Closed-Loop Blood Sugar Control for Automated Insulin Delivery

Automatically calculating insulin doses for diabetes management without user input of factors like correction factors and insulin/carb ratios. The method involves two techniques: 1) during online operation (with glucose measurements) it generates personalized insulin doses based on glucose trends, and 2) during offline operation (without glucose measurements) it uses information from previous online periods to calculate insulin doses. This allows regulating blood sugar without manual inputs once the system learns the user's insulin requirements.

Source

Automated method for maintaining blood glucose homeostasis. The method includes generating a glucose target that approaches a desired steady state setpoint from a current glucose value over a prediction horizon, determining an insulin delivery pattern that is more similar to the desired glucose trajectory than any other predicted glucose trajectory, and comparing the desired glucose trajectory against each predicted glucose trajectory and selecting a predicted glucose trajectory that is more similar to the desired glucose trajectory than any other predicted glucose trajectory.

Source

An artificial pancreas system that can improve glycemic control without requiring user input for meals and exercise. The system uses a modified internal model controller (IMC) that automatically adjusts insulin delivery based on food and exercise detection. The IMC eliminates the need for user announcements by generating an insulin correction signal without readjusting the main controller. It also suggests rescue carbohydrates to mitigate hypoglycemia. The IMC has parameters adjusted through optimization to achieve good postprandial control.

Source

Improving glycemic control in people with type 1 diabetes using closed-loop insulin delivery systems like artificial pancreases. The method involves automatically adjusting insulin delivery when a meal is detected but not reported, to prevent hyperglycemia. It also modifies insulin basal rates based on predicted blood glucose levels and detects changes in glucose trends. This helps compensate for uninformed meals and unannounced carb intake. The system uses a machine learning model to predict glucose levels and calculate insulin doses, with adaptive control strengths based on recent glucose trends. It also has a meal priming system that delivers extra insulin when a meal is detected but not reported. This prevents large glucose spikes from unannounced meals. The goal is to provide improved glycemic control in closed-loop insulin delivery systems that can handle un

Source

Adaptive glucose level control system for insulin pumps that optimizes insulin dosing based on historical data. The system adapts basal insulin rates, prediction models, and total daily doses over time based on feedback from actual glucose levels. It learns how much insulin is needed during certain time periods and adjusts future doses accordingly. This allows the pump to adapt to a patient's changing insulin needs over time without constant user input.

Source

Adaptive meal bolus calculation for closed-loop insulin pumps to improve glucose control in people with type 1 diabetes. The adaptation uses case-based reasoning and run-to-run control to automatically adjust meal insulin requirements based on previous meals and insulin responses. This avoids the need for manual bolus adjustments and addresses the high variability of insulin requirements in type 1 diabetes. The adaptive bolus calculator communicates with the closed-loop controller to coordinate insulin delivery.

Source

A method for optimizing insulin dosage in diabetes patients to improve glycemic control and reduce hypoglycemic events. The method involves tracking patient blood glucose levels over time, identifying when and why measurements were taken, and determining if insulin dosage adjustments are needed to bring the patient closer to their desired glucose range without increasing hypoglycemia. The adjustments may involve changing insulin distribution rather than the total dose.

Source

Improving the accuracy of predictive control for closed-loop insulin delivery in artificial pancreas systems. The system uses a physiological model to predict future glucose levels based on insulin, meals, and heart rate. It calculates the optimal insulin dose to avoid hyper/hypoglycemia. The model parameters are calibrated periodically to account for variations. The cost of predicted glucose levels is minimized to find the optimal dose. The maximum insulin dose is limited by a function of patient sensitivity, which can be affected by factors like heart rate.

Source

Reducing the risk of hypoglycemia in insulin pump users by considering the rate of change in blood glucose levels. The system calculates an initial insulin bolus dose based on the current glucose level. It then calculates a revised dose taking into account the rate of change in glucose levels over time. A function is applied to compare the two doses and determine a final insulin value. This final value is used to set the actual insulin bolus delivered to the user. The idea is to account for the delay between when the glucose level is measured and when the user takes the insulin, and to adjust the dose accordingly to avoid overshooting and causing hypoglycemia.

