Glucose Sensor Calibration for Accurate Monitoring

Compensating for varying accuracy of medical sensors over their lifetime to improve performance of devices like insulin pumps. The technique involves adjusting sensor readings based on estimated accuracy levels at different points in the sensor's lifespan. This accounts for the fact that sensors can have lower accuracy early on and higher accuracy later. By taking into account the sensor age, the adjustment aims to provide more accurate readings to devices like insulin pumps to improve dosing decisions.

Source

Method for accurately determining glucose levels in a person's body using a continuous glucose monitoring system, with reduced computational resources and simpler implementation compared to existing methods. The method involves estimating glucose levels in the blood by first estimating glucose levels in the tissue surrounding the blood using a sensor, then using a model to convert the tissue glucose estimates to blood glucose levels. Kalman filters are used to account for measurement and process noise.

Source

Correcting time lag in measurements of analyte concentration, like glucose, in interstitial fluid, to improve accuracy. The method involves adjusting the lag correction based on the patient's analyte variability and range. This balances maximizing lag correction versus minimizing output noise. Higher variability patients benefit more from lag correction as it compensates for lag between interstitial and blood levels. But lower variability patients risk reduced accuracy due to noise amplification. So the lag correction is scaled based on variability.

Source

Method for constructing a calibration model for dynamic blood glucose monitoring that improves the accuracy of glucose level measurements using dynamic glucose sensors. The method involves fitting a nonlinear model between sensor current and glucose concentration using target data subsets representing different glucose levels. This initial calibration model is then updated with a time-varying parameter function to create a final calibration model that can adapt to changing glucose levels over time.

Source

An adaptive calibration system for continuous glucose monitoring (CGM) sensors that improves accuracy by optimizing the sensor's response to different blood glucose levels and eliminating attenuation effects. The calibration involves segmented processing of the sensor data, weighted least squares fitting, and multiple rounds of fine-tuning. It accounts for factors like temperature, concentration changes, and sensor batch variation. The segmented fitting allows localized optimization instead of global solutions.

Source

Optimizing and adjusting initialization sequences for glucose sensors based on parameters relevant to manufacturing the sensor and environmental conditions. The goal is to shorten the time from sensor insertion to accurate glucose readings. The initialization sequence involves applying specific voltages and durations tailored to the sensor and patient conditions. This allows faster hydration, electrical equilibrium, and stabilization. The sequence is calculated based on factors like platinum surface area, glucose oxidase activity, and current slope.

Source

A calibration system and method for continuous glucose monitoring that improves the accuracy of glucose readings from continuous glucose monitors (CGM) by correcting for the delay and hysteresis between interstitial fluid glucose and blood glucose. The system involves using a separate blood glucose meter along with the CGM to collect paired data, then calculating prediction models for each patient's unique glucose dynamics. This targeted modeling improves accuracy compared to generic models. The system divides patients into groups based on factors like age, BMI, etc, and calculates prediction models for each group. It also combines three calculation methods and adjusts weights based on real-time pairing to further improve accuracy.

Source

Continuous blood glucose monitoring systems can provide detailed time series data to help manage blood sugar levels for people with diabetes. However, the glucose sensors used in these systems can drift over time, causing inaccuracies. To address this, the disclosed method for calibrating continuous glucose monitors involves using weighted linear regression to calculate adjustment factors for the sensor's output. The method involves collecting multiple sets of blood samples and corresponding glucose readings from the monitor over a period of time. The readings are then used to calculate a weighted linear regression line to find the adjustment factors that minimize the error between the true blood glucose values and the monitor's readings. These factors are then applied to subsequent readings from the monitor to compensate for any drift and improve accuracy.

Source

Reducing blanking of sensor glucose signals in continuous glucose monitoring (CGM) systems by using multiple machine learning models trained for specific conditions. The CGM device inputs sensor data into multiple models, each trained for different data characteristics or abnormal conditions. The models generate predicted glucose values. The device combines the predictions to generate the displayed glucose value. This improves accuracy and reduces blanking compared to using a single model.

Source

Personalized calibration of glucose sensors to improve accuracy by estimating subject-specific in-vivo calibration parameters for sensors with different diffusion characteristics. The method involves obtaining glucose measurements from two sensors in a subject, estimating the in-vivo calibration parameters for each sensor based on the measurements, and using the estimated parameters to calculate blood glucose levels. The calibration parameters compensate for differences in sensor response times due to implant depths.

