Artificial Intelligence Based Robust Clinical Data Classification for Intradialytic Hypotension Diagnosis

• Main Authors: Mohammed, Ibraheem Hatem, Obaid, Mohammed Kadhim, Hadi, Maysam Reyad, Alzubaidi, Laith H., Mohammed, Aymen, Hassan, Ahmed R.
• Format: Conference Proceeding
• Language: English
• Subjects: Artificial intelligence Clinical data classification; Biomedical data processing; Classification algorithms; Data models; Deep learning; Internet of Things; Intradialytic hypotension; Medical services; Statistical analysis; Sustainability; Training data; Transforms
• ISSN: 2831-753X
• DOI: 10.1109/IICETA57613.2023.10351218

The demands for communication and computation technologies are growing in daily lives, which drives the utilization of sustainable artificial intelligence (AI) in various applications, particularly healthcare. Intradialytic hypotension (IDH) becomes a common problem at the time of hemodialysis treatment. Forecasting whether patients would face IDH during hemodialysis (HD) was not a simple task. IDH can be linked with many risk factors, means that conventional statistical methods cannot identify the relations that affect it. In this regard, the usage of methods related to machine learning (ML) could allow the detection of complex relations, as they could solve complexities without being explicitly programmed. The deep learning (DL) method will use training data for discovering underlying patterns and helpful in building methods and fitting the optimal method for prediction. In this aspect, this article presents a Robust Deep Learning based Clinical Data Classification for IDH (RDLCDC-IDH) diagnosis. The presented RDLCDC-IDH model aims to detect and classify IDH effectually and timely manner. In the presented RDLCDC-IDH model, data preprocessing is performed to transform it into a compatible format. To properly detect and classify IDH, the RDLCDC-IDH technique make use of deep convolutional neural network (DCNN) method. For effectual hyperparameter adjustment of the DCNN model, the sparrow search optimization (SSO) algorithm is exploited in this work. To demonstrate the enhanced outcomes of the RDLCDC-IDH technique, a wide-ranging simulations can be made. The experimentation outcomes highlighted the enhanced results of the RDLCDC-IDH method on IDH diagnosis.