ECG Diagnosis Criteria Extraction Using AI and Rule-Based Analysis

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The process of extracting rules and criteria for rule-based ECG analysis models is arduous, inefficient, and often results in inaccurate or incomplete outcomes, while AI models, though efficient and accurate, lack interpretability and are prone to bias.

Innovation Solution

A system and method that uses an AI model to extract interpretable criteria for a rule-based ECG analysis model, integrating features from ECGs to determine diagnoses, enhancing the model's performance and reducing bias.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If a rule-based ECG analysis model is used to determine diagnoses with fully-interpretable criteria, then the interpretability and ease of operation are improved, but the efficiency and accuracy of diagnosis extraction deteriorate

Engineering Contradiction:
Improveinterpretability of diagnosis criteriaVSAvoidefficiency of rules extraction
Core Design Contradiction:
Ease of operationVSProductivity

Solution Approach 1:

An AI model serves as an intermediary system that automatically extracts diagnosis criteria from ECG data, bridging the gap between raw ECG information and interpretable diagnostic rules. This intermediary AI extractor handles the complex extraction task that would otherwise be arduous and inefficient for human experts to perform manually.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Ease of operation

If manual extraction of rules and criteria is performed by subject-matter experts, then the interpretability is maintained, but the efficiency and accuracy of extraction deteriorate

Engineering Contradiction:
Improveinterpretability of diagnosis criteriaVSAvoidtime required for rules extraction
Core Design Contradiction:
Ease of operationVSLoss of time

Solution Approach 1:

The patent replaces the manual mechanical process of expert extraction with an automated AI-based extraction system. The AI model processes ECG data and generates diagnostic criteria automatically, eliminating the time-consuming manual extraction process while maintaining the interpretability advantage of rule-based systems.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

3Productivity

If AI models are used for ECG interpretation to improve efficiency and accuracy, then the productivity is improved, but the interpretability and reliability deteriorate

Engineering Contradiction:
Improveefficiency of ECG analysisVSAvoidinterpretability of decision-making
Core Design Contradiction:
ProductivityVSEase of operation

Solution Approach 1:

The system segments the ECG analysis function into two distinct components: an AI-based extractor that efficiently identifies potential diagnoses and criteria, and a rule-based analyzer that provides interpretable reasoning and verification. This segmentation allows each component to optimize for its specific strength while working together to achieve both efficiency and interpretability.

Inventive Principle:
Principle #1Segmentation

4Measurement precision

If AI models are used for ECG interpretation to improve accuracy, then the measurement precision is improved, but the reliability deteriorates due to bias from training data

Engineering Contradiction:
Improveaccuracy of diagnosisVSAvoidbias in diagnosis
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The rule-based analyzer provides feedback verification of the AI extractor's outputs against established diagnostic criteria and clinical knowledge. This feedback mechanism allows the system to identify and correct potential biases in the AI model's reasoning, ensuring that diagnoses align with reliable medical standards while maintaining the accuracy benefits of AI processing.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP4670632A1System and method for extracting, using an artificial intelligence model, criteria for determining a diagnosis by a rule-based ECG analysis model
Publication Date: 2025.12.31 GE PRECISION HEALTHCARE LLC
  • EP4670632A1 patent drawingFigure 1
  • EP4670632A1 patent drawingFigure 2
  • EP4670632A1 patent drawingFigure 3

AI summary

A system and method for determining, by a rule-based ECG analysis model, a diagnosis of an ECG using criteria extracted by an AI model are provided. An ECG may be received by the rule-based ECG analysis model. Features of the ECG may be determined by the rule-based ECG analysis model. The diagnosis may be determined by the rule-based ECG analysis model using the features of the ECG and the criteria extracted by the AI model. The diagnosis may be transmitted.