mass, simultaneous exposure of patients. Additionally, the SNP (single nucleotide polymorphism) data from the genomics challenge show some genetic variability among outbreak strains which would be less likely to support the hypothesis of simultaneous release. Additionally there is no intelligence evidence to indicate intention to release a biological agent.

From this evidence we hypothesize that the meeting in Dubai served as the epicenter for this disease outbreak. We estimate with moderate confidence that Dombrovski and an associate (perhaps Kuryakin), are at the center of this arms network, and they met with representatives from each of the outbreak countries between April 15 and April 23. These face to face meetings provided the opportunity for the virus to spread to representatives from all the outbreak locations. Upon returning to their respective countries, the general population was exposed to the infectious virus.

Origin and Timeline of Disease Spread:

There is high confidence that exploitable genomics data collected from patients infected with Drafa Fever indicate that the recent outbreak originated in Nigeria and is contagious. The genetic analysis of the outbreak strain indicates it evolved from the Nigeria B strain of the Drafa virus. Additionally, “local health issues” are referenced in an intercepted communication between Kuryakin and George Ngoki, the Nigerian member of the arms network.

Hospitalization records from nine different countries indicate the presence of an epidemic. Intelligence data also shows that buyer networks representing each of these countries travelled to Dubai between April 15 and April 23 for meetings regarding the purchase of illegal arms. The timing of the disease spread shows a direct correlation between the known arrival and meeting dates in Dubai from the intelligence data, and the liftoff dates for Drafa Fever mortalities in the outbreak countries represented in the hospitalization records. There is intelligence data linking the presence of at least one arms dealer from each of the outbreak countries with the meeting in Dubai.

The timeline of meetings and outbreaks are as follows. Two Kenyans arrived in Dubai on April 17, 2009 and had a scheduled meeting with Kuryakin and Dombrovski on the same day. Kenya was the first to experience a deaths liftoff of all the outbreak countries on April 29. Three Pakistanis were in Dubai from April 18 to April 20 and probably met with a Russian arms dealer during that time period. Pakistan was the next to experience deaths liftoff on May 1. Three Syrians arrived in Dubai on April 18 and probably met with a Russian arms dealer sometime on or after that date. Syria had a deaths liftoff on May 1. Three Lebanese arrived in Dubai on April 18 and probably met with a Russian arms dealer sometime on or after that date. Lebanon had a deaths liftoff on May 2. One Yemeni, who is known to also operate in Saudi Arabia, arrived in Dubai on April 19 and probably met with Dombrovski that same day. Yemen and Saudi Arabia both had a deaths liftoff on May 3. One Iranian arrived in Dubai on April 21 and met with Kuryakin the same day. Iran had a deaths liftoff on May 4. One Venezuelan arrived in Dubai on April 22 and met with Dombrovski the same day. Venezuela had a deaths liftoff on May 5. One Colombian arrived in Dubai on April 23 and met with Dombrovski the same day. Colombia was the last to have a deaths liftoff on May 5.

With moderate confidence it is hypothesized that Ngoki or one of his associates likely became infected in Nigeria prior to their travels to Dubai, and transferred the virus to Dombrovski at the earliest known meetings in Dubai (April 15). Dombrovski subsequently transferred the virus to Nicolai and then they proceeded to infect the Pakistanis, Syrians, Lebanese, and Yemenis. Since one of the Yemenis (Saleh Ahmed) also frequently travels to Saudi Arabia, with low confidence it is assessed that he was responsible for the Saudi outbreak. The Iranians were next to met with Dombrovski, followed by the Venezuelans and Columbians.

Not all dates established in the hospitalization records correlate with the meeting dates. The liftoff dates for hospital admissions do not follow the precise timeline of the Dubai meetings. The hospital admission information contains some number of patients that have similar symptoms but do not have Drafa fever and this baseline noise confounds the timeline especially in countries like Columbia which have fewer cases of Drafa. For this reason the mortality liftoff rates were assessed to be more reliable and were used for this analysis.

