CN115798596B - Tumor marker identification method based on machine learning - Google Patents
Tumor marker identification method based on machine learning Download PDFInfo
- Publication number
- CN115798596B CN115798596B CN202310059970.8A CN202310059970A CN115798596B CN 115798596 B CN115798596 B CN 115798596B CN 202310059970 A CN202310059970 A CN 202310059970A CN 115798596 B CN115798596 B CN 115798596B
- Authority
- CN
- China
- Prior art keywords
- tumor
- marker
- markers
- association
- identification
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The application discloses a tumor marker identification method based on machine learning, which comprises the following steps: constructing a marker association network in a plurality of tumor markers by utilizing maximum correlation and minimum redundancy analysis based on the dynamic characteristics of the tumor marker content; constructing a correlation marker measuring and calculating model by utilizing a BP neural network based on the correlation degree among the tumor markers so as to realize the quantification of the tumor species correlation; performing parallel tumor seed marking by using tumor markers and related markers; serial tumor species labeling was performed using tumor markers and cognate markers. The tumor type identification method and the tumor type identification device have the advantages that tumor markers and associated markers are used for carrying out parallel tumor type marking so as to achieve tumor type identification comprehensively and improve identification efficiency, and tumor markers and associated markers are used for carrying out serial tumor type marking so as to achieve tumor type identification comprehensively and reduce invalid identification, and the defects of strong randomness and poor pertinence of tumor marker identification are overcome.
Description
Technical Field
The application relates to the technical field of marker identification, in particular to a tumor marker identification method based on machine learning.
Background
Tumor markers are substances that are characteristic of the production of malignant tumor cells, or that are produced by the host's response to a stimulus of a tumor. These substances are present in tissues, body fluids and excretions of patients with tumor errors, and are not contained or contained in very small amounts in normal cells. There are many tumor markers in common use today: AFP (alpha fetoprotein), CEA (carcinoembryonic antigen), PSA (prostate-specific antigen), CA series (CA 199, CA125, CA50, CA 7-24), and the like.
In the current tumor marker identification, the tumor disease probability is judged mostly according to the instantaneous tumor marker content, the randomness of the judgment is strong, and meanwhile, the single tumor marker can only be used for carrying out disease judgment on single tumor, or a plurality of tumor markers are randomly selected to carry out comprehensive identification on a plurality of tumors, the comprehensive difference is caused by single identification, and the multiple redundant identifications are caused by the poor comprehensive identification pertinence, so that the randomness of the tumor marker identification is strong, and the pertinence is poor.
Disclosure of Invention
The application aims to provide a tumor marker identification method based on machine learning, which aims to solve the technical problems that in the prior art, single tumor markers can only be used for carrying out disease judgment on single tumor, or a plurality of tumor markers are selected randomly to carry out comprehensive identification on a plurality of tumors, the comprehensive performance is poor due to the single identification, and multiple redundant identifications are caused due to the poor pertinence of the comprehensive identification, so that the tumor marker identification has strong randomness and poor pertinence.
In order to solve the technical problems, the application specifically provides the following technical scheme:
a tumor marker identification method based on machine learning, comprising the following steps:
s1, selecting a multi-tumor sample, acquiring dynamic characteristics of the contents of a plurality of tumor markers in the multi-tumor sample, and constructing a marker association network in the plurality of tumor markers by utilizing maximum correlation minimum redundancy analysis based on the dynamic characteristics of the contents of the tumor markers, wherein the marking specificity of the plurality of tumor markers to a plurality of tumor seeds in the multi-tumor sample is strong;
s2, extracting the association degree among the tumor markers in a marker association network, and constructing an association marker measuring and calculating model by using a BP neural network based on the association degree among the tumor markers so as to realize the quantification of tumor species association;
step S3, performing parallel tumor seed marking by using tumor markers and related markers, wherein the step comprises the following steps: respectively and synchronously inputting dynamic characteristics of tumor markers and associated marker contents into a pre-established corresponding tumor species disease identification model to determine the disease tumor species so as to realize comprehensive tumor species identification and improve identification efficiency;
serial tumor species labeling with tumor markers and cognate markers, comprising: and inputting the dynamic characteristics of the tumor marker content into a pre-established corresponding tumor disease identification model, and determining whether the corresponding tumor disease identification model of the related marker is identified according to the tumor disease result of the tumor marker so as to realize comprehensive tumor identification and reduce ineffective identification.