Source

Personalized calibration method for glucose sensing devices that improves accuracy by accounting for subject-specific differences in glucose diffusion between sensors. It involves using two glucose sensors at different depths in a subject to estimate personalized time constants for glucose diffusion from blood to each sensor site. This allows more accurate glucose level estimation using the sensors by accounting for the subject's unique interstitial glucose dynamics. The method involves obtaining glucose measurements from both sensors over a time interval, estimating the personalized time constants based on the measurements, and using the estimated time constants to calculate glucose levels from the sensors.

Source

Calibrating continuous glucose monitors more accurately by using monitor electrodes with the same chemistry stacks as the working electrodes to compensate for environmental effects on the working electrodes. The monitor electrodes are calibrated before installation and the changes in their operating parameters are used to determine environmental effects on the working electrode chemistry stacks. These effects are then applied to the working electrode calibration to correct glucose readings.

Source

Automatic blood sugar correction for continuous glucose monitoring (CGM) devices to improve accuracy over time as sensor performance degrades. The method involves regularly collecting the transient and steady-state output currents of the sensor. By analyzing the transient current characteristics, the sensitivity coefficient of the sensor is determined. Then, the steady-state output current is corrected using the sensitivity coefficient and used to calculate blood glucose levels. This allows automatic and efficient blood sugar correction as sensor performance degrades over time.

Source

Calibrating glucose sensors using a barcode to improve patient experience and convenience. The method involves binding each sensor with a unique barcode, testing the sensors to generate calibration data, and storing the association between barcode and calibration info. When using the sensor, the barcode is scanned to retrieve the calibration data, which is then applied to compensate for any sensor drift or error. This allows calibration to be done before use, eliminating the need for fingerstick calibration. The barcode provides a way to track and associate sensor performance data for calibration purposes.

Source

Calibrating glucose sensors using electrical parameters to improve accuracy and reduce variability. The technique involves measuring electrical parameters like voltage, current, or impedance from the sensor in vitro. These parameters are used to cluster the sensors based on similar electrical characteristics. Each cluster represents a configuration for calibrating the sensor. When the sensor is implanted, it is configured using the identified cluster's calibration parameters. This allows customizing sensor calibration based on its electrical properties, improving accuracy and consistency compared to generic calibration methods.

Source

Method for calibrating non-invasive biometric data like glucose levels measured by devices that don't require blood samples. The calibration involves comparing the non-invasive readings with continuous readings from a device that uses a sensor inserted under the skin. By learning the relationship between invasive and non-invasive measurements over time, the non-invasive readings can be personalized and corrected for individual users. This allows more accurate determination of biometric trends without needing invasive sensors.

Source

Calibrating analyte data from medical sensors using artificial intelligence to provide more accurate readings and extend calibration intervals. The method involves using an AI model to predict correction values over time based on historical data. This allows calibration to be done less frequently since the AI can compensate for drift. By receiving initial calibration data, storing it, then using it along with current sensor data to predict and apply corrections, the AI calibration can handle time-dependent errors better than fixed mapping tables.

Source

Robust glucose monitoring system using orthogonally redundant sensors for improved accuracy and reliability compared to single sensor systems. The system has two glucose sensors with distinct technologies, like electrochemical and optical, implanted in the body. The sensors measure glucose levels separately and the optical sensor's output is calibrated based on the electrochemical sensor's reading. This allows accounting for environmental effects and sensor anomalies. The redundant sensors provide true redundancy with unique failure modes that don't intersect.

Source

Method for stabilizing continuous glucose monitoring systems to improve accuracy of measured blood sugar levels. The method involves dynamically adjusting the stabilization time required before displaying the sensor-transmitter data at the communication terminal. Instead of using a fixed stabilization time, it allows variable stabilization times based on the sensor data received. The stabilization steps include an initial stabilization phase set by the terminal, followed by a second phase where stabilization can occur for any time within the initial phase. If the second phase stabilization is not complete, a third stabilization phase is initiated. This allows the sensor data to stabilize without forcing users to wait an arbitrary fixed time. If stabilization is not achieved within the third phase, the terminal terminates the connection with the sensor.

