Inspiration
Every ten minutes, an American dies from drug addiction. Two-thirds of these fatal overdoses are from opioids such as heroin and prescription pain killers. What makes the opioid epidemic uniquely difficult to address is that there is no one profile of an addict, so it's challenging for community organizations to target their limited resources toward those who are most in need.
What it does
Our web app maps data points related to drug emergencies and provides basic demographic information about the drug abuser (age, gender, income, etc). It empowers mobile health clinics and nonprofits to direct their services toward addiction "hotspots" and tailor their messaging to specific demographic groups. For instance, a mobile health clinic may distribute a different opioid resource kit to middle-aged women than to college-age male students.
How I built it
We used HTML and JavaScript to process and display the data, and we used Python to scrape CSV files. Using CSV data from the CDC about the demographics of drug abusers, we still had to mock the geographic data using a random number generator because the CDC did not provide accurate data on this. They provided information such as gender and age, but all geographic data points were the same (potentially due to privacy concerns).
Challenges I ran into
The biggest challenge was parsing the data files.
One of the biggest challenges was finding comprehensive data sets that integrate both geographic and demographic data. Many local police departments and first responders in states where the opioid crisis is particularly severe do not collect geographical data in a centralized database, nor do they release the data to the public, due to valid privacy concerns. There is no centralized database to which all researchers have access. Therefore, data must be culled from a variety of sources in order to provide a complete picture of the crisis.
Moreover, the mass volume of data is often overwhelming and causes the computer to lag. The difficulty of gathering data was so immense that approximately 15+ hours were required to gather, transform, and parse the data. The majority of time was spent transforming data from the CDC's data sheet to a 2-dimensional array of features. The CDC's data sheet was extremely irregular; much of the data did not follow a particular set of guidelines, which made parsing a tedious task. Due to irregularities and inconsistencies in the data's format, our group transitioned from trying to convert an XML file, then to using JSON to extract the data, and finally to scraping data from a CSV file with python and reading it into a 2 dimensional array. Furthermore, the data set included 90,000+ individual entries, increasing the time to perform conversions drastically.
Accomplishments that I'm proud of
I'm proud of my team's ability to adapt to our situation over the course of the hackathon. Especially in dealing with the difficult task of parsing over 90,000 entries, we were able to realize that there were other options and we weren't stuck using an ineffective method.
What I learned
I believe we have learned much during our time here at HSHacks. First, we learned how much effort is required to interpret and analyze datasets, especially those which contain enormous amounts of individual entries of data. We also experienced frustration in dealing with this type of problem, for more than 15 hours. This more or less taught us that patience is key in dealing with datasets. In the future, I expect important data to be of similar magnitude, and this has showed us how we should prepare to handle them. One of our team members, Meilinda, began this competition with no experience in front end development and the use of python. But, over the course of the hackathon, her first, she learned how to handle both effectively.
What's next for Aureal
Gathering data from local sources (e.g. counties and municipalities) who often don't partake in such practices will be especially important because the opioid crisis has hit small rural and urban communities the hardest. Moreover, hooking up the web app to emergency services will enable real-time and automatic data entry of 911 phone calls.
It is also important to add additional filters. The CDC splits opioids into four categories, and it would be beneficial to health and social workers to see the maps for heroin, synthetic opioids, and prescription opioids, since the reasons for addiction may differ depending on the drug. This will enable health workers to sharpen their focus on specific drugs.
Log in or sign up for Devpost to join the conversation.