In June 2015, the United States Geospatial Intelligence Foundation hosted the first GEOINT Hackathon. Students, along with professional developers and data scientists competed. The competitors were challenged to predict where Ebola outbreaks might occur and determine why certain areas of West Africa were not affected by the Ebola outbreak. The goal was to develop a solution that could be modified to a new set of conditions and used by other teams.
Out of 30 participants, including industry professionals, a student team, named “Team Intern” took first place, earning a $15,000 award and free admission to the GEOINT conference by developing a predictive analysis model that revealed a likely pathway for Ebola outbreak. In this webinar, the team will discuss the problem and their new approach to solving it by creating using AWS to create a model that employed multiple data sources to predict outbreaks and epidemics.
3. Investing in cloud evangelists speeds IT innovation
Startups
Corporations
Research orgs
Nonprofits
Government
4.
5. AWS Educate value proposition and goals
Labs and training on
cloud topics and AWS
products
Open course
content by leading
professors and
AWS
Grants for free
usage of AWS
services
Communities that
share best practices
virtually and in
person
• Positively impact many
thousands of students
• Curriculum change
• Student app development
and entrepreneurship
• Growth in AWS
Certifications / Badges
• Accelerate hiring pipeline
8. The goal was to bring together and introduce both non-
GEOINT and GEOINT-savvy coders and data scientists to
interesting problems requiring inventive coding solutions.
9. • Population Density Data
• Hospital Locations
• Streets and Railways
• Access to Water
• Poverty Level
• Humanitarian Aid Areas
**All data is preferable on a 1km x 1km scale to maintain accuracy.
10. Required Sources
Country Outlines/Polygon Shapefiles - http://www.mapmakerdata.co.uk.s3-website-eu-
west-1.amazonaws.com/library/stacks/Africa/index.htm
API Data Source
(ESRI, DigitalGlobe, NGA)
Other Related Data
Vector
Raster
Database/
Dataset
Social
Media
Poverty Data Analysis -
http://povertydata.worldbank.org/poverty/region/SSA
Roads & Railways - http://opendata.arcgis.com/
Timeline of Ebola Spread - http://www.healthmap.org/ebola/#
Population Density - http://www.worldpop.org.uk/
Locations of Hospitals- http://nga.maps.arcgis.com/apps/PublicGallery/
15. Create a
Network Model
Develop a model that
predicts where ebola will
spread and how many
people it will affect
based on how
contagious people
travel.
Inputs Outputs End Product
Take in data regarding
fatality rate, immunity
rate, average travel
distance, transmission
rate, as well as geo-
referenced statistics to
determine virus
movement.
Rasterize Data
into 1km x 1km
Connect to
Neighbors in
Cardinal
Direction
Designate
each Pixel as
a Node
Develop Data-
driven Model of
Infected Indv.
Model Travel,
Infection Rate,
& Disease
Progression
Output
GeoTIFFs of
Infected Areas
& Deceased
16. ● The only travel that we are modeling is the travel of contagious people
● Each contagious person can travel approximately 2,000 km per 10-day timestep
dependent on transit time and travel options, specifically roads and railways
● Disease control conducted at water ports and airports is sufficient to prevent this
method of the spread of infection
● Immunity rate, fatality rate, and average distance are user inputs. Past data
indicates a wide range of values and studies, to date, are inconclusive.
http://allafrica.com/stories/201409082247.html
17. Population Density of
Continental Africa
Highlight the area of interestClip to West Africa to reduce data quantity
Population Density of
West Africa
Convert Raster to a 1km x 1km grid to represent
each pixel
Each pixel represents a node in a node & edge
system
N
E
S
W
Travel paths are limited to the Von Neumann
neighborhood
18. D/R
Exposure Incubation Period Death/Recovery
Asymptomatic Symptomatic
Contagious Phase
First exposure to symptom onset is 2 to 21 days.
http://www.biomedcentral.com/1741-7015/12/196
http://www.who.int/mediacentre/factsheets/fs103/en/
The illness lasts 6 to 10 days.
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3766904/
http://www.huffingtonpost.com/2014/08/02/ebola-symptoms-infection-virus_n_5639456.html
I₁ I₂ I₃ I₄ C₁ C₂
5 days
Key:
19. Contagious
Leave Stay
Max travel is 2,050 km/10 days. Assume 34 km/hr (55miles/hr) traveling 6 hrs/day.
During each 10-day timestep, members
of the contagious population have the
option of traveling up to 2,000km or
staying in their current location.
Contagious travelers are more likely to
go to areas with roads, higher population
densities, closer proximities to hospitals,
and closer proximities to cities.
North East South West
41. ● Threading input data processing, contagious travel, and output
visualization
● Using HDF5 for data storage
● Vectorizing code with Numpy to improve speed
● Generalizing model for use with other data sources
● Open source our modeling software:
https://github.com/pawarren/PyDemic
42. Thank you to all of the judges for being here today as well as Hackathon sponsors and
data providers.
Notes de l'éditeur
Good Afternoon –
We have a skills gap for cloud and AWS
Evangelists
- startups, corps, govt, research, and nonprofits
Schools as entrepreneurial engines
All of a sudden, w the possibility a reality that you can try new ideas:
Move teams from learned helplessness where no point using shower cycles
To a world where employees are motivated to think of new ideas for customers
And instead of only getting these ideas from select senior folks, come from all over org
People often ask us what does cloud mean for our IT people
Reality is they don’t go away…work on value-added activities on top of infrastructure instead of undifferentiated racking and stacking
Imo, better question is how do we empower more of our employees to invent/improve cust exp
Truth is, people who work at enterprises want to invent as much as start-ups, just been hamstrung
Cloud unleashes this innovation…lets you be more agile, get more ideas all over org, and RECRUIT more talented folks in process
Better for customers, companies, and business—WIN ACROSS BOARD
***review Slide