Personalized Neural Network Models for Consistent Bio-Information Prediction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Deep learning models used in the medical field often produce inconsistent results for individual users due to variations in bio-information acquisition methods and user-specific biological characteristics, necessitating a personalized approach.

Innovation Solution

A method for acquiring a user-customized neural network model that updates itself based on user-specific bio-information, including first bio-information acquired non-invasively and second bio-information acquired invasively or by medical professionals, through continuous learning processes to enhance accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If a pre-trained deep learning model is used for medical diagnosis, then the model can provide real-time health analysis without specialized medical personnel, but the model produces inconsistent results for individual users due to biological variations

Engineering Contradiction:
Improvereal-time health analysis capabilityVSAvoiddiagnosis consistency
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The system performs preliminary personalization by updating the neural network model with user-specific bio-information before making diagnostic predictions. This preliminary adaptation to individual user characteristics ensures that the model accounts for biological variations before providing real-time analysis, thereby maintaining both productivity and reliability

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The neural network model is designed to be dynamic and adaptable, continuously updating its parameters based on newly acquired user-specific bio-information. This dynamic personalization allows the model to maintain consistent and reliable diagnostic results for each individual user while preserving real-time analysis capabilities

Inventive Principle:
Principle #15Dynamics

2Ease of operation

If a pre-trained deep learning model is provided to individual users, then the model can be deployed widely without specialized personnel, but the model may produce different results for the same type of bio-information across different users

Engineering Contradiction:
Improvemodel deployment accessibilityVSAvoidbio-information prediction accuracy
Core Design Contradiction:
Ease of operationVSMeasurement precision

Solution Approach 1:

The neural network model performs self-personalization by automatically updating its parameters using user-specific bio-information that is easily accessible through mobile devices. This self-service mechanism allows the model to maintain high prediction accuracy for each user without requiring specialized medical personnel for calibration, thus preserving ease of operation while improving measurement precision

Inventive Principle:
Principle #25Self-service

3Reliability

If the neural network model is updated continuously with new user data, then the model becomes personalized and accurate for individual users, but the computational resources and time required for updating increase

Engineering Contradiction:
Improveuser-specific prediction accuracyVSAvoidmodel updating time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The system implements periodic updating of the neural network model rather than continuous updating. The model is updated at scheduled intervals or when sufficient new user-specific bio-information is accumulated, balancing the need for personalization accuracy with the constraint of computational time and resources

Inventive Principle:
Principle #19Periodic action

Data Source

PatentEP4682896A1Method, program, and device for acquiring user-customized neural network model for identifying biometric information
Publication Date: 2026.01.21 MEDICAL AI CO LTD
  • EP4682896A1 patent drawingFigure 1~2
  • EP4682896A1 patent drawingFigure 3~4
  • EP4682896A1 patent drawingFigure 5

AI summary

According to one embodiment of the present disclosure, there is disclosed a method of acquiring a user-customized neural network model that predicts bio-information of a user, the method being performed by a computing device including at least one processor, the method including: acquiring a neural network model trained to predict second bio-information of a different type from that of first bio-information based on the first bio-information; and, when new data for the user is acquired, updating the neural network model based on the acquired new data so that the neural network model is personalized for the user; wherein the new data includes at least one of first bio-information for the user and second bio-information for the user.