Dynamic L-RNN recovery of missing data in IoMT applications

One of the most important factors of success of the Internet of Medical Things (IoMT) applications is reliable data delivery. The high quality of data delivery is a vital issue for IoMT applications to provide a high-quality of services to the end users. However, IoMT applications may suffer from low quality of data delivery due to several reasons, such as sensing errors, bad connections or outside attacks. As a result, the collected data is incomplete. IoMT applications require a complete data to provide a high-quality of services to the end users; otherwise, the performance will decrease and not meet the main requirements of IoMT applications. In reality, missing data should be intelligently recovered to save time and cost. In this paper, we propose a Dynamic Layered-Recurrent Neural Network (Dynamic L-RNN) approach to recover missing data from IoMT applications. The main idea is to perform a dynamic L-RNN to predict any missing value in a simple fast manner to save time and cost. The collected data is divided into two categories, complete and incomplete data. A dynamic L-RNN is trained based on complete data, which is used to predict the missing data from incomplete data. This proposed method is able to recover the missing data for IoMT applications with high AUC value when applied to two different datasets. The obtained results show great enhancement in the AUC values after recovering the missing data.