The ambulatory glucose profile (AGP) lacks sufficient statistical metrics and insightful graphs; indeed, it is missing important information on the temporal patterns of glucose variations. The AGP graph is difficult to interpret due to the overlapping metrics and fluctuations in glucose levels over 14 days. The objective of this proposed work is to overcome these challenges, specifically the lack of insightful information and difficulty in interpreting AGP graphs, to create a platform for decision assistance. The present work proposes 20 findings built from decision rules that were developed from a combination of AGP metrics and additional statistical metrics, which have the potential to identify patterns and insightful information on hyperglycemia and hypoglycemia. The "CGM Trace" webpage was developed, in which insightful metrics and graphical representations can be used to make inferences regarding the glucose data of any user. However, doctors (endocrinologists) can access the "Findings" tab for a summarized presentation of their patients' glycemic control. The findings were implemented for 67 patients' data, in which the data of 15 patients were collected from a clinical study and the data of 52 patients were gathered from a public dataset. The findings were validated by means of MANOVA (multivariate analysis of variance), wherein a p value of < 0.05 was obtained, depicting a strong significant correlation between the findings and the metrics. The proposed work from "CGM Trace" offers a deeper understanding of the CGM data, enhancing AGP reports for doctors to make treatment adjustments based on insightful information and hidden patterns for better diabetic management.
Keywords: CGM device; clinical metrics; decision assistive system; standardized metrics.