Source

CGM-based automated basal insulin titration system for Type 2 diabetes patients to improve insulin dosing accuracy and safety. The system uses historical CGM data, basal insulin doses, hypoglycemia reports, and past recommendations to generate adjusted insulin doses. It creates personalized dose-response models from CGM metrics like glucose percentiles and estimated fasting levels. Regularization biases early fit towards safe doses. CGM variability guards against overdosing. Checks ensure safe dosing like coherence with estimated BG levels and reductions after severe hypo.

Source

An insulin delivery system for people with diabetes that can automatically manage insulin during fasting periods like overnight sleep. The system calculates correction bolus doses based on blood glucose measurements during the fasting period, compensating for a determined insulin on board. Initially, the insulin on board is assumed based on a calculated correction dose before fasting started. This allows gradual lowering of insulin needs during fasting. The system uses a user's insulin sensitivity factor and other techniques to accurately calculate correction doses during fasting.

Source

Optimizing insulin delivery for diabetes patients using closed-loop data to adjust manual mode settings. The method involves analyzing closed-loop insulin delivery data to generate recommendations for adjusting manual mode settings like basal rates. This analysis is done using a computing device in communication with the patient's insulin pump. The recommendations are then communicated to the pump to automatically adjust manual mode settings during manual delivery. This allows leveraging closed-loop optimization to improve manual mode insulin delivery.

Source

Improving glucose control in diabetes by combining oral medications like SGLT2 inhibitors with automated insulin delivery systems. The approach involves using subtherapeutic doses of SGLT2 inhibitors like empagliflozin, dapagliflozin, or sotagliflozin in conjunction with insulin pumps to mitigate post-meal glucose spikes. The insulin pump adjusts basal rates and parameters based on CGM data to maintain target range. This helps prevent ketoacidosis risk from SGLT2 inhibitors. By adding low-dose oral meds, insulin delivery systems can improve daytime glycemic control beyond what insulin alone provides.

Source

Preemptively alerting a user during the operation of a fluid infusion device like an insulin pump to provide actionable alerts that avoid non-actionable alarms and allow timely intervention. The alerts are based on a homeostasis metric that predicts future glucose levels accounting for insulin on board, future insulin deliveries, and current glucose. This provides alerts for potential hypoglycemia or hyperglycemia before levels change. The alerts recommend actions like insulin adjustments or carb intake to avoid issues. The alerts are cleared when conditions change or higher priority alerts replace them.

Source

Method for adjusting insulin injections in diabetic patients using real-time blood sugar monitoring to optimize insulin dosing and prevent hyperglycemia. The method involves closed-loop control where an algorithm adjusts insulin infusion based on real-time blood sugar levels. It aims to maintain blood sugar within normal range by dynamically infusing insulin in response to real-time monitoring. The algorithm uses blood sugar data to determine the optimal infusion volumes and timings.

Source

Optimizing insulin delivery for new diabetes patients using an artificial pancreas system. The technique involves initially setting an adjusted total daily insulin factor based on user characteristics and comparing it to a maximum delivery threshold. If the factor exceeds the threshold, it is adjusted. Over time, blood glucose levels are monitored to determine glycated hemoglobin. The total daily insulin dose is then modified based on the glycated hemoglobin level. This adaptive approach provides personalized insulin delivery optimization for new diabetes patients using an artificial pancreas system.

Source

Closed-loop control system for managing glucose levels in people with diabetes using an insulin pump. The system uses a multi-model predictive controller that can adapt to rapidly changing glucose dynamics. It executes multiple models with varying parameters to predict future glucose levels. It selects the best model based on measured glucose data to determine optimal insulin doses. This allows the system to handle sudden glucose changes that regular controllers may struggle with.

