The honey bee colony losses are a major problem of the industry and society. In order to mitigate its effects or even eradicate it, mathematical modelling is to be employed. In our study, we adopt a compartment model accounting for the population size of the brood, hive bees, foragers and the quantity of food stores. Since the process of pupation and emerging does not happen instantly, the respective equation is delayed with the required time. The aim of the paper is to suggest an efficient algorithm to numerically recover the values of the vital parameters that are not directly measurable—the rate of consumption of food by the honey bees. To find their fair values, a quadratic cost functional is minimized via the trust region method. The computational analysis, including perturbation simulations, demonstrates the stability and robustness of the algorithm. Furthermore, the whole approach gives insights how to come up with suitable measures and how to implement them in prevention of colony collapse.