Model, analysis code, data and output files for FDA opioid systems model, as utilized in:
- Lim, Stringfellow, et al. 2022, "Modeling the evolution of the U.S. opioid crisis for national policy development", published in PNAS at 10.1073/pnas.2115714119
- Reports technical aspects of model development, estimation process, and replication of historical data from 1999-2020, as well as analyses of historical patterns and dynamics
- Stringfellow, Lim, et al. 2022, "Reducing opioid use disorder and overdose deaths in the United States: A dynamic modeling analysis", forthcoming in Science Advances at 10.1126/sciadv.abm8147
- Reports analyses of 11 potential policy strategies for addressing the opioid crisis, with projected effects on overdose mortality and opioid use disorder from 2022-2032
NOTE: This is the live, updated version of the model and associated files. For the archival version corresponding to Lim et al., please see the Zenodo repository at zenodo.org/record/6544836; for the archival version corresponding to Stringfellow et al., please see the Zenodo repository at zenodo.org/record/6599589.
For any questions, please contact Tse Yang Lim, Erin Stringfellow, Mohammad Jalali, or Reza Kazemi.
Vensim model files and Python code (in .ipynb format) used for model estimation, analysis, and graphing of results in the published papers & supplementary materials. This is the 'working' directory of this repository.
Excel data and documentation files, as well as R code for pre-processing of NSDUH data. Explanations of data calculations, assumptions, literature reviews, etc. are in these files.
Note: Vensim data format (VDF) versions of the relevant portions of these data files used in the model are found in Vensim Files and the Model & Analysis Replication Package.
Final reports from FDA-commissioned third-party reviews. For documentation of model structure, estimation process, and supplementary analyses, see SI Appendix of Lim et al. at 10.1073/pnas.2115714119.
Minimal file and code package needed to replicate the analysis 'from scratch', without any pre-generated results or figures. Use this to directly replicate the complete model estimation & analysis process. To do so:
- Make a local copy of this folder
- Update the ControlFile OICC*.txt according to the instructions here
- Run OIC-OO v7.ipynb, loading the updated ControlFile when prompted; make sure any necessary Python packages are installed before attempting to run!
- Then run OSM Results Processing.ipynb
- Then run the desired Graphing .ipynb code in any order
Important: The model estimation code is intended to work with an experimental parallelised Vensim engine. With appropriate modifications to the main function calls (but not the analytical procedure), the same analysis can be run on regular commercially available Vensim DSS, though it will take much longer, particularly with all supplementary analyses (e.g. sensitivity, synthetic data). Please contact Tom Fiddaman for information on the experimental Vensim engine.
Main estimation results and output tables and figures used in the paper, including summary output files from various robustness and sensitivity analyses. Publication figures and results can be downloaded directly here.
The main Vensim model file (.mdl) and other supplementary Vensim files used for model estimation (e.g. optimization control, payoff definition, savelist files, and so on).