ARTIFICIAL NEURAL NETWORK MODEL FOR CLASSIFICATION OF IOT DEVICE STATES IN MEDICAL INDUSTRY

Abstract

The rapid integration of Internet of Things (IoT) devices in healthcare has created a critical need for intelligent systems capable of accurately classifying device operational states. With over 60% of hospitals adopting IoT for patient monitoring and the global medical IoT market projected to reach USD 254.2 billion by 2026, ensuring device reliability is essential. Studies report that around 15% of medical IoT devices experience undetected malfunctions due to inadequate classification, while traditional manual monitoring methods struggle with real-time, large-scale data and often miss transient faults that can jeopardize patient safety. This study proposes a hybrid classification framework combining an Artificial Neural Network (ANN) with an Extra Trees Classifier (ETC) to categorize device states as Normal or Anomaly. Data from hospital telemetry logs and the Medical IoT Device Dataset (MIDD) undergo preprocessing including null value removal, min-max normalization, and time-series segmentation. A baseline Gaussian Naïve Bayes Classifier demonstrates moderate performance but cannot capture complex nonlinear relationships. In contrast, the ANN with ETC model effectively extracts temporal features and provides robust, efficient classification, achieving superior accuracy and anomaly detection. This approach offers a reliable, real-time solution for monitoring medical IoT devices, enhancing operational safety and healthcare quality.