As a preferred embodiment of the present application, the obtaining dynamic characteristics of the contents of multiple tumor markers in the multiple tumor samples includes:
setting a plurality of detection time sequences in a time period, and measuring the content of each tumor marker in a multi-tumor sample at each detection time sequence;
and (3) arranging the content of each tumor marker at each detection time sequence according to the time sequence to obtain a content time sequence as the dynamic characteristic of the content of each tumor marker.
As a preferred scheme of the present application, the method for constructing a marker association network from a plurality of tumor markers by using maximum correlation minimum redundancy analysis based on the dynamic characteristics of the tumor marker content comprises the following steps:
performing relevance screening on a plurality of tumor markers by using dynamic characteristics of maximized relevance based on tumor marker content to realize the maximization of the marker relevance in a marker relevance network, wherein the function expression of the maximized relevance is as follows:
；
wherein W is a relevance value, max is a maximize operator, h i ，h j Respectively the firsti，jAnd the dynamic characteristics of the tumor marker content, m is the total category of tumor markers, I (h i , h j ) Is h i And h j Is mutually trusted of (1)The function of the information is that,i，jis a count variable;
redundancy screening is carried out on a plurality of tumor markers by using the dynamic characteristics of minimized redundancy based on tumor marker content so as to realize the minimization of marker association in a marker association network, and the functional expression of minimized redundancy is as follows:
；
wherein V is a redundancy value, min is a minimization operator, h i ，h j Respectively the firsti，jAnd the dynamic characteristics of the tumor marker content, m is the total category of tumor markers, I (h i , h j ) Is h i And h j Is a function of the mutual information of (a),i，jis a count variable;
；
wherein x is t ，y t Respectively is h i ，h j The t component in (2) represents the content of the tumor marker at the t detection time sequence, p (x) t ,y t ) For joint probability, p (x t )、p(y t ) Is the edge probability, t is the counting variable, n is h i ，h j Total number of components in (a);
solving the maximized correlation and minimized redundancy to achieve screening of tumor markers from a plurality of tumor markers for constructing a marker-associated network, and constructing the marker-associated network.
As a preferred scheme of the present application, the construction of the marker-associated network includes:
taking a tumor marker used for constructing a marker-associated network as a network node of the marker-associated network, taking mutual information between any two tumor markers used for constructing the marker-associated network as correlation between any two tumor markers used for constructing the marker-associated network, and taking correlation between any two tumor markers used for constructing the marker-associated network as a connection weight between any two corresponding network nodes;
and carrying out network connection on each network node according to the connection weight to obtain a marker associated network.
As a preferred embodiment of the present application, the extracting the association degree between the tumor markers in the marker association network includes:
and sequentially extracting each tumor marker in the marker association network, extracting a plurality of tumor markers with connection relation with each tumor marker as the association markers of each tumor marker, and extracting the connection weight of each tumor marker and the association markers as the association degree of each tumor marker and the association markers.
As a preferable scheme of the application, the method for constructing the association marker measuring and calculating model by utilizing the BP neural network based on the association degree among the tumor markers comprises the following steps:
taking the tumor marker as an input item of the BP neural network, and taking the association marker of the tumor marker and the association degree of the association marker and the tumor marker as an output item of the BP neural network;
performing convolution training on the input item and the output item by using a BP neural network to obtain the associated marker measuring and calculating model, wherein the model expression of the associated marker measuring and calculating model is as follows:
[re_h,P]=BP(h)；
in the formula, re_h is a correlation marker, P is a correlation degree, h is a tumor marker, and BP is a BP neural network.
As a preferable scheme of the application, the tumor type disease identification model is in one-to-one correspondence with the tumor markers according to the tumor types, and the tumor type disease identification model is obtained by training big data by using a neural network.
