Probabilistic attack reconstruction and resource estimation in “reload” scenarios

Faced with a bio-terrorist attack with an aerosolized pathogen preparation, an emergency manager will need to make medical resource allocations early in the aftermath, with large uncertainties in the estimates of the resource demand, and in ignorance of what subsequent attacks might occur at other sites. In this work we use earlier results on the reconstruction of bio-attacks from partial observations of the patient stream to calculate a probability density function of the resource demand. Thereafter, we employ an optimization-under-uncertainty approach to perform resource allocations for both a single-site attack and a multiple-site attack, i.e., a reload scenario. We test the approach in situations where resources are scarce. Our approach can develop allocations profiles that have the potential to reduce the probability of an extremely adverse outcome in exchange for a more certain, but less adverse outcome. We also explore the effect of placing limits on daily allocations.