An auxiliary screening method for genetic metabolic diseases based on multi-domain fusion learning

A metabolic disease and screening technology, applied in the field of information technology applications, can solve the problems of difficult screening data, waste of medical resources, reducing the screening accuracy of the threshold method, etc., and achieve the effect of reducing the false positive rate

Active Publication Date: 2021-10-08
ZHEJIANG UNIV
View PDF6 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] The high false positive rate is a major shortcoming of existing screening methods, that is, most of the recalled suspected cases are actually healthy people, which brings unnecessary waste of medical resources
The reason is that the main problems causing the high false positive rate are concentrated in three aspects: (1) The threshold method sets a separate abnormal concentration cut-off value for each metabolite, and this linear method ignores the correlation between metabolites (2) There are certain differences in the concentration distribution of metabolites in the population of different regions, and it is difficult to integrate and analyze the screening data. Therefore, each screening center can only rely on the data samples in the region to establish cut-off value standard, and the reduction of statistical samples, especially positive cases, will reduce the screening accuracy of the threshold method; (3) The huge birth population in my country increases the work pressure of each pediatrician, and at the same time, qualifications, emotions, etc. will also become factors. Potential Factors Affecting Interpretation

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • An auxiliary screening method for genetic metabolic diseases based on multi-domain fusion learning
  • An auxiliary screening method for genetic metabolic diseases based on multi-domain fusion learning
  • An auxiliary screening method for genetic metabolic diseases based on multi-domain fusion learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0036] The technical solution of the present invention will be further described below in conjunction with examples.

[0037] The assisted screening method for genetic metabolic diseases based on multi-domain fusion learning of the present invention includes the following two stages:

[0038] The first stage:

[0039] For data from m regions or screening centers, train m neural networks with the same structure but different network parameters;

[0040] In this example, each neural network is set to include four hidden layers, and the number of neurons in each layer is 16, 8, 8, and 4 in sequence; the hidden layer uses ReLU as the activation function, and the output layer uses Sigmoid as the activation function.

[0041] second stage:

[0042] Establish the main neural network. The main neural network includes a frozen layer, a springboard layer, and a specific domain layer; in this example, the main neural network is set to include m*2 frozen layers, 2 springboard layers, an...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses an auxiliary screening method for genetic metabolic diseases based on multi-domain fusion learning. The method projects the screening data nonlinearly into the latent space for representation, and establishes the constraint on the distribution difference of metabolites in different regions to realize multi- Unified modeling of regional screening data, the invention uses the nonlinear mapping of the neural network to represent the correlation between different metabolites; uses the multi-domain fusion technology in the main neural network to establish a unified model based on multi-regional screening data model; and the total amount of data is increased through data fusion, so that the model reduces the false positive rate while keeping the recall rate unchanged.

Description

technical field [0001] The invention belongs to the field of information technology applications, and relates to an auxiliary screening method for genetic metabolic diseases, in particular to an auxiliary screening method for genetic metabolic diseases based on multi-domain fusion learning. Background technique [0002] Hereditary metabolic diseases are a large class of genetic diseases with metabolic function defects. The mutated gene changes the original protein code and affects the synthesis of enzymes. Due to the lack of enzymes, biomolecules in some metabolic pathways cannot be effectively decomposed, and the concentration of metabolites upstream and downstream of the pathway is out of the normal range, resulting in abnormal symptoms in the body, such as mental retardation, developmental delay and epilepsy, and severe cases may even face death. Early diagnosis and early treatment can greatly improve the prognosis, improve the quality of life of patients, and reduce the...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Patents(China)
IPC IPC(8): G16H50/70G06N3/08
CPCG06N3/08G16H50/70
Inventor 尹建伟林博舒强李莹邓水光蒋萍萍杨茹莱张鹿鸣尚永衡
Owner ZHEJIANG UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products