As a preferred embodiment of the present application, the determining whether the corresponding tumor type disease identification model of the associated marker identifies according to the tumor type disease result of the tumor marker includes:
if the tumor type disease result corresponding to the tumor marker is diseased, carrying out disease identification on the associated marker by utilizing a corresponding tumor type disease identification model according to the association degree;
if the tumor type disease result corresponding to the tumor marker is non-diseased, the associated marker does not need to be used for disease identification by using a tumor type disease identification model.
As a preferred scheme of the application, the dynamic characteristics of the tumor marker content are normalized.
In a preferred embodiment of the present application, in the step S3, the tumor marker and the associated marker are obtained from an associated marker measurement model.
Compared with the prior art, the application has the following beneficial effects:
according to the application, a marker association network is constructed to represent the association degree among the tumor markers, an association marker measuring and calculating model is constructed by utilizing the BP neural network based on the association degree among the tumor markers so as to realize the quantification of tumor species association, the tumor markers and the association markers are utilized to carry out parallel tumor species marking so as to realize the comprehensive recognition of tumor species and improve the recognition efficiency, and the tumor markers and the association markers are utilized to carry out serial tumor species marking so as to realize the comprehensive recognition of tumor species and reduce the ineffective recognition at the same time, thereby reducing the defects of strong randomness and poor pertinence of tumor marker recognition.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below. It will be apparent to those of ordinary skill in the art that the drawings in the following description are exemplary only and that other implementations can be obtained from the extensions of the drawings provided without inventive effort.
Fig. 1 is a flowchart of a tumor marker identification method according to an embodiment of the present application.
Detailed Description
The following description of the embodiments of the present application will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
As shown in fig. 1, the application provides a tumor marker identification method based on machine learning, which comprises the following steps:
step S1, selecting a multi-tumor sample, obtaining dynamic characteristics of the contents of a plurality of tumor markers in the multi-tumor sample, and constructing a marker association network in the plurality of tumor markers by utilizing maximum correlation minimum redundancy analysis based on the dynamic characteristics of the contents of the tumor markers, wherein the marking specificity of the plurality of tumor markers to a plurality of tumor types in the multi-tumor sample is strong, for example: tumor markers of primary liver cancer in AFP, tumor markers of stomach, colorectal and lung in CEA, CA199 is tumor marker with strong pancreatic cancer specificity, PSA is tumor marker of prostate cancer, etc.;
acquiring dynamic characteristics of multiple tumor marker contents in multiple tumor samples, including:
setting a plurality of detection time sequences in a time period, and measuring the content of each tumor marker in a multi-tumor sample at each detection time sequence;
and (3) arranging the content of each tumor marker at each detection time sequence according to the time sequence to obtain a content time sequence as the dynamic characteristic of the content of each tumor marker.
The probability of random content characteristics of a certain time sequence of the tumor marker exists, so that the probability of tumor disease is judged by utilizing the transient content of the tumor marker, and the reliability is poor.
