Adaptive CGM Alert Horizon for Earlier Glucose Warnings
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing continuous glucose monitoring (CGM) systems generate alarms only when blood glucose levels are already problematic, providing insufficient advance warning for users to take preventive action.
Innovation Solution
A prediction system modifies the prediction horizon for glucose level alerts based on user feedback, monitored glucose levels, and additional data to provide timely warnings before threshold values are reached, adjusting the alert timing to avoid nuisance alerts and enhance user intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If glucose alerts are generated only when threshold values are reached, then false alarms are reduced, but users receive insufficient advance warning and cannot take preventive action
Solution Approach 1:
The system performs preliminary action by predicting future glucose values using a prediction model before the actual threshold is reached. The prediction horizon extends into the future to identify when glucose levels are likely to cross thresholds, allowing the system to generate advance warnings that enable users to take preventive action before hyperglycemia or hypoglycemia occurs.
Solution Approach 2:
The system implements feedback by monitoring user responses to alerts and using this information to dynamically modify the prediction horizon. When users interact with alerts or take corrective actions, the system adjusts the prediction horizon length to optimize future alert timing, balancing advance warning with alert accuracy based on actual user behavior and glucose level patterns.
2Loss of time
If the prediction horizon is extended to provide more advance warning, then users can take preventive action, but the risk of nuisance alerts increases
Solution Approach 1:
The system applies dynamics by making the prediction horizon adjustable and adaptive rather than fixed. The prediction horizon is dynamically modified based on user feedback, glucose level trends, and prediction confidence levels. This allows the system to extend the horizon when high confidence predictions indicate genuine risk, while shortening it or suppressing alerts when prediction uncertainty increases the likelihood of nuisance alerts.
Solution Approach 2:
The system changes parameters by modifying the prediction horizon length based on multiple factors including user feedback, glucose variability, and prediction model confidence. The prediction horizon is adjusted as a variable parameter rather than a constant, allowing optimization of advance warning time while minimizing nuisance alerts through data-driven parameter tuning.
3Reliability
If multiple prediction horizons are used to increase confidence in predictions, then alert reliability improves, but system complexity increases
Solution Approach 1:
The system segments the prediction process by using multiple prediction horizons (e.g., short-term, medium-term, long-term predictions) to evaluate glucose levels at different future time points. Each horizon provides information about different time scales, and the system integrates these segmented predictions to form a comprehensive view of future glucose trends, improving confidence while maintaining manageable complexity through modular prediction evaluation.
Data Source
AI summary
Data describing glucose measurements is received from a continuous glucose monitoring (CGM) system worn by a user and predicted glucose values during a future time period are generated for the user based on the data. A determination is made that at least one of the predicted glucose values satisfies a threshold value for an alert, which is associated with a prediction horizon that defines an amount of time prior to satisfaction of the threshold value for communicating the alert to the user. Output of the alert is caused responsive to determining that the at least one predicted glucose value satisfies the threshold value for the alert within the prediction horizon, relative to a current time. The prediction horizon is modified based on a user response to the alert. Output of a subsequent instance of the alert is caused based on the modified prediction horizon.