A Complex Event Processing Framework for Patients with Diabetes

Managing diabetes requires continuous monitoring of blood glucose levels to prevent complications. In this context, this paper focuses on analyzing time-series glucose data to model different glycemic conditions, ranging from normal to abnormal or extremely abnormal. On the one hand, the developed system exploits Python tools to perform data cleaning and segmentation into intervals, and on the other hand it leverages the ISEQL language for complex event detection. More specifically, our proposed architecture includes a Data Crawler for raw data extraction, an Interval Action Detector for event annotation, and a High-Level Event Detector for complex event identification. Results highlight effective glucose level analysis for diabetes management, emphasizing pattern recognition and anomaly detection across various time intervals, thus both in the short and in the long term. Consequently, this approach significantly enhances understanding of glucose behaviors critical for personalized treatment strategies.