The single tumor marker identification can only be used for identifying single tumor species, and in a patient with concurrent tumor species, only single tumor species identification is carried out, so that other tumor species are missed to be identified, treatment of the tumor species is ignored, and treatment delay danger is caused, and if all tumor markers are identified for the patient at first, although a plurality of tumor species existing in the patient can be comprehensively identified, the residual tumor markers belong to ineffective identification, medical resources are wasted, medical burden of the patient is increased, and further, the problems of poor pertinence and strong randomness exist in identification of the single tumor markers and identification of the comprehensive tumor markers are solved, so that the relevance analysis is carried out on each tumor marker to analyze the disease relevance of each tumor species, namely the disease of the A tumor species, the disease occurrence of other tumor species is generated, the relevance of the disease occurrence among the tumor species is excavated by utilizing the dynamic characteristics of the tumor markers, the relevance of the tumor markers is marked by constructing a marker relevance network, and the relevance among the tumor species is further represented, and the specific steps are as follows:
constructing a marker association network in a plurality of tumor markers by utilizing maximum correlation and minimum redundancy analysis based on dynamic characteristics of tumor marker content, comprising:
carrying out relevance screening on a plurality of tumor markers by using dynamic characteristics of maximized relevance based on tumor marker content so as to realize the maximization of the marker relevance in a marker relevance network, wherein the function expression of the maximization of the relevance is as follows:
；
wherein W is a relevance value, max is a maximize operator, h i ，h j Respectively the firsti，jAnd the dynamic characteristics of the tumor marker content, m is the total category of tumor markers, I (h i , h j ) Is h i And h j Is a function of the mutual information of (a),i，jis a count variable;
redundancy screening is carried out on a plurality of tumor markers by using the dynamic characteristics of minimized redundancy based on the content of the tumor markers so as to realize the minimization of the association of the markers in a marker association network, and the functional expression for minimizing the redundancy is as follows:
；
wherein V is a redundancy value, min is a minimization operator, h i ，h j Respectively the firsti，jAnd the dynamic characteristics of the tumor marker content, m is the total category of tumor markers, I (h i , h j ) Is h i And h j Is a function of the mutual information of (a),i，jis a count variable;
；
wherein x is t ，y t Respectively is h i ，h j The t component in (2) represents the content of the tumor marker at the t detection time sequence, p (x) t ,y t ) For joint probability, p (x t )、p(y t ) Is the edge probability, t is the counting variable, n is h i ，h j Total number of components in (a);
solving the maximized correlation and minimized redundancy to achieve screening of tumor markers from a plurality of tumor markers for constructing a marker-associated network, and constructing the marker-associated network.
The construction of the marker-associated network comprises the following steps:
taking a tumor marker used for constructing a marker-associated network as a network node of the marker-associated network, taking mutual information between any two tumor markers used for constructing the marker-associated network as correlation between any two tumor markers used for constructing the marker-associated network, and taking correlation between any two tumor markers used for constructing the marker-associated network as a connection weight between any two corresponding network nodes;
and carrying out network connection on each network node according to the connection weight to obtain a marker associated network.
The maximum correlation and the minimum redundancy are utilized to construct a marker correlation network, so that the correlation among tumor markers existing in the marker correlation network is the maximum, the redundancy of the tumor markers contained in the whole correlation network is the minimum, namely the number of irrelevant markers is the minimum, the correlation among various tumor markers can be accurately and simply represented by minimizing the redundancy and maximizing the correlation, and the correlation among various tumor types can be further accurately and simply represented. For example, the association markers of the tumor markers a are B, C and D, and the respective association degrees are 80%,65% and 77%, which indicates that there is an association between the tumor seeds a and B, C and D, so that in order to ensure the comprehensiveness and pertinence of the tumor seed identification, the identification of the association markers B, C and D is required, and whether the tumor seeds B, C and D are ill or not is further determined, so that a marker association network is constructed, a plurality of tumor markers to be identified can be acquired pertinently, and the comprehensiveness of the identification can be improved when the tumor marker/tumor seed identification is performed pertinently.
S2, extracting the association degree among the tumor markers in a marker association network, and constructing an association marker measuring and calculating model by using a BP neural network based on the association degree among the tumor markers so as to realize the quantification of tumor species association;
extracting the association degree among the tumor markers in the marker association network, wherein the method comprises the following steps of:
and sequentially extracting each tumor marker in the marker association network, extracting a plurality of tumor markers with connection relation with each tumor marker as the association markers of each tumor marker, and extracting the connection weight of each tumor marker and the association markers as the association degree of each tumor marker and the association markers.
Constructing a correlation marker measuring model by using a BP neural network based on the correlation degree among the tumor markers, wherein the method comprises the following steps:
taking the tumor marker as an input item of the BP neural network, and taking the association marker of the tumor marker and the association degree of the association marker and the tumor marker as an output item of the BP neural network;
performing convolution training on the input item and the output item by using the BP neural network to obtain a related marker measuring and calculating model, wherein the model expression of the related marker measuring and calculating model is as follows:
[re_h,P]=BP(h)；
in the formula, re_h is a correlation marker, P is a correlation degree, h is a tumor marker, and BP is a BP neural network.
