A Novel Analysis and Classification of Pressure Ulcer Prediction for Coma Patients in ICU Using Deep Learning Techniques

Pressure Ulcer or bed sore is a common disease caused due to the constant pressure applied over the skin. It primarily affects coma patients and patients in the Intensive Care Unit (ICU). This can be identified as lesions or deep cuts on the affected areas. This occurs because the blood flow in those skin areas gets stopped by the constant pressure applied there, which causes the death of cells and damages the skin. Evolving technologies emerging with influential ideas solve various problems that arise on a daily basis. The proposed system uses two evolving technologies such as deep learning and augmented reality. With the use of deep learning algorithms, the lesions are examined and the existence of the pressure ulcer can be analysed. The Deep Learning model developed here uses a dataset that compares the images of the lesions from the affected patient. The results are displayed with the use of Augmented Reality (AR), which can benefit individuals affected by the bed sore. The proposed model is developed as a hybrid model in which the deep learning algorithms are Dense Convolutional Network ( DenseNet), Recurrent Neural Network (RNN), and Long Short - Term Memory Networks (LSTM). The accuracies obtained from these algorithms are 89%, 94%, and 97% respectively. By comparing the results obtained by the model the algorithm Long Short-Term Memory Networks (LSTM) is considered efficient and accurate among the other algorithms employed. The output of the model is fed into the AR module where the results are displayed in AR format.