Contextual Data Imputation for Diabetes Testing Gaps

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Users with diabetes, particularly pre-diabetics and non-insulin dependent Type 2 diabetics, often fail to consistently capture and record blood glucose and other relevant data during testing periods, leading to incomplete datasets that hinder accurate disease progression analysis and therapy optimization.

Innovation Solution

A system and method utilizing a user device to capture structured testing data, including blood glucose, dietary intake, and activity data, with intelligent imputation of missing values using contextual information and predictive learning, employing generic and time-weighted blood glucose models to generate substitute data based on population and user-specific patterns.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If users are tasked with structured testing periods to capture complete data, then data reliability is improved, but user compliance deteriorates due to the burden of habitual testing and recording

Engineering Contradiction:
Improvedata reliabilityVSAvoiduser compliance
Core Design Contradiction:
ReliabilityVSEase of operation

Solution Approach 1:

The system performs self-service by automatically imputing missing data values using predictive learning models and contextual information without requiring user intervention. The system monitors itself for data completeness and autonomously fills gaps using learned patterns from user behavior and contextual factors, eliminating the burden on users to manually track every data point while maintaining data reliability for disease progression analysis

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system implements feedback mechanisms by continuously monitoring data completeness and using predictive models to anticipate missing values before they affect analysis. The system learns from user behavior patterns and contextual information to automatically adjust data capture strategies and fill missing values, creating a closed-loop system that maintains reliability without increasing user burden

Inventive Principle:
Principle #23Feedback

2Loss of information

If data imputation is performed using predictive learning, then data completeness is improved, but system complexity increases due to the need for contextual analysis and model training

Engineering Contradiction:
Improvedata completenessVSAvoidsystem complexity
Core Design Contradiction:
Loss of informationVSDevice complexity

Solution Approach 1:

The system performs preliminary action by pre-training predictive learning models on historical data and contextual information before actual data collection begins. The system pre-establishes patterns for data imputation based on user behavior, meal types, activity levels, and other contextual factors, so that when data gaps occur during testing, the models are already ready to quickly generate appropriate imputed values without requiring complex real-time computation

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system uses predictive learning models as intermediary components that bridge the gap between raw data collection and final analysis. These models act as mediators that translate contextual information and patterns into concrete imputed values, simplifying the overall system architecture by encapsulating complex imputation logic within specialized modules that can be trained once and applied repeatedly

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP4256570B1Imputation of data using contextual information
Publication Date: 2025.10.08 F HOFFMANN LA ROCHE & CO AG
  • EP4256570B1 patent drawingFigure 1
  • EP4256570B1 patent drawingFigure 2
  • EP4256570B1 patent drawingFigure 3

AI summary

A user device (e.g., a mobile device) receives testing period data during a testing period. The user device determines that a portion of the data is missing based on an analysis of the received testing period data. The analysis of the received data includes an analysis of one or more blood glucose levels and an analysis of the dietary intake data, the medication data, and the activity data. The user device imputes the missing portion of the data with substitute data determined using predictive learning. The user device calculates a confidence level associated with the substitute data. The user device identifies, using the substitute data, a progression or a regression in a diabetic condition associated with the user.