Dispense Data Quality Enhancement via Pattern Weighting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems for enhancing data quality of drug dose dispense data sets struggle to reliably differentiate between prime and therapeutic doses, leading to inaccurate insulin tracking and dosing recommendations, particularly in patients with varying insulin requirements.
Innovation Solution
A computing system that segments dispense data into sessions, applies pattern recognition algorithms to identify dispense patterns, and calculates combined pattern weights to label dispense events as either prime or injection events, utilizing historical data and confidence metrics for improved accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If all dispensed insulin amounts are tracked without differentiation, then the tracking system is simple, but the therapy tracking and insulin on board calculations become incorrect
Solution Approach 1:
The system segments dispensed doses into different categories (prime doses vs. therapy doses) based on temporal patterns and dosage characteristics. By dividing the homogeneous set of all dispenses into distinct segments with different clinical meanings, the system maintains both operational simplicity and therapeutic accuracy.
Solution Approach 2:
The system introduces an intermediary classification layer between the raw dispense data and the therapy tracking calculations. This intermediary module analyzes temporal patterns and dosage characteristics to determine whether each dispense should be counted in insulin on board calculations, thereby protecting the reliability of therapy tracking while keeping the user interface simple.
2Extent of automation
If prime doses are distinguished from therapy doses using temporal proximity thresholds, then automated classification is achieved, but accuracy decreases when therapy doses are much smaller than prime doses
Solution Approach 1:
The system changes from using a single temporal proximity threshold parameter to using multiple parameters including temporal patterns, dosage magnitude relationships, and sequential positioning. By adjusting and combining multiple parameters rather than relying on a single threshold, the system achieves both automation and precision in dose classification.
Solution Approach 2:
The classification system is made dynamic by considering the sequence and timing of doses relative to each other. Rather than applying a static threshold, the system evaluates whether a small dose occurs in a position consistent with therapy dosing patterns (e.g., not immediately following another small dose), allowing accurate classification even when dosage magnitudes are similar.
3Reliability
If multiple classification methods are used to improve dose identification accuracy, then reliability improves, but device complexity increases
Solution Approach 1:
The system merges multiple classification indicators (temporal patterns, dosage relationships, sequential positioning) into a unified classification decision. By combining these indicators in an integrated evaluation framework rather than using separate independent systems, the system achieves high reliability while avoiding the complexity of multiple separate classification modules.
Solution Approach 2:
The classification system is designed to be universal by using a single integrated algorithm that handles multiple dose scenarios (prime followed by therapy, multiple primes, single therapy dose) through the same set of evaluation criteria. This multi-functional approach achieves high reliability across diverse dosing patterns without requiring separate specialized classifiers for each scenario.
Data Source
AI summary
A method for enhancing data quality of a drug dose dispense data set to automatically provide dispense data reflecting actually injected dose amounts. For a given dispense session the method comprises the steps of creating a list of possible dispense patterns in accordance with a set of pattern rules and then for each pattern in the list calculate a weight allowing a winning pattern to be determined. Each dispense pattern is a particular sequence of priming events and injection events, with each dispense pattern being a possible interpretation of the dispense events in the current session. For each pattern a combined pattern weight being the product of weight factors for each dispense in the pattern is calculated, wherein each weight factor is determined in accordance with a weight factor vs dispense size function for the given dispense type and dispense size, wherein the larger the dispense size, the more likely it is to represent an injection event and the less likely it is to represent a priming event.