Source

Improving accuracy and reliability of glucose monitoring systems by providing data processing and control features. The method involves sampling a batch of in vivo glucose sensors, determining the sensitivity of each sensor, calculating a mean sensitivity, and using that mean value to validate sensor data. Sensor codes are assigned based on sensitivity ranges. During monitoring, the sensor code is entered and the system checks data against the corresponding mean sensitivity. If outside a tolerance, it disables glucose trend calculations and alerts the user. This accounts for sensor drift and ensures accurate trend estimation.

Source

Improving continuous glucose monitoring (CGM) systems by using multiple machine learning models trained for specific conditions to generate more reliable sensor glucose values. The models are trained separately for factors like sensor data availability, accuracy, and probabilistic reliance. During CGM, the models are applied and outputs averaged to generate the sensor glucose value. For outlier conditions, signatures in sensor data are identified and models adjusted to prioritize those associated with the signature. This allows customization of models for situations like limited data or high variability to improve accuracy.

Source

Simplifying blood glucose measurement confirmation in insulin pumps to improve user experience and prevent calibration expiration. The technique involves automatically initiating glucose sensor calibration and insulin bolus calculation when the user confirms a fingerstick glucose measurement, without requiring separate confirmation steps. This eliminates the possibility of outdated sensor calibration and redundant confirmations.

Source

Implantable medical device calibration method that allows automatic sensor calibration without external equipment. The method involves adding an identifiable electronic component to the sensor assembly. The component's value is detected by the sensor circuit. A parameter library with calibration parameters for different component values is stored. The sensor circuit reads the calibration parameter corresponding to the detected component value and uses it to correct the sensor's original parameters. This calibrates the sensor without external devices.

Source

Calibrating blood glucose measurements from a continuous glucose monitoring system to improve accuracy and reliability. The calibration method selects between two modes based on the difference between blood glucose values measured by the continuous monitor and a reference device. In the first mode, the reference value must fall within a calculated range based on the continuous measurement to calibrate. In the second mode, multiple reference values are used to calibrate if the first mode is not applicable. This prevents inaccurate calibration when single reference values are outliers.

Source

Automatic system for blood sugar control in patients using an improved model calibration technique to improve accuracy and reliability. The system predicts future blood glucose levels based on a physiological model and adjusts insulin delivery accordingly. To calibrate the model, it minimizes the error between estimated and measured blood glucose during a past observation period. This involves estimating the model's initial state and parameters simultaneously. If the model quality is unsatisfactory, insulin delivery is adjusted without using the model predictions.

Source

Closed-loop insulin infusion systems using orthogonally redundant glucose sensors for improved accuracy and reliability. The system has two glucose sensors, one optical and one electrochemical, to provide orthogonal redundancy. An algorithm combines the sensor data to improve accuracy and reliability. If one sensor fails, the other can provide glucose values. The sensors have features like distributed electrodes and membrane barriers to reduce drift and fouling. The system uses on-demand calibration rather than frequent fingersticks.

Source

Optional calibration of calibration-free glucose sensors using internal sensor data and electrochemical impedance spectroscopy (EIS) to reduce or eliminate the need for external calibration references. The method involves measuring the sensor's current signals and performing EIS procedures to calculate glucose values. By fusing the results from multiple predictive models and applying filtering, a calibrated glucose value is obtained without a fingerstick. The sensor's data provides internal calibration and stability feedback.

Source

More accurate and efficient method to determine blood glucose levels using a rolling time domain estimation technique. The method involves continuously monitoring interstitial glucose levels and using a rolling time domain estimation algorithm to calculate the current blood glucose level. This provides improved accuracy compared to traditional continuous glucose monitoring systems that rely on calibration using separate blood glucose measurements. The rolling time domain estimation allows estimation of blood glucose levels based on historical interstitial glucose data, taking into account diffusion dynamics and noise models.

Source

A reference offset calibration algorithm for blood glucose monitoring that reduces the frequency of fingerstick calibrations by leveraging cloud computing to improve accuracy over time. The algorithm extracts stable segments of blood glucose data from historical readings to predict and calibrate current values. It calculates weights based on time and variability to prioritize older, more stable segments. This allows extending sensor life by adjusting current readings instead of frequent fingerstick calibrations.

Source

Calibrating a noninvasive sensor device to accurately estimate physiological parameters like glucose levels without using a continuous glucose monitor. The calibration involves using data from the invasive monitor and noninvasive sensors simultaneously during a calibration period. By correlating the data, a transfer function is generated to estimate glucose levels from the noninvasive sensor output alone. This allows accurate glucose monitoring without needing a continuous sensor.

