A Neural-Network-Based Model for the Removal of Biomedical Equipment From a Hospital Inventory

This article puts forward an accurate and robust model based on an artificial neural network that guarantees a warning when a piece of medical equipment requires replacement. A perceptron neural network composed of 1 input layer with 2 neurons is described. The artificial neural network can classify data in groups. In this research, 3 groups were classified. These groups depend on numerical values of service cost/acquisition cost and usage time/useful lifetime ratios. A supervised learning rule to train the artificial neural network was selected. The training process was carried out by collecting typical data from 200 high-performance and 100 low-performance devices from 4 hospitals under study. The network was tested by collecting data (998 high-performance and 765 low-performance devices) in 4 hospitals. In 100% of the cases, the artificial neural network classified the equipment in the expected groups. It can be concluded that the network had a great level of data discrimination and an excellent performance level.