Name: bayeshack-transportation-ems
Owner: Bayes Hack 2016
Description: Starter kit for the Bayes Hack 2016 Transportation prompt on emergency services
Forked from: bayesimpact/bayeshack-transportation-ems
Created: 2016-04-23 02:29:44.0
Updated: 2017-01-25 22:12:47.0
Pushed: 2016-04-23 00:32:49.0
Homepage: http://bayeshack.org/transportation.html
Size: 6151
Language: Jupyter Notebook
GitHub Committers
User | Most Recent Commit | # Commits |
---|
Other Committers
User | Most Recent Commit | # Commits |
---|
How can data optimize emergency services?
In 2014 there were more than 6 million traffic crashes in the United States, resulting in 32,675 deaths and 2.3 million injuries. Based on incident and traffic conditions, emergency responders have a number of options for transportation (including helicopter ambulance transport in lieu of ground ambulance). There aren't standardized methods or technologies designed for emergency medical services personnel to predict and optimize transport times by selecting the right response to every incident.
Can rich EMS response data coupled with incident state?including weather conditions, road and traffic conditions, and hospital locations?be used to develop a real-time tool for paramedics and dispatchers?
Two datasets that may come in handy for this prompt are the traffic-fatalities dataset from FARS (the Fatality Analysis Reporting System) and the EMS response dataset from NEMSIS (the National EMS Information System).
FARS has detailed data regarding crashes, vehicles, drivers, passengers, and pedestrians involved in all traffic fatalities from 1994 to 2014. You can use the web interface to query the data fields you want for a given year, or take a look at data/fatalities-2014-allfields.tsv
, a dump that we prepared of all 2014 crash/driver/occupant/vehicle/pre-crash data fields from FARS, in TSV format.
We have prepared a brief introduction to the FARS dataset, in an iPython (Jupyter) notebook.
NEMSIS has compiled a 4GB dataset of data pertaining to all EMS responses in the United States in 2014, organized into about 20 tables. To download and extract it, run make data/nemsis
(or you can manually download it from here).
It may be easier to start with just the Illinois subset (150 MB) of the NEMSIS 2014 data, which is available separately. To download and extract it, run make data/nemsis-illinois
(or you can manually download it from here).
Note that there are some slight differences in data format between the national and Illinois datasets ? for example, the tables in the Illinois dataset are CSV, while those in the national dataset use a different delimiter.
To learn more about what the dataset contains and where it comes from, see the ReadMeFirst.txt
file in the Illinois dataset.
data/
nemsis/
? this is where the Makefile will download the national NEMSIS 2014 data dump.nemsis-illinois/
? this is where the Makefile will download the Illinois NEMSIS 2014 data dump.fatalities-2014-allfields.tsv
? a dump of all 2014 crash/driver/occupant/vehicle/pre-crash data fields from FARS, in TSV format.fatalities-fieldnames.tsv
? a mapping of field-codes to descriptive field names for the FARS dataset.fatalities-codes.txt
? a raw text dump of the meanings of all relevant codes for fatalities-2014-allfields.tsv
.analysis/
fars-exploration.ipynb
? iPython (Jupyter) basic exploration notebook for the FARS dataset.Makefile
? scripts to download the NEMSIS dataset.