Fatal Injury and Violence Analytics PoC

Welcome to the Fatal Injury and Violence Analytics Proof-of-Concept (PoC) challenge! CDC’s National Center for Injury Prevention and Control (NCIPC) works to understand how injury and violence impact society and what can be done to prevent it. In this proof-of-concept challenge we aim to help them create a prototype solution for detecting and exploring trends in fatal injury and violence data.

Provided in the challenge forum is an artificial dataset detailing death counts by geography, cause of death, and demographic variables. NCIPC is interested in a solution for analyzing the data and:
 

• Detecting upticks and downticks (time series spikes, drops, other anomalies, upward- and downward trends) in the data. It should support different statistical methods, and queries (e.g. identification of trends by deviation from historical baseline; changing the window of historic baseline; selection of state/city for the analysis; comparison between states and cities, etc.).
• Generating statistical summary reports (e.g. the largest identified deviation of death counts from historic baseline; suicide deaths for ages 10-24 by state; etc.).
• Customizing specific analysis/query options mentioned in the previous two points.
• Providing an interactive graphic user interface (GUI) to inspect discovered trends, create summaries, and control all relevant parameters.
• Supporting exports of trends and summaries as static documents (Word or PDF).
   

The goal of this challenge is to identify open source software solutions that meet the requirements, to figure out the high-level architecture of the overall solution for NCIPC, and to create a PoC demo of such a system. The preferred technology requirements are the following:

• The core data analysis is performed remotely (on a cloud server, presumably deployed into AWS in production use); the core analysis is handled by Python or R code.
• The system uses a web interface for end users.
• The solution relies on software licensed under permissive free open-source licenses (e.g. Apache, MIT, etc.)
• These may not be hard requirements. If you have a different approach / technology / software with different licensing  and additional benefits, you should consult with the copilot first, either via the challenge forum, or via email dr.pogodin@gmail.com (in case the question is too specific and may give away your ideas to other members)
   

We expect you to submit a PoC solution that demonstrates how your technologies work together to address the requirements explained above, and what the final look and feel of such a solution will be. Along with the code you will submit a PDF write-up, explaining the approach and requirements of your solution, and commenting on every relevant detail that could not be implemented within the timeline and budget of this challenge, but is feasible to implement with your approach in the future.

Your submissions will be judged subjectively, based on how they fulfil the expectations explained above. Detailed feedback on the review and challenge outcomes will be shared to competitors.

The winning solutions will be presented to NCIPC. Assuming they are approved, we expect to run several follow-up challenges to create a production-ready solution, based on their feedback.

Final Submission Guidelines

Submit a ZIP file containing your PoC code and PDF write-up. Optionally, you may also make a video to demonstrate some aspects of your solution; in such case the video should be uploaded to YouTube, or similar video hosting, as unlisted (accessible by everybody via a direct URL, but not visible in any public search), and the URL to the video should be included with your write-up.