Source

A calibration method for implantable medical sensors that simplifies and automates calibration without requiring external devices. The method involves using a recognizable electronic component integrated into the sensor to determine calibration parameters. The sensor detects the component value, reads the corresponding calibration parameter from a stored library, and adjusts the sensor parameters accordingly. This self-calibration eliminates the need for external devices like meters or scanners, as the sensor can calibrate itself based on the internal component value.

Source

Calibration-free glucose monitoring using a glucose sensor with internal electronics for measuring glucose levels without external calibration. The method involves calculating sensor glucose (SG) values using electrode current (Isig) signals, electrochemical impedance spectroscopy (EIS) data, and calibration factors. It fuses multiple SG predictions to generate a single, calibrated SG value. Error detection and correction is done. This internal calibration compensates for variations in sensor parameters and provides accurate glucose readings without external calibration devices.

Source

A personalized non-invasive glucose measurement system using machine learning to improve accuracy of non-invasive glucose meters for diabetes management. The system initially measures glucose invasively and non-invasively, calculates the error between the two, and builds a personalized error function model using machine learning to compensate for the error in future non-invasive measurements. The model is trained based on factors like skin color, thickness, diet, activity, etc. to customize glucose estimation for each patient.

Source

Retrospective sensor calibration method for implantable glucose sensors that eliminates the need for external calibration devices and finger pricks. The method involves measuring the sensor current over time, preprocessing it, applying discrete wavelet decomposition, and using machine learning models to calculate the final glucose value. This allows retrospective calibration without external references. The sensor is also stabilized by applying voltage pulses in anode-cathode cycles.

Source

Calibrating continuous glucose monitors (CGM) to improve accuracy by using a cross-calibration method. The method involves comparing currents from a new CGM with a previously inserted CGM to estimate the new sensor's glucose level. This compensates for background currents and sensor sensitivity changes that occur after insertion. The new sensor's current is converted to a glucose value based on the comparison, rather than relying solely on the new sensor's own measurements. This helps reduce errors during the initial run-in period after sensor insertion.

Source

Blood glucose monitoring system that adjusts the calibration interval of a minimally invasive glucose sensor based on the sensor's error characteristics. The system calculates the time it takes for the sensor error range to reach a threshold level. This time is used to set the calibration interval. This allows personalized calibration intervals that minimize error without unnecessary blood draws. The system also predicts the sensor error range at each measurement time and displays it along with the glucose level.

Source

Correcting blood glucose concentration measurements using machine learning and big data analysis to improve the accuracy of blood glucose monitoring devices. The method involves training a blood glucose change model using continuous glucose data, predicting future glucose levels based on current readings, then correcting those predictions using actual future readings. This correction data is used to fine-tune the blood glucose change model. The corrected model is then used to provide more accurate future glucose predictions.

Source

Calibrating and compensating for drift in continuous glucose monitors using only the sensor data, without external reference points, to improve accuracy over time. The calibration involves correlating sensor readings at known steady-state glucose levels to determine the sensor's sensitivity. It then uses that sensitivity to calculate glucose levels without external calibration. Recalibration is done by re-correlating at new steady-state points. Drift compensation is done by comparing slow-moving average glucose levels over time. Adjustments are made based on the difference between averages. This allows calibration and drift correction entirely from sensor data, improving accuracy without external calibration points.

Source

Improving the accuracy of continuous in-body analyte monitoring by calibrating the sensor readings using known physiological levels. The method involves collecting sensor data over a period, identifying the most frequently occurring signal points, and using those as reference points to derive calibrated analyte levels. This leverages the fact that the body naturally maintains stable analyte levels for most of the time. By identifying the most frequently occurring signals, it can determine the normal physiological level and use that as a reference to calibrate the readings.

Source

Enhancing the accuracy of glucose sensors used in continuous glucose monitoring (CGM) systems for diabetes management by leveraging information from insulin pumps to improve sensor accuracy, particularly during hypoglycemia where CGM accuracy is lowest. The method involves using insulin delivery data along with glucose sensor readings and a filtering algorithm to estimate glucose levels. This estimated glucose is then weighted more heavily than the sensor reading during hypoglycemia to account for the sensor's accuracy issues in that range. The weighting scheme balances sensor and estimated glucose based on factors like insulin delivery and sensor error indices.

