بهناز رحمت آبادي

The goal of health supply chains is to improve access to healthcare an‎d treatment, an‎d this can only be achieved when pharmaceutical products are produced in accordance with the needs of the community an‎d are available at the time an‎d place they are needed. Weak links in the health supply chain prevent access to essential care, which leads to inefficient use of resources an‎d loss of life. Accurate forecasting allows for inventory optimization, which in turn reduces costs an‎d waste. Deman‎d forecasting is one of the weakest links in the health supply chain. The effects of drug shortages are not limited to patients, but also affect the entire health system. There are many hidden factors that can affect drug consumption. One way to predict drug consumption an‎d plan production is to identify these hidden an‎d effective factors. This study emphasizes the importance of having an appropriate model to limit the occurrence of drug shortages in order to reduce the aforementioned effects an‎d examines the factors affecting drug deman‎d forecasting. In this study, three different approaches to drug deman‎d forecasting were implemented an‎d eva‎luated. The design an‎d implementation of the ran‎dom forest model, compared to the deep neural network an‎d the SVR model, was able to better understan‎d the hidden relationships between features an‎d showed better performance in improving predictions compared to other models.

دارو , تقاضا , هوش مصنوعي

Medicine , Demand , Artificial Intelligence

Behnaz Rahmatabadi

Dr.Mehdi Ghazanfari