Anomalies Associated with Arms Dealers and Outbreak Countries:

According to health data records neither Turkey nor Thailand experienced an outbreak of the Drafa virus, despite having representatives in Dubai at the time of the arms meetings. We have no hospital records from Ukraine, and therefore no evidence of a outbreak. However Ukrainians are at the center of this arms network. Representatives from Turkey arrived in Dubai the day after Ngoki and the other Nigerians, but it’s not explicitly clear that they had any direct dealings with either of the Ukrainian dealers. Intercepted communications with an individual from Syria, indicate that Turkish dealers were staying at a different hotel from the other members of the arms network were possibly present only as financiers.

In contrast, there is direct evidence to support the fact that one representative from Thailand (Boonmee Khemkhaengare) met with Kuryakin. In support of this evidence, the genetic data also revealed that Nicolai infected another individual with Drafa Fever in a hospital in Paris.

The speculation as to why these three countries were spared from the pandemic differs. The hypotheses are derived from ancillary evidence and thus are afforded low confidence of their accuracy. If Turkish individuals had no face to face interaction with anyone infected with the virus they would not have been able to carry the virus back to Turkey, which would explain the lack of outbreak there.

For Thailand, since it is known that Khemkhaengare did meet with Kuryakin, who was infected, it would be expected that Thailand would experience an outbreak. Intelligence data indicates that Khemkhaengare and Kuryakin travel together regularly to visit arms exhibitions in various countries. Since the genetic data revealed that Kuryakin infected another individual with Drafa Fever in a hospital in Paris it is hypothesized that this individual was Khemkhaengare. We also theorize that Khemkhaengare was travelling with Kuryakin when they were both admitted to the hospital in Paris. Genetic evidence also indicates that the person infected in Paris by Kuryakin had a particularly severe case of Drafa fever. If this were in fact Khemkhaengare, then he likely died in Paris and did not have the chance to return to spread the virus in Thailand.

Since we know Kuryakin was hospitalized in Paris, this could explain why he did not transmit the virus back to Ukraine. It is also possible that Dombrovski died as well. There is some evidence based on information about the deaths of the Kenyans and Saleh Ahmed, that the incubation period of the virus was approximately one week. If this is true, then Dombrovski was likely hospitalized in Dubai since he was meeting with other members of the arms network over approximately 8 days (4/15 to 4/23). This would also support the lack of an outbreak in Ukraine.

Recommendations:

At the end of the reporting period and the outbreak there are still residual arms subnetworks (buyers, suppliers, and intermediaries). From the health records the mortality rate of Drafa is estimated at 8-11%. Since intelligence reports and the genomics data set confirm the death of at least 4 members of the arms network (Kuryakin, Ahmed, and the two Kenyans), and indicate Dombrovski may also have died, it could reasonably be concluded that most other Dubai meeting attendees survived. It is recommended that intelligence collection activities and targeting focus on disrupting the remaining arms network which was not severely impacted by the pandemic. The following key individuals and countries are assessed to be important to the continued proliferation of the arms network. Additionally there are public health concerns that also warrant additional attention.

* The outbreak, although large did not disrupt existing arms subnets in Bosnia, Burma, Iran, Kenya, North Korea, Pakistan, Palestine, Saudi Arabia, Sudan, Ukraine, and Venezuela. These subnets should also be targeted for additional intelligence collection.

* Intelligence reports indicate Nigeria, Syria, and Turkey purchased arms during the Dubai meeting for unknown purposes. Additional collections could uncover the intended shipments and the purpose of these deals.

* It is recommended that targeting should also be concentrated on specific individuals, including: Borodinski (who may inherit the entire Ukrainian network as the only known surviving member), Thatcher and Onyango (Kenya), Aden-al-Sallal (Yemen), Ngoki (Nigeria), Khushid (Iran) as well as Chanarong with the Burmese Shan State Army-South rebels.

* Finally, it is also recommended that public health agencies introduce a more robust monitoring of local epidemics, especially in these regions, to be monitored for their potential to develop into new pandemics.