By constructing the associated marker measuring and calculating model, the associated markers of the tumor markers can be obtained according to the tumor markers, and then the comprehensively aimed marker identification is carried out, the possibility of tumor occurrence is judged, the modeling identification is carried out, the identification efficiency is high, and the accuracy is high.
Step S3, performing parallel tumor seed marking by using tumor markers and related markers, wherein the step comprises the following steps: respectively and synchronously inputting dynamic characteristics of tumor markers and associated marker contents into a pre-established corresponding tumor species disease identification model to determine the disease tumor species so as to realize comprehensive tumor species identification and improve identification efficiency;
serial tumor species labeling with tumor markers and cognate markers, comprising: and inputting the dynamic characteristics of the tumor marker content into a pre-established corresponding tumor disease identification model, and determining whether the corresponding tumor disease identification model of the related marker is identified according to the tumor disease result of the tumor marker so as to realize comprehensive tumor identification and reduce ineffective identification.
The parallel tumor type mark can be suitable for tumor type recognition of a patient with a determined multiple tumor type and coexistence, can realize comprehensive recognition, improve recognition efficiency, and can be suitable for tumor type recognition of a patient with a determined multiple tumor type and coexistence, and can reduce ineffective recognition caused by direct parallel tumor type mark while realizing comprehensive recognition.
The tumor type disease identification model is obtained by training big data by utilizing a neural network according to one-to-one correspondence of tumor types and tumor markers.
Determining whether the corresponding tumor type disease identification model of the associated marker is identified according to the tumor type disease result of the tumor marker, comprising the following steps:
if the tumor type disease result corresponding to the tumor marker is diseased, carrying out disease identification on the associated marker by utilizing a corresponding tumor type disease identification model according to the association degree;
if the tumor type disease result corresponding to the tumor marker is non-diseased, the associated marker does not need to be used for disease identification by using a tumor type disease identification model.
And carrying out normalization treatment on the dynamic characteristics of the tumor marker content.
In step S3, tumor markers and associated markers are obtained from an associated marker measurement model.
According to the application, a marker association network is constructed to represent the association degree among the tumor markers, an association marker measuring and calculating model is constructed by utilizing the BP neural network based on the association degree among the tumor markers so as to realize the quantification of tumor species association, the tumor markers and the association markers are utilized to carry out parallel tumor species marking so as to realize the comprehensive recognition of tumor species and improve the recognition efficiency, and the tumor markers and the association markers are utilized to carry out serial tumor species marking so as to realize the comprehensive recognition of tumor species and reduce the ineffective recognition at the same time, thereby reducing the defects of strong randomness and poor pertinence of tumor marker recognition.
The above embodiments are only exemplary embodiments of the present application and are not intended to limit the present application, the scope of which is defined by the claims. Various modifications and equivalent arrangements of this application will occur to those skilled in the art, and are intended to be within the spirit and scope of the application.
Claims (6)
1. The tumor marker identification method based on machine learning is characterized by comprising the following steps of:
s1, selecting a multi-tumor sample, acquiring dynamic characteristics of the contents of a plurality of tumor markers in the multi-tumor sample, and constructing a marker association network in the plurality of tumor markers by utilizing maximum correlation minimum redundancy analysis based on the dynamic characteristics of the contents of the tumor markers, wherein the marking specificity of the plurality of tumor markers to a plurality of tumor seeds in the multi-tumor sample is strong;
s2, extracting the association degree among the tumor markers in a marker association network, and constructing an association marker measuring and calculating model by using a BP neural network based on the association degree among the tumor markers so as to realize the quantification of tumor species association;
step S3, performing parallel tumor seed marking by using tumor markers and related markers, wherein the step comprises the following steps: respectively and synchronously inputting dynamic characteristics of tumor markers and associated marker contents into a pre-established corresponding tumor species disease identification model to determine the disease tumor species so as to realize comprehensive tumor species identification and improve identification efficiency;
