Back to results
US 12,597,517 B2Granted
Method for extracting intrinsic properties of cancer cells from gene expression profiles of cancer patients and device for the same
AssigneeKorea Advanced Institute of Science and Technology
InventorsKwang-Hyun Cho, Dongsan Kim
Life of this patent
Estimating the term…
Abstract
Provided is a method of predicting a condition of a patient, including decomposing an input matrix to produce a residual matrix, the input matrix representing patients and expression levels of genes of the patients, training a classifier by using health condition values the patients as learning criteria and inputting the residual matrix into the classifier, and obtaining a value for predicting a health condition of a first patient, by inputting a first input matrix including the expression levels of genes of the first patient into the classifier, the first residual matrix being a residual matrix obtained from the first input matrix by using the predetermined algorithm.
Claims9
A method of predicting a condition of a patient, comprising:
- analyzing, by a gene expression level analysis device, expression levels of genes of the patient to obtain a first input matrix, a column of the first input matrix corresponding to the expression levels of genes of the patient;
- and
- obtaining, by a computing device, a value for predicting the health condition of the patient by inputting the column of the first input matrix into an input layer of a classifier using a neural network, wherein the classifier has been trained using a residual matrix obtained from gene expression data of a plurality of patients, such that the classifier is capable of assigning a same group to test samples collected from different sites of a body of a same patient;
- wherein the classifier has been trained by a predetermined supervised learning method,
- wherein the predetermined supervised learning method comprises:
- analyzing, by the gene expression level analysis device, expression levels of genes from respective cells obtained from a second plurality of patients;
- calculating, by the computing device, a second input matrix representing gene expression levels for the second plurality of patients, by receiving the expression levels of genes analyzed for the second plurality of patients from the gene expression level analysis device;
- decomposing, by the computing device, the second input matrix into a sum of a product of a plurality of element matrices and a second residual matrix by using a non-negative matrix factorization (NMF) algorithm;
- and
- training, by the computing device, the classifier by supervised learning by using health condition as learning criteria and inputting the second residual matrix into the input layer of the classifier,
- wherein the non-negative matrix factorization (NMF) algorithm decomposes a given matrix into a sum of a product of a plurality of element matrices and one residual matrix,
- and wherein the second residual matrix represents information that is different from a product of the plurality of element matrices.
9 more elements
A method of predicting a health condition of a patient, comprising:
- analyzing, by a gene expression level analysis device, expression levels of genes from respective cells obtained from a second plurality of patients to obtain a second input matrix representing gene expression levels for the second plurality of patients;
- decomposing, by a computing device, the second input matrix into a sum of a product of a plurality of element matrices and a second residual matrix by using a non-negative matrix factorization (NMF) algorithm, wherein the second residual matrix is a matrix having the same size as the second input matrix and is different from a product of the plurality of element matrices;
- training, by the computing device, a classifier using a neural network by a supervised learning method by using health condition as learning criteria and inputting the second residual matrix into the input layer of the classifier, wherein the classifier is trained such that test samples collected from different sites of a body of a same patient are assigned to a same group;
- and
- obtaining, by the computing device, a value for predicting the health condition of the patient, by inputting a column of a first input matrix into the input layer of the trained classifier, the column of the first input matrix including expression levels of genes of the patient,
- wherein the non-negative matrix factorization (NMF) algorithm decomposes a given matrix into a sum of a product of a plurality of element matrices and one residual matrix.
3 more elements
A system for predicting a health condition of a patient, comprising:
- a gene expression level analysis device;
- and
- a computing device,
- wherein the gene expression level analysis device is configured to analyze expression levels of genes of the patient and analyze expression levels of genes from respective cells obtained from a given plurality of patients,
- the computing device comprises:
- a residual generation unit configured to output a residual matrix by receiving an input matrix representing the expression levels of genes for each of the given plurality of patients and decomposing the input matrix into a sum of a product of a plurality of element matrices and a residual matrix, wherein the residual matrix is a matrix having the same size as the input matrix and is different from a product of the plurality of element matrices;
- and
- a classifier using a neural network, and the classifier being trained by a predetermined supervised learning method, wherein the classifier is trained such that test samples collected from different sites of a body of a same patient are assigned to a same group;
- and
- a processing unit,
- the processing unit is configured to:
- generate a first input matrix including a column corresponding to the expression levels of genes of the patient;
- and
- calculate a value for predicting the health condition of the patient by inputting the column of the generated first input matrix into an input layer of the classifier, and
- the predetermined supervised learning method comprises:
- analyzing, by the gene expression level analysis device, expression levels of genes from respective cells obtained from a second plurality of patients;
- calculating, by the computing device, a second input matrix representing gene expression levels for the second plurality of patients by receiving the expression levels of genes analyzed for the second plurality of patients from the gene expression level analysis device,
- decomposing, by the residual generation unit of the computing device, the second input matrix into a sum of a product of a plurality of element matrices and a second residual matrix by using a predetermined non-negative matrix factorization (NMF) algorithm;
- and
- training, by the computing device, the classifier by supervised learning by using health condition as learning criteria and inputting the second residual matrix into the input layer of the classifier.
17 more elements
Dependencies are inferred from the claim text.