Source

Intelligent control method for artificial pancreas systems that improves blood sugar management for diabetes patients. The method involves switching between basal rate insulin delivery and miniature bolus insulin delivery based on real-time glucose trends. If glucose is not rising, basal rate is used. If glucose is rising, miniature bolus delivers extra insulin in smaller doses over time to prevent spikes from foods with less carbs absorbed. This adaptive insulin delivery aims to maintain normal blood sugar levels.

Source

Artificial pancreas control method to improve safety and efficiency of closed-loop insulin delivery for diabetes management. The method involves using a hybrid approach that combines model predictive control (MPC) with safety checks. It judges the rate of blood sugar change and determines if it's below a threshold. If so, it stops insulin infusion until stability is restored. If the sugar is rising, it uses MPC to optimize insulin delivery. This hybrid approach balances MPC's accuracy with safety checks to prevent hypoglycemia and minimize MPC calculations.

Source

A closed-loop artificial pancreas system for insulin infusion in diabetes management that uses a modified Proportional-Integral-Derivative (PID) algorithm to precisely control insulin delivery. The modified PID algorithm converts the asymmetric blood glucose levels in the physical space to a symmetric risk space using a conversion function. This allows using the simple and robust PID algorithm while gaining the flexibility and precision of the original conversion. The closed-loop system uses a glucose sensor, program module, and infusion pump.

Source

Closed-loop artificial pancreas system for insulin infusion that accurately calculates the required insulin amount without relying on external devices. The system has a detection module with an algorithm to directly calculate the current insulin requirement based on the user's blood glucose level. This avoids issues with delayed or misaligned transmission of blood glucose data that can lead to inaccurate insulin dosing. The calculated insulin amount is then sent to the infusion module for delivery.

Source

Closed-loop artificial pancreas insulin infusion control system that provides more accurate insulin dosing for diabetes management. The system uses three algorithms: a first algorithm to calculate initial insulin dose, a second algorithm to calculate secondary insulin dose, and a third optimization algorithm to combine and refine the doses. The optimization algorithm improves dose accuracy by further adjusting the initial and secondary doses based on risk conversion of blood glucose levels. This compensates for asymmetry in glucose risk versus deviation and enables more precise insulin infusion in closed-loop artificial pancreas systems.

Source

Closed-loop artificial pancreas insulin infusion system that combines the Proportional-Integral-Derivative (PID) algorithm and Model-Predictive-Control (MPC) algorithm to improve insulin infusion accuracy. The PID algorithm is used with intermediate MPC values, and the MPC uses PID output. This hybrid approach leverages the strengths of both algorithms to provide more precise insulin dosing compared to using just PID or MPC alone.

Source

Compound artificial pancreas algorithm for closed-loop insulin infusion in diabetes management. The algorithm optimizes the insulin infusion calculation by converting asymmetric blood glucose levels into a more symmetric risk space. This involves methods like segmented weighting, relative value, and improved control variability grid analysis. It allows accurate insulin dosing in complex scenarios where the classic PID and MPC algorithms struggle. The optimized insulin amounts from two algorithms are further combined to provide the final infusion. This compound algorithm provides more reliable and precise insulin dosing for closed-loop artificial pancreas systems.

Source

Closed-loop artificial pancreas system for precise insulin delivery to diabetic patients. The system uses optimized algorithms to convert asymmetric blood glucose risks into symmetric risks, allowing reliable drug infusion calculations. The algorithms include the rMPC algorithm for segmented weighting, relative value, BGRI, and CVGA blood glucose risk conversions. They also have compound artificial pancreas algorithms to optimize infusion amounts from multiple algorithms. This addresses complex scenarios with asymmetric blood glucose risks.

Source

Closed-loop artificial pancreas insulin infusion control system for diabetics that accurately calculates the required insulin infusion volume without relying on external devices. The system has a detection module that calculates the required insulin based on the current blood sugar level. This avoids delays and errors in transmitting blood sugar values between devices. The calculated insulin is then infused. The calculation algorithms convert asymmetric blood sugar risk to symmetric risk space using methods like relative value conversion, blood sugar index conversion, and improved control variability grid analysis.

