OUD simulation model is built in FRED, an agent-based simulation modeling tool that uses a synthetic population to represent the entire US population. Significant clinical detail can be incorporated into descriptions of diseases, locations, and interventions
Investigators contracted the simulation model under the guidance of a national committee of addiction experts and opioid use disorder clinicians who met several times a year with the investigators to assure the model represented the most current knowledge regarding OUD
Simulation model of OUD represents the disease as a series of clinical states through which individuals may pass through. The model described on these pages determines the individual trajectories of the entire population in the model
OUD model is clinically complex, and it requires a large number of estimates of the probabilities of transitioning from one state to another, for example the risk of a non-user becoming a person with OUD, or the risk of a person with OUD developing an overdose. Given the complexity of the model, there are nearly 100 parameters that must be estimated from the literature and finalized by calibration.
Process of calibrating a complex simulation model has required advances in the methodologies for searching over a large number of possible parameters and finding the set that best produces the outcome streams we have observed in overdoses and overdose deaths.