serial tumor species labeling with tumor markers and cognate markers, comprising: inputting dynamic characteristics of tumor marker content into a pre-established corresponding tumor disease identification model, and determining whether the corresponding tumor disease identification model of the related marker is identified according to a tumor disease result of the tumor marker so as to realize comprehensive tumor identification and reduce ineffective identification;
the parallel tumor type mark is used for identifying tumor types of the patient with the determined multiple tumor types, so that the comprehensive identification of the multiple tumor types is realized and the identification efficiency is improved; the serial tumor type mark is used for identifying tumor types of patients with multiple tumor types at the same time so as to realize comprehensive identification and reduce ineffective identification caused by direct parallel tumor type marks;
constructing a marker association network in a plurality of tumor markers by utilizing maximum correlation and minimum redundancy analysis based on dynamic characteristics of tumor marker content, comprising:
carrying out relevance screening on a plurality of tumor markers by using dynamic characteristics of maximized relevance based on tumor marker content so as to realize the maximization of the marker relevance in a marker relevance network, wherein the function expression of the maximization of the relevance is as follows:
；
wherein W is a relevance value, max is a maximize operator, h i ，h j Respectively the firsti，jAnd the dynamic characteristics of the tumor marker content, m is the total category of tumor markers, I (h i , h j ) Is h i And h j Is a function of the mutual information of (a),i，jis a count variable;
redundancy screening is carried out on a plurality of tumor markers by using the dynamic characteristics of minimized redundancy based on the content of the tumor markers so as to realize the minimization of the association of the markers in a marker association network, and the functional expression for minimizing the redundancy is as follows:
；
wherein V is a redundancy value, min is a minimization operator, h i ，h j Respectively the firsti，jAnd the dynamic characteristics of the tumor marker content, m is the total category of tumor markers, I (h i , h j ) Is h i And h j Is a function of the mutual information of (a),i，jis a count variable;
；
wherein x is t ，y t Respectively is h i ，h j The t component in (2) represents the content of the tumor marker at the t detection time sequence, p (x) t ,y t ) For joint probability, p (x t )、p(y t ) The edge probability is given, and t is a counting variable;
solving the maximized correlation and minimized redundancy to screen out tumor markers used for constructing a marker-associated network from a plurality of tumor markers, and constructing the marker-associated network;
the construction of the marker-associated network comprises the following steps:
taking a tumor marker used for constructing a marker-associated network as a network node of the marker-associated network, taking mutual information between any two tumor markers used for constructing the marker-associated network as correlation between any two tumor markers used for constructing the marker-associated network, and taking correlation between any two tumor markers used for constructing the marker-associated network as a connection weight between any two corresponding network nodes;
network connection is carried out on each network node according to the connection weight to obtain a marker associated network;
determining whether the corresponding tumor type disease identification model of the associated marker is identified according to the tumor type disease result of the tumor marker, comprising the following steps:
if the tumor type disease result corresponding to the tumor marker is diseased, carrying out disease identification on the associated marker by utilizing a corresponding tumor type disease identification model according to the association degree;
if the tumor type disease result corresponding to the tumor marker is non-diseased, the associated marker does not need to be subjected to disease recognition by using a corresponding tumor type disease recognition model;
in step S3, tumor markers and associated markers are obtained from an associated marker measurement model.
2. The machine learning based tumor marker identification method of claim 1, wherein: the method for obtaining dynamic characteristics of the contents of various tumor markers in various tumor samples comprises the following steps:
setting a plurality of detection time sequences in a time period, and measuring the content of each tumor marker in a multi-tumor sample at each detection time sequence;
and (3) arranging the content of each tumor marker at each detection time sequence according to the time sequence to obtain a content time sequence as the dynamic characteristic of the content of each tumor marker.
3. The machine learning based tumor marker identification method of claim 2, wherein: the extracting the association degree among the tumor markers in the marker association network comprises the following steps:
and sequentially extracting each tumor marker in the marker association network, extracting a plurality of tumor markers with connection relation with each tumor marker as the association markers of each tumor marker, and extracting the connection weight of each tumor marker and the association markers as the association degree of each tumor marker and the association markers.