Source

Closed-loop artificial pancreas insulin infusion control system using an optimized Proportional-Integral-Derivative (PID) algorithm for precise insulin delivery. The PID algorithm is modified to convert asymmetric blood sugar levels in physical space to approximately symmetrical risk levels in a risk space. This allows retaining the simplicity and robustness of PID while improving precision and flexibility for closed-loop insulin delivery. It converts blood sugar deviations into risk values that are symmetrical around zero. The converted risk is used to calculate insulin infusion amounts. This compensates for the asymmetric distribution of blood sugar levels and risks in diabetes.

Source

Improved closed-loop artificial pancreas system for more accurate insulin dosing in diabetic patients using a composite algorithm. The system involves three algorithms: a classic PID algorithm, a risk-modified PID algorithm, and a third optimization algorithm. The classic PID calculates an initial insulin dose based on current glucose. The risk-modified PID converts glucose deviations into risk space. The optimization algorithm combines the initial and modified doses to find the final dose. This further processing improves accuracy by compensating for delays and reducing errors.

Source

Closed-loop artificial pancreas system for precise insulin delivery to diabetic patients. The system uses a composite algorithm to calculate insulin infusion volumes. It involves two algorithms, a classic PID algorithm and a modified algorithm like risk-based PID (rPID) or risk-based model predictive control (rMPC). The first algorithm calculates an initial insulin amount. The second algorithm calculates another insulin amount. These are optimized and averaged to get the final insulin infusion volume. This improves accuracy compared to using a single algorithm. The system also has features like meal and motion recognition.

Source

A closed-loop artificial pancreas system for precise insulin delivery to diabetic patients. The system uses an optimized algorithm called rPID (risk-based proportional-integral-derivative) to calculate insulin doses. The algorithm converts blood glucose deviations from target into a risk space, allowing accurate dosing across the range. It also factors in delays to compensate for insulin absorption and glucose sensing. The system has a detection module, program module, and infusion module. The program module uses rPID to calculate insulin based on blood glucose. It can also have a secondary algorithm for optimization. The infusion module infuses insulin based on the calculated dose.

Source

Closed-loop artificial pancreas system for precise insulin delivery in diabetes management. The system uses an optimized model predictive control (MPC) algorithm called rMPC to convert asymmetric blood sugar into a symmetrical risk space. This allows precise control of insulin infusion while considering the asymmetry of blood sugar distribution. The rMPC algorithm calculates insulin infusion instructions based on risk conversion instead of absolute blood sugar values. This improves accuracy compared to traditional MPC algorithms. The system also compensates for delays in insulin absorption, interstitial fluid glucose sensing, and blood glucose sensing.

Source

Artificial pancreas system that can be used for therapeutic purposes. The system includes a wearable blood glucose monitoring device, an insulin delivery system, and a computing device communicatively coupled to perform various aspects of the system.

Source

Intelligent insulin management system that adapts insulin infusion rates based on glucose trends and meal intake to prevent hypoglycemia and hyperglycemia. The system uses a glucometer to measure blood glucose levels at intervals. It calculates insulin infusion rates using glucose readings and adaptive multipliers based on glucose trends. If glucose drops rapidly, it shortens the measurement interval. If glucose stability is good, it increases the interval. This prevents over-infusion during fast drops and under-infusion during slow drops. It also coordinates meal boluses with intravenous infusions. The system can transition to subcutaneous insulin delivery if glucose stability is sustained.

Source

Automated insulin delivery system for people with diabetes that provides a lower and narrower target glucose range during normal activity to improve glycemic control without increasing hypoglycemia risk. The system lets users select between a standard and alternate normal mode with tighter glucose ranges. It calculates and delivers insulin doses based on CGM data according to the chosen mode. The alternate mode has a lower and narrower range but prevents automatic correction boluses when glucose is high and falling rapidly to reduce hypo risk.

