Location allocation approach based on influence of points of dispensing

Public health emergencies, such as disease outbreaks, require health authorities to set up points of dispensing (PODs) to efficiently distribute vaccines and drugs to attempt to reduce infection spread within a short timeframe. A considerable amount of research has been conducted to determine the number and location of new PODs while keeping existing ones operational. A query-based approach needs data on the number of new PODs in advance, and an optimisation-based approach requires the specification of distance and percentage gap as constraints or objective to ensure the required population coverage is achieved. The model presented here overcomes such issues. It takes an influence-based approach for POD location-allocation (IPLA) based on a new Influence Affinity Propagation (IAP) clustering algorithm to designate the number and location of new PODs where PODs already exist. IAP takes the influence of existing PODs as a novel weight, that is, the total population size attracted by existing PODs with minimum distance, in addition to geographical information. The proposed model is applied to three synthetic datasets and a real case, namely: the COVID-19 pandemic in Jeddah, Saudi Arabia. IPLA demonstrates better population coverage with a lower number of new PODs than other approaches. It also obtained the closest results to the case study in terms of number, location and predictive performance.