4. A machine learning based tumor marker identification method according to claim 3, characterized in that: the method for constructing the association marker measuring and calculating model based on the association degree among the tumor markers by using the BP neural network comprises the following steps:
taking the tumor marker as an input item of the BP neural network, and taking the association marker of the tumor marker and the association degree of the association marker and the tumor marker as an output item of the BP neural network;
performing convolution training on the input item and the output item by using a BP neural network to obtain the associated marker measuring and calculating model, wherein the model expression of the associated marker measuring and calculating model is as follows:
[re_h,P]=BP(h)；
in the formula, re_h is a correlation marker, P is a correlation degree, h is a tumor marker, and BP is a BP neural network.
5. The machine learning-based tumor marker identification method according to claim 4, wherein the tumor type disease identification model is obtained by training big data by using a neural network according to one-to-one correspondence of tumor types and tumor markers.
6. The machine learning-based tumor marker identification method according to claim 5, wherein the dynamic characteristics of the tumor marker content are normalized.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310059970.8A CN115798596B (en) | 2023-01-18 | 2023-01-18 | Tumor marker identification method based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310059970.8A CN115798596B (en) | 2023-01-18 | 2023-01-18 | Tumor marker identification method based on machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115798596A CN115798596A (en) | 2023-03-14 |
| CN115798596B true CN115798596B (en) | 2023-10-13 |
Family
ID=85429756
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310059970.8A Active CN115798596B (en) | 2023-01-18 | 2023-01-18 | Tumor marker identification method based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115798596B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| AU2001268602A1 (en) * | 2000-06-19 | 2002-03-21 | Halliburton Energy Services, Inc. | Apparatus and methods for applying time lapse VSP to monitor a reservoir |
| JP2010220621A (en) * | 2002-09-06 | 2010-10-07 | Agensys Inc | Nucleic acid entitled as 98p4b6 and corresponding protein useful in treatment and detection of cancer |
| CN106407742A (en) * | 2016-08-26 | 2017-02-15 | 赵毅 | Method for screening tumor protein markers on basis of multilayer complex network |
| CN106706912A (en) * | 2015-07-21 | 2017-05-24 | 中国科学院上海生命科学研究院 | Marker for diagnosis of inflammation-associated HCC and application thereof |
| CN107058574A (en) * | 2017-05-26 | 2017-08-18 | 郴州市第人民医院 | A kind of related tumor markers of nasopharyngeal carcinoma and application |
| CN107282145A (en) * | 2016-04-01 | 2017-10-24 | 葛宇杰 | The female tumor mark joint inspection chip apparatus of many type of drive coupling runnings |
| TW201804348A (en) * | 2016-07-29 | 2018-02-01 | 長庚醫療財團法人林口長庚紀念醫院 | Method for analyzing cancer detection result by establishing cancer prediction model and combining tumor marker kits analyzing the cancer detection result by using the established cancer prediction model and combining the detection results of the tumor marker kits |
| CN110751203A (en) * | 2019-10-16 | 2020-02-04 | 山东浪潮人工智能研究院有限公司 | Feature extraction method and system based on deep marker learning |
| CN112396616A (en) * | 2020-12-14 | 2021-02-23 | 南京信息工程大学 | Osteosarcoma recurrence risk prediction model based on tissue morphology analysis |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| AU2001268602B2 (en) * | 2000-06-19 | 2006-04-06 | Halliburton Energy Services, Inc. | Apparatus and methods for applying time lapse vsp to monitor a reservoir |
-
2023
- 2023-01-18 CN CN202310059970.8A patent/CN115798596B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| AU2001268602A1 (en) * | 2000-06-19 | 2002-03-21 | Halliburton Energy Services, Inc. | Apparatus and methods for applying time lapse VSP to monitor a reservoir |
| JP2010220621A (en) * | 2002-09-06 | 2010-10-07 | Agensys Inc | Nucleic acid entitled as 98p4b6 and corresponding protein useful in treatment and detection of cancer |
| CN106706912A (en) * | 2015-07-21 | 2017-05-24 | 中国科学院上海生命科学研究院 | Marker for diagnosis of inflammation-associated HCC and application thereof |