Source

Automated insulin delivery system with a user-selectable feature to provide more aggressive glucose control during normal activity without increasing hypoglycemia risk. The feature is an alternate normal activity mode with a lower and narrower glucose range compared to the standard range. It also has a bolus lockout to prevent over-correction boluses if glucose is rising quickly. This allows users to choose a more aggressive mode for better glucose control during normal activity without the risk of hypoglycemia.

Source

A feature for closed-loop insulin pumps to better handle meals and prevent high post-meal glucose levels. It allows the user to activate an "Eating Soon" mode when they anticipate eating soon. This mode lowers the insulin target range and delivers a correction bolus to reduce glucose levels before the meal. After a period or meal entry, normal target range resumes. This prevents overcorrecting or undercorrecting for meals.

Source

Automated blood glucose regulation system for diabetes management that improves performance and reduces risks of hyper/hypoglycemia when users forget to declare meals. The system has a method to automatically detect and estimate missed meals. If a missed meal is detected, it estimates the likely meal size and time based on historical probabilities. If the missed meal probability is low, it falls back to a non-meal-specific regulation. If the user confirms a missed meal, it uses the estimated values. This prevents long hyperglycemia periods from non-specific regulation if meals are omitted.

Source

Applying metaheuristic optimization techniques like ant colony optimization and genetic algorithms to design controllers for managing blood glucose levels in people with diabetes. The controllers are based on the PID control method and are optimized using metaheuristic algorithms to find the best parameters for controlling glucose. The goal is to find the optimal controller settings that can balance the tradeoff between response time and insulin dosage. By using metaheuristics to optimize the controllers, it enables finding the best parameters for specific patient conditions like parameter uncertainties, meal disruptions, and sensor noise.

Source

Personalized blood sugar control system for diabetes management that learns and adapts insulin dosing based on a patient's unique glucose metabolism characteristics. The system continuously learns a patient's blood sugar trends, history, and individual glucose response to insulin. It then uses this personalized model to determine optimal insulin dosage and rate. This allows tailoring insulin delivery to each patient's unique glucose dynamics, improving glycemic control and reducing risk of hypoglycemia or hyperglycemia. The system consists of a continuous glucose monitor, insulin pump, and learning model that adapts over time as new data is received.

Source

Customizing insulin delivery in closed loop artificial pancreas systems to better suit individual user needs. The closed loop algorithm uses a cost function to determine insulin dosages. The cost function balances penalties for glucose and insulin excursions. To avoid over or under dosing based on generic parameters, the cost function coefficients for glucose and insulin penalties are customized based on the user's daily insulin needs. This tailoring of the cost function helps smooth insulin delivery and prevent hypo/hyperglycemia.

Source

Adaptive closed-loop insulin delivery system for managing diabetes that can accommodate different types of insulin with varying pharmacokinetic profiles. The system uses a multi-model predictive controller (MMPC) algorithm that adapts to changes in insulin type and pharmacokinetics. The MMPC algorithm integrates multiple models representing different insulin types with varying kinetics. It selects the best fitting model based on past data to determine optimal insulin dosing. This improves control of glucose levels when switching between insulins with different kinetics. The system also has adaptability to changing pharmacokinetics over time, such as the Tamborene effect from long-term continuous subcutaneous infusion. The MMPC algorithm can handle these changes by adjusting the selected model based on historical data.

Source

A blood sugar monitoring and control system that uses a continuous glucose monitor, a warning device, and a control device to automatically inject insulin when glucose levels get too high. The system collects real-time glucose values, determines control status based on target ranges, and responds with emergency measures like insulin injection if needed. It integrates glucose data and patient characteristics to calculate optimal insulin doses. This helps prevent hyperglycemia complications by proactively adjusting insulin when levels get too high.