Source

Calibrating a continuous glucose monitor (CGM) implanted in the body to improve accuracy by using a second sensor as a reference. The method involves measuring a parameter like pH by the first sensor and converting it to blood glucose. The second sensor also measures the parameter and converts it to blood glucose. The ratio of the first sensor's blood glucose to its parameter is stored. To calibrate the first sensor, its parameter is multiplied by the second sensor's parameter-to-blood glucose ratio and the inverse of the first sensor's ratio. To calibrate the second sensor, its parameter is multiplied by the first sensor's ratio and the inverse of the second sensor's ratio. This corrects the CGM readings.

Source

Factory calibration method for glucose sensors that enables consistent calibration of multiple glucose sensors without requiring frequent user calibration. The method involves acquiring a batch of glucose sensors with consistent parameters, selecting one sensor for analysis, obtaining its glucose response curve and sensitivity decay curve, generating a compensation model based on that, and embedding the model into all the sensors for automatic calibration. This allows accurate factory calibration of multiple sensors that can compensate for variations in enzyme, membrane, etc. without user calibration.

Source

Compensating for temperature effects on glucose sensors to improve accuracy. The compensation involves adjusting estimated glucose levels based on sensor temperature readings. The adjustment accounts for the known relationship between glucose sensor output and temperature. By compensating for temperature variations, it reduces errors caused by fluctuations in body temperature at the sensor site. The compensation can involve models, calibrations, and algorithms to determine the temperature effect on glucose readings.

Source

Estimating blood glucose levels from tissue glucose sensor signals in a way that overcomes the limitations of conventional methods. The disclosed estimator operates in an anti-causal mode where tissue sensor signals lead to serial estimates of blood glucose concentration. It isolates errors in matched blood glucose reference values to determine residual error. The estimator adds this residual error to the reference values to produce estimated blood glucose. This allows more accurate blood glucose estimation from tissue sensors compared to direct correlation.

Source

Optional external calibration of a calibration-free glucose sensor for measuring the level of glucose in a body of a user, wherein the glucose sensor includes physical sensor electronics, a microcontroller, and a working electrode. The method involves periodically measuring electrode current signals, performing an EIS procedure, calculating sensor glucose values using calibration-free models, fusing the values, filtering, and displaying the calibrated sensor glucose. The EIS data provides additional sensor health information beyond the raw current signals. This allows calculating sensor glucose without needing fingerstick calibration references.

Source

Improving calibration reliability and analyte measurement accuracy for medical sensors like continuous glucose monitors (CGM) by mitigating errors in historical calibration points and avoiding over-fitting of new calibration points. The method involves weighting historical reference measurements based on a cost function and using weighted averages for sensor calibration. It also updates lag parameters used for converting interstitial fluid analyte levels to blood levels over time.

Source

Estimating actual glucose levels in a diabetic person when sensor noise or failure is present. The method involves using a probability analysis tool to determine the accuracy of the glucose sensor based on the measured results. The glucose results are then analyzed weighted by the sensor accuracy. A recursive filter estimates the actual glucose level using the weighted results. This allows more reliable glucose level estimation in the presence of sensor noise or failure.

Source

Non-invasive device for measuring glucose levels in the body without drawing blood. The device uses a group of sensors to acquire different biological information from the body. It calculates a regression coefficient vector from the sensor readings and actual glucose levels to generate a calibration model. This model is stored and used to estimate glucose levels based on the sensor readings. The device improves glucose quantification accuracy by generating a personalized calibration model for each user, accounting for factors like skin characteristics, mental state, etc.

Source

Intelligent implantable glucose sensor that improves accuracy by learning and compensating for user-specific errors. The sensor has a training mode where it collects data and learns models for correcting errors due to unmodeled variables like disease state, lifestyle, medications, etc. This training is done in vivo. The sensor then switches to detection mode where it applies the learned models to correct the glucose readings in real time. This allows customized error correction tailored to each user's unique physiology and context. The training and correction logic can be performed on the sensor itself or offloaded to a nearby receiver. The sensor data is also transmitted to a cloud server for population-level modeling and error correction.

Source

Improving accuracy of continuous glucose monitoring system calibration by considering insulin delivery information. The method involves delaying or modifying calibration routines when insulin is being delivered, as rapid glucose changes can introduce error. During calibration, factors like insulin dose, time since delivery, and sensitivity are considered to account for insulin effects. This improves calibration accuracy by factoring in events that impact glucose levels.

Source