Multi-task Learning Model for Chronic Disease Prediction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods fail to predict multiple chronic diseases simultaneously by leveraging potential relationships among them.

Innovation Solution

A chronic disease prediction system based on a multi-task learning model, comprising a shared layer convolutional neural network and multiple chronic disease branch networks, which preprocesses and codes physical examination data using a five-fold cross validation method to perform feature extraction and prediction across various diseases.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If separate prediction models are used for each chronic disease, then the prediction accuracy for individual diseases can be maintained, but the system cannot leverage potential relationships among various chronic diseases and cannot predict multiple diseases simultaneously

Engineering Contradiction:
Improvecapability to predict multiple chronic diseases simultaneouslyVSAvoidprediction accuracy
Core Design Contradiction:
Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The patent merges multiple single-disease prediction models into a unified multi-task learning framework. The shared convolutional neural network layers extract common features from physical examination data that are relevant across multiple chronic diseases, while task-specific branches handle disease-specific predictions. This combining approach enables simultaneous prediction of multiple diseases while leveraging their potential relationships through shared feature representations.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The shared convolutional neural network layers serve multiple functions by extracting general features that are applicable to various chronic disease predictions. These universal feature extraction capabilities are then utilized by different task-specific branches, allowing the system to handle multiple prediction tasks with a single unified structure rather than requiring separate specialized models for each disease.

Inventive Principle:
Principle #6Universality (Multi-functionality)

2Adaptability or versatility

If a multi-task learning model with shared layers is used to predict multiple diseases simultaneously, then the system can leverage relationships among diseases, but the model complexity increases

Engineering Contradiction:
Improvecapability to predict multiple chronic diseases simultaneouslyVSAvoidmodel structure complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The multi-task learning model is segmented into shared convolutional neural network layers for common feature extraction and task-specific branches for disease-specific predictions. This segmentation allows the complex model to be organized in a modular fashion, where the shared layers handle general pattern recognition and the task-specific branches handle disease-specific logic, making the overall complex system more manageable and interpretable.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentUS20220254493A1Chronic disease prediction system based on multi-task learning model
Publication Date: 2022.08.11 ZHEJIANG UNIV
  • US20220254493A1 patent drawing
  • US20220254493A1 patent drawing
  • US20220254493A1 patent drawing

AI summary

A chronic disease prediction system based on a multi-task learning model. The system includes a computer memory, a computer processor and a computer program which is stored in the computer memory and executable on the computer processor, wherein a trained chronic disease prediction model is stored in the computer memory, and the chronic disease prediction model is composed of a shared layer convolutional neural network and a plurality of chronic disease branch networks; and when executing the computer program, the computer processor implements the following steps: preprocessing a to-be-predicted physical examination record and then inputting the record into the shared layer convolutional neural network of the chronic disease prediction model for feature extraction to obtain a feature map, and inputting the obtained feature map into each chronic disease branch network and performing feature extraction and prediction respectively to obtain a chronic disease prediction result.