Quantification of the Variability of Continuous Glucose Monitoring Data

Glucose Data Classification for Diabetic Patient Monitoring

Living longer and healthier is the wish of all patients. Therefore, to design effective solutions for this objective, the concept of Big Data in the health field can be integrated. Our work proposes a patient monitoring system based on Internet of Things (IoT) and a diagnostic prediction tool for diabetic patients. This system provides real-time blood glucose readings and information on blood glucose levels. It monitors blood glucose levels at regular intervals. The proposed system aims to prevent high blood sugar and significant glucose fluctuations. The system provides a precise result. The collected and stored data will be classified by using several classification algorithms to predict glucose levels in diabetic patients. The main advantage of this system is that the blood glucose level is reported instantly; it can be lowered or increased.

Knowledge-driven dictionaries for sparse representation of continuous glucose monitoring signals

Continuous glucose monitoring (CGM) of patients with diabetes allows the effective management of the disease and reduces the risk of hypoglycemic or hyperglycemic episodes. Towards this goal, the development of reliable CGM models is essential for representing the corresponding signals and interpreting them with respect to factors and outcomes of interest. We propose a sparse decomposition model to approximate CGM time-series as a linear combination of a small set of exemplar atoms, appropriately designed through parametric functions to capture the main fluctuations of the CGM signal. Sparse decomposition is performed through the orthogonal matching pursuit (OMP). Results indicate that the proposed model provides 0.1 relative reconstruction error with 0.8 compression rate on a publicly available dataset containing 25 patients diagnosed with Type 1 diabetes. The atoms selected from the OMP procedure can be further interpreted in relation to the clinically meaningful components of the CGM signal (e.g. glucose spikes, hypoglycemic episodes, etc.

Development of the Likelihood of Low Glucose (LLG) algorithm for evaluating risk of hypoglycemia: a new approach for using continuous glucose data to guide therapeutic decision making

The objective was to develop an analysis methodology for generating diabetes therapy decision guidance using continuous glucose (CG) data. The novel Likelihood of Low Glucose (LLG) methodology, which exploits the relationship between glucose median, glucose variability, and hypoglycemia risk, is mathematically based and can be implemented in computer software. Using JDRF Continuous Glucose Monitoring Clinical Trial data, CG values for all participants were divided into 4-week periods starting at the first available sensor reading. The safety and sensitivity performance regarding hypoglycemia guidance "stoplights" were compared between the LLG method and one based on 10th percentile (P10) values. Examining 13 932 hypoglycemia guidance outputs, the safety performance of the LLG method ranged from 0.5% to 5.4% incorrect "green" indicators, compared with 0.9% to 6.0% for P10 value of 110 mg/dL. Guidance with lower P10 values yielded higher rates of incorrect indicators, such as 11.7% to 38% at 80 mg/dL. When evaluated only for periods of higher glucose (median above 155 mg/dL), the safety performance of the LLG method was superior to the P10 method. Sensitivity performance of correct "red" indicators of the LLG method had an in sample rate of 88.3% and an out of sample rate of 59.6%, comparable with the P10 method up to about 80 mg/dL. To aid in therapeutic decision making, we developed an algorithm-supported report that graphically highlights low glucose risk and increased variability. When tested with clinical data, the proposed method demonstrated equivalent or superior safety and sensitivity performance.