Ensuring Accurate Measurement in Continuous Glucose Monitors

Retrospective algorithm to improve accuracy and precision of continuous glucose monitoring (CGM) data by exploiting reference blood glucose (BG) measurements and a new constrained regularized deconvolution method. The algorithm takes CGM data, BG references, a sensor accuracy degradation model, and a glucose kinetics model as input. It outputs a retrofitted quasi-continuous glucose concentration signal that is more accurate and precise than the original CGM data, along with a calibrated CGM time series and flagged outliers. The algorithm uses BG references sparsely and frequently to correct the noisy, biased CGM data.

Source

Accurately determining glucose levels from continuous glucose monitors without requiring calibration with fingerstick blood tests involves subtracting the sensor's zero-signal level—obtained from calibration data—from the continuous sensor signal. This adjustment eliminates baseline drift, enabling direct determination of glucose levels from the sensor readings alone.

Source

Monitoring glucose levels from interstitial fluid using continuous glucose monitoring systems to overcome noise and lag issues. The method involves calculating glucose levels using a lag correction algorithm that minimizes errors between interstitial and blood glucose. It also calculates smoothed glucose levels using a smoothing algorithm to reduce noise. The lag-corrected and smoothed signals are then weighted and combined to provide accurate glucose levels and rates of change.

Source

Dynamic glucose monitoring system that optimizes blood glucose testing frequency based on sensor reliability and current glucose level. It determines the time between fingerstick tests based on factors like sensor reliability, glucose trend, and current level versus target range. This allows tailoring testing to patient needs rather than fixed schedules.

Source

Analyte detection and treatment dosing system for monitoring and regulating analyte levels in bodily fluids like blood. The system uses a pump to withdraw samples for analysis, a sensor to measure analyte concentration, and a treatment pump to infuse medication based on the concentration. It aims to provide accurate, personalized dosing of drugs like insulin to maintain glycemic control. The system also has features like calibration, interferent compensation, and user feedback to improve accuracy and safety.

Source

Compensating for errors in continuous glucose monitoring (CGM) by leveraging sensor progression parameters derived from past glucose measurements. The method involves computing ratios of current glucose to past glucose levels, which are used as sensor progression parameters. These parameters are then fed into a gain function along with the current glucose level to calculate a compensated glucose value. This compensated value has reduced error compared to the raw CGM reading. The gain function allows adjusting for factors like sensor drift and interstitial fluid lag. By compensating for errors at each point using historical data, the overall CGM accuracy can be improved.

Source

Personalized non-invasive glucose measurement using machine learning to accurately calculate blood glucose levels without a needle. The device collects invasive and non-invasive glucose readings, calculates errors between them, and builds a customized error function model using machine learning. This model compensates for variations in non-invasive glucose measurements specific to the individual. It then uses the model to provide more accurate personalized non-invasive glucose readings.

Source

A blood glucose measuring device that adjusts the calibration interval based on sensor error to improve accuracy while minimizing pain. The device compares blood glucose levels measured by the sensor and a separate blood draw during a fixed calibration interval. It calculates the time for sensor error to reach a threshold and then sets the calibration interval based on that time. This allows personalized calibration intervals based on sensor error to provide accurate readings while reducing the need for painful blood draws.

Source

Intermittent use of continuous glucose monitoring (CGM) systems to manage diabetes with lower cost and burden compared to continuous CGM. The method involves using a combination of continuous and non-continuous monitoring techniques to manage diabetes. During periods when CGM is not being used, glucose levels are estimated based on correlations learned during CGM periods. This allows managing diabetes between CGM sessions using less expensive and less intrusive methods.

Source

Analyzing the reliability of a glucose sensor in a closed-loop insulin pump system to determine when the sensor needs to be replaced or when manual control is needed. The reliability is assessed based on trends in the sensor data like reduced sensitivity, anomalies, drift, and noise. A metric or indicator representing the reliability is calculated and compared to a threshold. If the metric falls below the threshold, it indicates the sensor is less reliable and prompts actions like requesting calibration, transitioning to manual mode, or sensor replacement.

Source