Shane Henderson, Cornell University
Typical ambulance response times are inadequate for improving out-of-hospital cardiac arrest (OHCA) survival rates. Recently, volunteer systems such as GoodSam and Pulsepoint have arisen in which volunteers are alerted by their smartphones to a nearby OHCA and can choose to respond or not. With a sufficient density of volunteers, response times to OHCA can be dramatically reduced, thereby materially increasing survival rates. How many volunteers are needed? And what policy should be followed to alert volunteers to ensure a timely volunteer response while not overly burdening volunteers with too many alerts? Ideally we would use volunteer location data to estimate volunteer density and thence infer survival rates, but privacy considerations preclude this approach. We use a combination of Poisson point processes, convex optimization and other modelling techniques to provide answers to these questions.
Joint work with Caroline Jagtenberg, Hemeng (Maggie) Li and Pieter van den Berg.