Source

Automated insulin delivery system with safety constraints to prevent under- or over-delivery of insulin based on sensor inputs. The constraints mitigate errors like missing or erroneous sensor data and unexpected events like unplanned meals. The system uses an artificial pancreas algorithm to monitor glucose levels and determine insulin doses. When sensor inputs are missing or erroneous, the system delivers insulin below a fixed personalized basal rate for a reliable time period. If glucose measurements are unreliable, insulin delivery is suspended. This prevents under-delivery during sensor gaps. The system also limits insulin delivery rate changes during events like meal detection to prevent over-delivery.

Source

Diabetes management system that combines automatic basal insulin delivery using a model predictive control algorithm with manual bolus insulin doses initiated by the user. The system modifies manual bolus doses based on the user's physiological data to avoid overinsulinization when both doses are delivered simultaneously. This prevents issues like hypoglycemia that can occur when the basal and bolus calculations aren't coordinated. The system calculates the total insulin dose based on the modified bolus and calculated basal, and delivers it.

Source

Automated insulin delivery system for managing blood glucose levels in people with diabetes using an artificial pancreas (AP) that can provide better post-meal glucose control without the need for prandial insulin injections. The AP adjusts insulin infusion rates based on real-time glucose levels using a model predictive control (MPC) algorithm. To improve post-meal glucose response, the MPC detunes parameters when an unannounced meal is detected. The detuning involves reducing penalties for insulin deviation and infusion differences to allow faster insulin infusion rates. This compensates for the delayed absorption of some insulins to better match physiological insulin profiles and prevent hyperglycemia. The AP can also use insulins with faster absorption properties to further enhance post-meal glucose control.

Source

Range-controlled failsafes for insulin pumps to mitigate risks from inaccurate glucose measurements and changes in insulin sensitivity in continuous glucose monitoring (CGM) systems. It calculates basal rate adjustments using an algorithm and failsafe constraints to account for CGM errors. The algorithm factors in glucose level and rate of change. The failsafes protect against unreliable increases/decreases in basal rates due to CGM inaccuracies or insulin sensitivity issues.

Source

Reducing hypoglycemia risk in automated insulin delivery systems by preventing excessive correction boluses that could cause low blood sugar levels. The system analyzes factors like glucose trend and future predictions before delivering a correction bolus. If delivery would potentially result in low blood sugar, it prevents the bolus for a period to allow stabilization. This avoids overcorrecting and countering rescue carbs, reducing oscillations and hypoglycemia.

Source

Velocity-weighted model predictive control algorithm for artificial pancreas (AP) systems in Type 1 diabetes treatment to prevent controller-induced hypoglycemia. The algorithm penalizes predicted glucose outcomes based on glucose velocity, with less penalty as glucose rates become more negative. This prevents overcorrection when glucose levels converge to normal. It allows independently shaping AP response to hyperglycemic excursions, especially downhill. The velocity-weighting is integrated into the MPC cost function.

Source

An artificial pancreas control system that minimizes and prevents hypoglycemia during and after exercise in people with type 1 diabetes. The system uses a multi-stage model predictive control (MS-MPC) algorithm that incorporates exercise profiles, anticipatory and reactive modes, and exercise-aware premeal boluses to optimize insulin infusion. The MS-MPC predicts glucose uptake during exercise based on individual profiles, then adjusts basal insulin in advance to prevent hypoglycemia. It also responds to real-time exercise signaling.

Source

Glucose control systems for managing blood sugar levels that include features like software update techniques to avoid interrupting therapy delivery, gesture-based control of therapy delivery, automatic resumption of therapy after pause, improved alarm management, display of autonomously calculated dosing recommendations, wide area network connectivity, and security features. The systems can have an infusion pump that delivers insulin and/or other glucose control agents. They allow modifying therapy settings like insulin doses and rate, meal doses, correction doses, and glucose targets. Eligibility for modifying settings can be determined based on factors like test results, history, or authorized access levels. The systems can also provide remote viewing of therapy data and reports.

Source