ECG Risk Prediction Model for Future Atrial Fibrillation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional ECG analysis is ineffective in forecasting the likelihood of future atrial fibrillation or other medical events, and current screening strategies for atrial fibrillation are inadequate, often missing cases due to its low incidence and paroxysmal nature.

Innovation Solution

A deep neural network model is employed to analyze electrocardiogram data, incorporating age and sex values, to predict a risk score for the likelihood of developing cardiovascular conditions within a predetermined period, using a 12-lead ECG and electronic health record data to generate a risk score for atrial fibrillation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If conventional ECG analysis is used, then the analysis process is simple, but it is ineffective in forecasting the likelihood of future atrial fibrillation

Engineering Contradiction:
Improveforecasting accuracyVSAvoidmodel complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent replaces conventional visual ECG analysis with a deep neural network model that automatically processes ECG signals. The model uses convolutional neural networks to extract features from ECG traces and predicts future atrial fibrillation risk, substituting manual mechanical analysis with automated intelligent systems.

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

Solution Approach 2:

The patent transforms the ECG analysis by changing from simple visual inspection parameters to multiple quantitative parameters including P-wave duration, P-wave morphology, heart rate variability, and other electrocardiographic features that are processed through the neural network to predict AF risk.

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If current screening strategies are used, then the screening process is straightforward, but it often misses cases due to low incidence and paroxysmal nature

Engineering Contradiction:
Improvedetection accuracyVSAvoidscreening system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent performs preliminary action by predicting future atrial fibrillation risk before the condition actually occurs. The deep learning model analyzes current ECG patterns to identify patients at high risk of developing AF in the future, enabling preventive intervention before the paroxysmal episodes begin.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent introduces an intermediary deep neural network system between the ECG data and the diagnosis. This intermediary model processes complex ECG patterns and translates them into risk predictions, acting as a mediator that enhances detection accuracy for subtle paroxysmal patterns that conventional methods miss.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Measurement precision

If deep neural network model is employed, then prediction accuracy improves, but computational requirements increase

Engineering Contradiction:
Improverisk prediction accuracyVSAvoidcomputational energy
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent segments the deep neural network into multiple specialized components including separate convolutional neural network branches for different ECG lead analyses, feature extraction layers, and prediction modules. This segmentation allows efficient parallel processing and reduces overall computational energy requirements while maintaining high prediction accuracy.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP4030997B1ECG based future atrial fibrillation predictor systems and methods
Publication Date: 2026.01.28 TEMPUS AI INC
  • EP4030997B1 patent drawingFigure 1
  • EP4030997B1 patent drawingFigure 2
  • EP4030997B1 patent drawingFigure 3

AI summary

A method and system for predicting the likelihood that a patient will suffer from atrial fibrillation is provided. The method includes receiving electrocardiogram data associated with the patient, providing at least a portion of the electrocardiogram data to a trained model, receiving a risk score indicative of the likelihood the patient will suffer from atrial fibrillation within a predetermined period of time from when the electrocardiogram data was generated, and outputting the risk score to at least one of a memory or a display for viewing by a medical practitioner or healthcare administrator. The system includes at least one processor executing instructions to carry out the steps of the method.