Prolonged hospital stays beyond the medical discharge date present significant challenges, including suboptimal patient care, bed shortages, and increased hospital costs. Addressing these issues is not a trivial task, as it depends on two simultaneous predictions: one to predict medical discharge dates and another to identify aftercare needs. Such predictions should rely on the latest patient information, enabling transfer nurses to make timely and efficient aftercare arrangements, thus potentially reducing unnecessary prolonged hospital stays. However, the interdependence of these predictions complicates their evaluation. To tackle this, we propose a novel methodology, named ADEM score, for assessing the performance of a model that makes such predictions. In this paper, we evaluate the performance of predictions done by healthcare professionals based on the ADEM score. This new approach emphasises critical periods (based on the two predictions) leading up to discharge, offering a more precise and realistic evaluation of the effectiveness of the models. By focusing on these critical periods, our proposed metric better represents a real-world scenario than previous evaluation methods, ultimately aiming to enhance patient care and operational efficiency.