162 Optimizing a Drone Network to Deliver Naloxone

Abstract

Treatment of opioid overdoses depends on timely administration of naloxone. However, administration is delayed in some cases due to response time. Improvement in response time could be overcome with the use of drones. Drones have the potential to deliver naloxone efficiently to the scene of opioid overdose for bystander use. We hypothesize that using a mathematical model to optimize drone placement will reduce time to naloxone administration compared to ambulance response time. Using geocoded retrospective data from suspected opioid overdoses in Durham County, North Carolina from January 2016 to February 2019, we created a drone geospatial network model based on current technological specifications and potential base locations. All 29 fire, paramedic and EMS administrative buildings within Durham County were considered potential drone bases. The drone network modelled response time and county coverage. Then a genetic optimization model was built to maximize county coverage by drones and the number of overdoses covered per drone base. From this model, we quantified the optimum reduction in average response time balanced with the number drone bases required. From 2016 to early 2019, 2,327 distinct incidents of suspected opioid overdose data were found. The average time from 911 call to ambulance arrival at scene was 10 minutes 46 seconds. In the drone network model, as the response time decreased, the number of drone bases increased. In leveraging a balance between reductions in response time and number of drone bases required, we determined that the optimum number of drone bases would be four. Four drone bases would reduce response time by 4 minutes 38 seconds, covering 64.2% of the county (Fig. 1). Our analysis found that in a mathematical model for geospatial optimization, implementing four drone bases could reduce response time of 911 calls or opioid overdoses. Therefore, in theory, drones can improve time to naloxone delivery.