Skip to content

Given previous times and locations of opioid overdose deaths, can we predict where future interventions would be effective?

License

Notifications You must be signed in to change notification settings

tufts-ml/opioid-overdose-models

Repository files navigation

Spatiotemporal Forecasting of Opioid-related Fatal Overdoses: Towards Best Practices for Modeling and Evaluation

This is code for the published paper:

Spatiotemporal Forecasting of Opioid-related Fatal Overdoses: Towards Best Practices for Modeling and Evaluation
Kyle Heuton, Jyontika Kapoor, Shikhar Shrestha, Thomas J Stopka, Michael C. Hughes
American Journal of Epidemiology, 2024
https://doi.org/10.1093/aje/kwae343

Please get in touch with Kyle Heuton if you have questions.

Data

Cook County

Cook County Data is located in the cook-county directory. Running extract_dataset.py will recreate the necessary files

Massachusetts

Massachusetts data is unable to be shared, but our cleaning scripts are included in the massachusetts directory

Models

Simple Python Models

Most models are located in the experiment-runner directory, and running recreate_table.py will re-create the results from Table 1

Other models

CASTNet, the Bayesian spatiotemporal model, and the negative binomial regression are more complex to run, and reside in their own directories

About

Given previous times and locations of opioid overdose deaths, can we predict where future interventions would be effective?

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 3

  •  
  •  
  •