The Design of an Integrated IoT and Artificial Intelligence System for Fish Quality Degradation Diagnosis

Abstract

In post-harvest handling, fish quality assessment is typically carried out using traditional sensory observations, which are potentially resulting in inconsistent diagnoses. Furthermore, prior research has not fully integrated the Internet of Things (IoT) with artificial intelligence for fish quality diagnostics, and it frequently concentrates on a single quality metric. This study aims to design and evaluate an integrated system based on IoT and artificial intelligence using a Case-Based Reasoning (CBR) approach for diagnosing fish quality degradation. The developed system utilizes IoT-based sensors to monitor physicochemical parameters, such as temperature, pH, and gas indicators, with real-time data transmission to a cloud platform. The collected data are analyzed using a CBR model as a decision support system. Performance evaluation was conducted using 120 testing data under controlled storage conditions and validated through expert assessment. The results show that the system achieves a diagnostic accuracy of 92.5%, with precision of 91.8%, recall of 93.2%, and an F1-score of 92.5%. In addition, the system has an average data transmission latency of 0.87 seconds, enabling near real-time diagnosis. These findings indicate that the system provides accurate and efficient diagnosis of fish quality degradation and supports post-harvest quality management