| CN107282145A (en) * | 2016-04-01 | 2017-10-24 | 葛宇杰 | The female tumor mark joint inspection chip apparatus of many type of drive coupling runnings |
| TW201804348A (en) * | 2016-07-29 | 2018-02-01 | 長庚醫療財團法人林口長庚紀念醫院 | Method for analyzing cancer detection result by establishing cancer prediction model and combining tumor marker kits analyzing the cancer detection result by using the established cancer prediction model and combining the detection results of the tumor marker kits |
| CN106407742A (en) * | 2016-08-26 | 2017-02-15 | 赵毅 | Method for screening tumor protein markers on basis of multilayer complex network |
| CN107058574A (en) * | 2017-05-26 | 2017-08-18 | 郴州市第人民医院 | A kind of related tumor markers of nasopharyngeal carcinoma and application |
| CN110751203A (en) * | 2019-10-16 | 2020-02-04 | 山东浪潮人工智能研究院有限公司 | Feature extraction method and system based on deep marker learning |
| CN112396616A (en) * | 2020-12-14 | 2021-02-23 | 南京信息工程大学 | Osteosarcoma recurrence risk prediction model based on tissue morphology analysis |
Non-Patent Citations (3)
| Title |
|---|
| Chi Yuan 等.TMDFM: A Data Fusion Model for Combined Detection of Tumor Markers.《IEEE》.2015,全文. * |
| 杨馨悦 等.生物信息学分析揭示主动脉瘤潜在的诊断标志物和 发病机制 .《岭南心血管病杂志》.2022,全文. * |
| 葛菁茹 等.以肺转移为首发表现的结肠癌预后分析.《医学信息》.2021,全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115798596A (en) | 2023-03-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Jutric et al. | Impact of lymph node status in patients with intrahepatic cholangiocarcinoma treated by major hepatectomy: a review of the National Cancer Database | |
| CN109906276A (en) | For detecting the recognition methods of somatic mutation feature in early-stage cancer | |
| CN1484806A (en) | A process for discriminating between biological states based on hidden patterns from | |
| CN109124660B (en) | Gastrointestinal stromal tumor postoperative risk detection method and system based on deep learning | |
| US20120191357A1 (en) | Discovering Progression and Differentiation Hierarchy From Multidimensional Data | |
| CN105574365B (en) | The statistics verification method of high-flux sequence abrupt climatic change result | |
| CN109528230B (en) | Method and device for segmenting breast tumor based on multistage transformation network | |
| CN105132407A (en) | Method for low-frequency mutant-enriched sequencing of DNA of exfoliative cells | |
| Bilal et al. | Novel deep learning algorithm predicts the status of molecular pathways and key mutations in colorectal cancer from routine histology images | |
| CN111028945B (en) | Classification prediction method and device based on data fusion and storage medium | |
| WO2021143422A1 (en) | Grain sampling method, readable storage medium and system | |
| CN109337957A (en) | The method for detecting genome multimutation type | |
| CN115375640A (en) | Tumor heterogeneity identification method and device, electronic equipment and storage medium | |
| Raajan et al. | Non-invasive technique-based novel corona (COVID-19) virus detection using CNN | |
| CN110767312A (en) | Artificial intelligence auxiliary pathological diagnosis system and method | |
| Feragen et al. | A hierarchical scheme for geodesic anatomical labeling of airway trees | |
| Borkowski et al. | Comparing artificial intelligence platforms for histopathologic cancer diagnosis | |
| CN106460045A (en) | Use of recurrent copy number variations in constitutional human genome for prediction of predisposition to cancer | |
| CN113903401A (en) | ctDNA length-based analysis method and system | |
| Li et al. | Artificial intelligence-assisted decision making for prognosis and drug efficacy prediction in lung cancer patients: A narrative review | |
| CN110189824A (en) | Prognosis situation group technology, the device and system of primary carcinoma of liver radical excision | |
| CN113380396A (en) | Method for evaluating risks of multiple intestinal diseases based on fecal microbial markers and human DNA content and application | |
| CN115798596B (en) | Tumor marker identification method based on machine learning | |
| CN113033667A (en) | Ultrasound image two-stage deep learning breast tumor classification method and device | |
| CN107850526A (en) | Assess the method for cell breast samples and the composition for putting into practice methods described |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |