CN119312114B - A method for predicting abnormalities during the hypoxia expansion phase of stem cells and related equipment - Google Patents
A method for predicting abnormalities during the hypoxia expansion phase of stem cells and related equipment Download PDFInfo
- Publication number
- CN119312114B CN119312114B CN202411852226.4A CN202411852226A CN119312114B CN 119312114 B CN119312114 B CN 119312114B CN 202411852226 A CN202411852226 A CN 202411852226A CN 119312114 B CN119312114 B CN 119312114B
- Authority
- CN
- China
- Prior art keywords
- variable data
- time
- history
- historical
- data
- 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
- 238000000034 method Methods 0.000 title claims abstract description 50
- 210000000130 stem cell Anatomy 0.000 title claims abstract description 49
- 206010021143 Hypoxia Diseases 0.000 title claims abstract description 28
- 230000005856 abnormality Effects 0.000 title claims abstract description 27
- 230000007954 hypoxia Effects 0.000 title claims abstract description 23
- 238000003199 nucleic acid amplification method Methods 0.000 claims abstract description 22
- 230000003321 amplification Effects 0.000 claims abstract description 21
- 230000004931 aggregating effect Effects 0.000 claims abstract description 9
- 238000004220 aggregation Methods 0.000 claims description 53
- 230000002776 aggregation Effects 0.000 claims description 49
- 239000011159 matrix material Substances 0.000 claims description 28
- 230000006870 function Effects 0.000 claims description 25
- 238000004590 computer program Methods 0.000 claims description 19
- 230000004913 activation Effects 0.000 claims description 10
- 230000010355 oscillation Effects 0.000 claims description 3
- 230000002159 abnormal effect Effects 0.000 abstract description 37
- 238000002360 preparation method Methods 0.000 abstract description 6
- 230000008569 process Effects 0.000 description 11
- 210000004027 cell Anatomy 0.000 description 9
- 238000003860 storage Methods 0.000 description 8
- CURLTUGMZLYLDI-UHFFFAOYSA-N Carbon dioxide Chemical compound O=C=O CURLTUGMZLYLDI-UHFFFAOYSA-N 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 6
- 229910052760 oxygen Inorganic materials 0.000 description 6
- 239000001301 oxygen Substances 0.000 description 6
- 238000001514 detection method Methods 0.000 description 5
- 230000004044 response Effects 0.000 description 4
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 3
- 229910002092 carbon dioxide Inorganic materials 0.000 description 3
- 239000001569 carbon dioxide Substances 0.000 description 3
- 238000013500 data storage Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000012549 training Methods 0.000 description 3
- 230000006978 adaptation Effects 0.000 description 2
- 230000010261 cell growth Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000007789 gas Substances 0.000 description 2
- 230000012010 growth Effects 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 230000002503 metabolic effect Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000008447 perception Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 101100272279 Beauveria bassiana Beas gene Proteins 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000004071 biological effect Effects 0.000 description 1
- 238000004113 cell culture Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000001973 epigenetic effect Effects 0.000 description 1
- 239000012530 fluid Substances 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000002906 microbiologic effect Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 230000003938 response to stress Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000004062 sedimentation Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000003756 stirring Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application relates to the technical field of stem cell preparation, and provides an abnormality prediction method and related equipment for a stem cell hypoxia amplification stage. The method comprises the steps of calculating data characteristics of each variable data, calculating global characteristics of each historical moment based on all the data characteristics, calculating similarity relations between every two variable data of the historical moment based on the global characteristics, building a directed graph according to all the similarity relations, calculating correlation attenuation rates between every two nodes in the directed graph according to all the historical moments, acquiring final characteristics of each variable data based on all the correlation attenuation rates, aggregating the final characteristics of all the variable data corresponding to the modes at each historical moment to obtain aggregated characteristics of the modes at each historical moment, updating the aggregated characteristics to obtain final aggregated characteristics, and carrying out abnormal prediction based on the final aggregated characteristics to obtain abnormal prediction results. The method can improve the accuracy of anomaly prediction.
Description
Technical Field
The application relates to the technical field of stem cell preparation, in particular to an abnormality prediction method and related equipment for a stem cell hypoxia amplification stage.
Background
The stem cell preparation is a highly refined work, and is characterized in that the key steps of raw material collection, passage separation, process parameter control and the like are performed. In the current process flow, the conditions affecting cell quality and amplification efficiency can be regarded as abnormal conditions, and the changes of the observable variables of the environmental parameters and biological properties can well reflect the abnormality. The hypoxia amplification stage involves continuous and fine multi-temporal multi-modal variables, the variation of a single variable can be monitored in time by various methods, but it cannot be used as a signal of abnormality generation. The reason is that abnormalities in a single variable do not directly affect cell growth, but rather occur by affecting other related variables to produce a chain reaction.
There are differences among stem cell individuals, which are derived from genetic diversity, epigenetic differences, cell metabolic states, etc., meaning that different stem cells have different demands on growth environments and the differences are difficult to determine. In addition, stem cells are very sensitive to the perception of the culture environment, and even small fluctuations may trigger the stress response of the cells, which makes professionals face a serious problem in cell culture. First, the adjustment of the process parameters according to a predetermined schedule does not completely ensure efficient expansion of cells, and may even lead to abnormal cell quality. Secondly, the control of relevant parameters of the microcarrier can lead to abnormal generation, a plurality of scale bioreactors can be used along with the increase of passage times, and in different bioreactors, different stirring speeds or fluid dynamic conditions can lead to the aggregation or sedimentation of the microcarrier, and can lead to the micro-change of local environment and influence related variables. Thirdly, the stem cell process flow can be applied to the intelligent tool in future scenes, the long-term non-fixed program operation of the stem cell process flow can have a perception error, so that a certain deviation occurs between the actual operation time and the planned operation time, meanwhile, the robot can be excessively adapted to specific conditions, and when the process requirement changes, the performance of the stem cell process flow can be poor or even abnormal. The problems are analyzed by capturing the characteristics of the time series data such as environment and biological attributes, the complexity and dimension of the multi-time series multi-mode variable are increased along with the intelligent degree of the process flow, the cost for automatically monitoring the data is higher and higher, and the time delay of the system is also changed. In addition, in view of the large cell size and excessive passage cost in the expansion stage, although occurrence of abnormality is a small probability event, it is still indispensable to perform abnormality prediction of the multiplex time series data in this scenario.
To our knowledge, there is no abnormal prediction of the hypoxia expansion phase of stem cells. In the conventional time series data anomaly detection algorithm, most of research works are modeled by capturing the spatial and time dependence between plant data streams, and a method based on a graph neural network is widely adopted on the basis of the modeling. Part of the study modeling dependencies based on distance or similarity methods, one significant drawback is dependency symmetry. The characteristics of direct or indirect unidirectional dependence, dynamic dependence among variables and the like are not comprehensively considered, and the stability is lacking. In the environment of stem cell preparation, there are various kinds of observation variables of microbiological safety, biological safety, cell biology, etc. which can intuitively reflect the growth state of stem cells. There is a highly complex dependency between these variables, e.g. a certain concentration of carbon dioxideCan react with water to generate carbonic acid, thereby influencingValue of dissolved oxygenThe level of (2) affects the metabolic activity of the cell, which will produceAnd water, the result is adversely affectedLevel and level ofValues. The abnormality is obviously determined by considering only the variation of each univariate, and thus, when the environment of the stem cell hypoxia amplification stage is predicted abnormally, the accuracy of the abnormality prediction is low.
Disclosure of Invention
The application provides an abnormality prediction method and related equipment for a stem cell hypoxia amplification stage, which can solve the problem of low accuracy of abnormality prediction.
In a first aspect, embodiments of the present application provide an abnormality prediction method for a stem cell hypoxia amplification stage, the abnormality prediction method comprising:
Acquiring a plurality of variable data of the environment where the target stem cells are located at a plurality of historical moments in a hypoxia amplification stage, wherein the plurality of variable data correspond to a plurality of modes, and each mode at least corresponds to one variable data;
Calculating the data characteristics of each variable data at each historical moment, and calculating the global characteristics of each historical moment based on all the data characteristics;
Calculating the similarity relation between every two variable data at each historical moment based on the global feature at each historical moment, and constructing a directed graph according to all the similarity relations;
Calculating the correlation attenuation rate between every two nodes in the directed graph according to all the historical moments, and acquiring the final characteristics of each variable data based on all the correlation attenuation rates;
Respectively aiming at each mode, aggregating the final characteristics of all variable data corresponding to the mode at each historical moment to obtain the aggregated characteristics of the mode at each historical moment;
updating all the aggregation features to obtain a plurality of final aggregation features, and carrying out exception prediction based on all the final aggregation features to obtain an exception prediction result, wherein the exception prediction result is used for describing the exception condition of a plurality of variable data of the environment in a future time period.
Optionally, calculating the data characteristic of each variable data at each historical moment includes:
by the formula:
Calculate the first The first time of historyData characteristics of individual variable data;
Wherein, Which is indicative of the characteristics of the sensor,,,The convolution is represented by a representation of the convolution,Represent the firstA matrix of variable data for each historical moment,Represent the firstThe 1 st variable data of the historical moment,Represent the firstThe 2 nd variable data of the historical moment,Represent the firstThe first time of historyThe data of the individual variables are stored,,The number of variable data representing each historical moment,,The number of times of the history is indicated,Represent the firstThe first time of historyPosition coding of individual variable data:
wherein, Represents an index of an integer number,Representing the oscillation frequency.
Optionally, calculating the global feature for each historical moment based on all the data features includes:
by the formula:
Calculate the first Global features for individual historical moments;
Wherein, Represent the firstThe first time of historyThe degree of information center of the individual variable data,Represent the firstThe first time of historyThe data characteristics of the individual variable data,Represent the firstThe first time of historyVariable data and the firstThe first time of historyThe degree of information sharing between the individual variable data,Representation ofIs used as a reference to the entropy of (a),Representation ofIs used as a reference to the entropy of (a),Representation ofAnd (3) withThe information center degree is the sum of the information sharing degree between the variable data and each other variable data at the historical moment.
Optionally, calculating the similarity relationship between each two variable data at each historical moment based on the global feature at each historical moment includes:
For each history time, the following steps are performed:
Calculating the comprehensive similarity between every two variable data at the historical moment based on the global features at the historical moment;
And acquiring an initial similarity relation according to the comprehensive similarity between every two variable data, and updating the initial similarity relation to obtain the similarity relation between every two variable data.
Optionally, calculating the comprehensive similarity between each two variable data at the historical moment based on the global feature at the historical moment includes:
by the formula:
Calculate the first The first time of historyVariable data and the firstThe first time of historyComprehensive similarity between individual variable data;
Wherein, 、AndAll of which represent the saturation rate and,Represent the firstVariable data and the firstThe degree of information sharing between the individual variable data,Represent the firstVariable data and the firstThe degree of information sharing between global features at each historical moment,Represent the firstVariable data and the firstCosine similarity between the individual variable data,Represent the firstThe first time of historyThe data characteristics of the individual variable data,Represent the firstThe first time of historyThe data characteristics of the individual variable data,,,The number of variable data representing each historical moment,,Representing the number of historical moments;
obtaining an initial similarity relation according to the comprehensive similarity between every two variable data, wherein the method comprises the following steps:
by the formula:
Acquisition of the first Variable data and the firstInitial similarity relationship between individual variable data;
Wherein, Represents the value of the function of the index,Represent the firstVariable data and the firstNormalized similarity between the individual variable data,Representing a similarity threshold;
Updating the initial similarity relationship to obtain the similarity relationship between every two variable data, wherein the updating comprises the following steps:
by the formula:
For initial similarity relationship Updating to obtain the firstVariable data and the firstSimilarity relationship between individual variable data;
Wherein, Representing the first priori knowledge in a directed graphVariable data and the firstThe adjacency between the individual variable data,Representing an activation function.
Optionally, calculating the correlation attenuation rate between every two nodes in the directed graph according to all the historical moments includes:
for each history time, the correlation decay rate between the node at the history time and the node at each other history time is defined as ;The value of (c) is the difference between the historical time and the other historical time minus one,Is a constant;
obtaining final characteristics of each variable data based on all the correlation decay rates, including:
by the formula:
Calculate the first The first time of historyFinal characteristics of individual variable data;
Wherein, Representing the parameters that can be learned,Represent the firstThe first time of historyVariable data and the firstThe first time of historyThe comprehensive adjacency between the individual variable data,Represent the firstThe first time of historyThe data characteristics of the individual variable data,The number of times of the history is indicated,The number of variable data representing each historical moment,Represent the firstThe first time of historyVariable data and the firstThe first time of historyAdjacency between variable data ifThen,Represent the firstThe first time of historyVariable data and the firstThe first time of historyComprehensive similarity between variable data, ifThen,Represent the firstNode and No. of historical timeAnd the comprehensive adjacency relation is the adjacency relation between two variable data after the introduction of the correlation attenuation rate for adjustment.
Optionally, aggregating final features of all variable data corresponding to the mode at each historical moment to obtain an aggregate feature of the mode at each historical moment, including:
by the formula:
Calculate the first The first mode is atAggregation features for individual historic moments;
Wherein, Represent the firstThe first mode is atA set of all variable data corresponding to each historical moment,Represent the firstThe first time of historyInformation-outputting centrality of the individual variable data.
Optionally, updating all the aggregation features to obtain a plurality of final aggregation features, including:
calculating the association degree between every two aggregation features;
and updating all the aggregation features according to all the association degrees to obtain a plurality of final aggregation features.
Optionally, calculating the degree of association between each two aggregation features includes:
by the formula:
Calculate the first Modality and the firstBetween the first modesDegree of correlation of individual historic moments;
Wherein, Represent the firstThe first mode is atA set of all variable data corresponding to each historical moment,Represent the firstThe first time of historyVariable data and the firstThe first time of historyThe association relationship between the individual variable data,,Represent the firstThe first time of historyVariable data and the firstThe first time of historyComprehensive similarity between the individual variable data;
updating all the aggregation features according to all the association degrees to obtain a plurality of final aggregation features, including:
by the formula:
updating all the aggregation characteristics;
wherein, Represent the firstThe feature matrix is aggregated in the middle of the layer,The degree of association matrix is represented by a matrix of degrees of association,Represent the firstThe matrix of learnable weights of the layer,Represent the firstThe feature matrix is aggregated in the middle of the layer,The activation function is represented as a function of the activation,,Representing the update of the last layer whenIn the time-course of which the first and second contact surfaces,Representing an aggregate feature matrix whenIn the time-course of which the first and second contact surfaces,Representing the final aggregate feature matrix.
In a second aspect, an embodiment of the present application provides an abnormality prediction apparatus for a stem cell hypoxia expansion stage, including:
the system comprises an acquisition module, a control module and a control module, wherein the acquisition module acquires a plurality of variable data of an environment in which a target stem cell is positioned at a plurality of historical moments in a hypoxia amplification stage, the variable data correspond to a plurality of modes, and each mode at least corresponds to one variable data;
The first calculation module is used for calculating the data characteristics of each variable data at each historical moment and calculating the global characteristics of each historical moment based on all the data characteristics;
The system comprises a building module, a directed graph, a plurality of nodes, a plurality of variable data storage modules, a plurality of data storage modules and a plurality of data storage modules, wherein the building module is used for calculating the similarity relation between every two variable data at each historical moment based on the global characteristic at each historical moment and building the directed graph according to all the similarity relations;
the second calculation module calculates the correlation attenuation rate between every two nodes in the directed graph according to all the historical moments and acquires the final characteristics of each variable data based on all the correlation attenuation rates;
the aggregation module is used for respectively aggregating the final characteristics of all variable data corresponding to the modes at each historical moment aiming at each mode to obtain the aggregation characteristics of the modes at each historical moment;
The anomaly prediction module is used for updating all the aggregation characteristics to obtain a plurality of final aggregation characteristics, and performing anomaly prediction based on all the final aggregation characteristics to obtain an anomaly prediction result, wherein the anomaly prediction result is used for describing the anomaly condition of a plurality of variable data of the environment in a future time period.
In a third aspect, an embodiment of the present application provides a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, where the processor implements the method for predicting an abnormality in a hypoxia-expansion phase of stem cells described above when the processor executes the computer program.
In a fourth aspect, embodiments of the present application provide a computer readable storage medium storing a computer program which, when executed by a processor, implements the above-described method for predicting abnormalities in a stem cell hypoxia-amplification stage.
The scheme of the application has the following beneficial effects:
in the embodiment of the application, the data characteristics of each variable data at each historical moment are calculated, the global characteristics of each historical moment are calculated based on all the data characteristics, the similarity relation between every two variable data at each historical moment is calculated based on the global characteristics of each historical moment, a directed graph is constructed according to all the similarity relations, the correlation attenuation rate between every two nodes in the directed graph is calculated according to all the historical moments, the final characteristics of each variable data are obtained based on all the correlation attenuation rates, the final characteristics of all the variable data corresponding to each mode are respectively aggregated for each mode, the aggregated characteristics of the mode at each historical moment are obtained, finally all the aggregated characteristics are updated, a plurality of final aggregated characteristics are obtained, and abnormal prediction is carried out based on all the final aggregated characteristics, so that an abnormal prediction result is obtained. The method comprises the steps of calculating global features based on data features of a plurality of variable data, comprehensively analyzing a plurality of variables of an environment where target stem cells are located, improving information richness of the global features, obtaining similarity relations between the variable data according to the global features, improving accuracy and global performance of the similarity relations, calculating final features of the variable data by using a correlation attenuation rate, considering correlation between historical moments, improving accuracy of the final features, further improving accuracy of final aggregated features of modes obtained according to final feature aggregation, carrying out anomaly prediction according to the accurate final aggregated features, and effectively improving accuracy of anomaly prediction.
Other advantageous effects of the present application will be described in detail in the detailed description section which follows.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the embodiments or the description of the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flowchart of a method for predicting abnormalities in a stem cell hypoxia-expansion phase according to an embodiment of the present application;
FIG. 2 is a schematic diagram of an abnormality prediction apparatus in a stem cell hypoxia amplification stage according to an embodiment of the present application;
fig. 3 is a schematic structural diagram of a terminal device according to an embodiment of the present application.
Detailed Description
In the following description, for purposes of explanation and not limitation, specific details are set forth such as the particular system architecture, techniques, etc., in order to provide a thorough understanding of the embodiments of the present application. It will be apparent, however, to one skilled in the art that the present application may be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present application with unnecessary detail.
It should be understood that the terms "comprises" and/or "comprising," when used in this specification and the appended claims, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.
It should also be understood that the term "and/or" as used in the present specification and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes such combinations.
As used in the present description and the appended claims, the term "if" may be interpreted as "when..once" or "in response to a determination" or "in response to detection" depending on the context. Similarly, the phrase "if a determination" or "if a [ described condition or event ] is detected" may be interpreted in the context of meaning "upon determination" or "in response to determination" or "upon detection of a [ described condition or event ]" or "in response to detection of a [ described condition or event ]".
Furthermore, the terms "first," "second," "third," and the like in the description of the present specification and in the appended claims, are used for distinguishing between descriptions and not necessarily for indicating or implying a relative importance.
Reference in the specification to "one embodiment" or "some embodiments" or the like means that a particular feature, structure, or characteristic described in connection with the embodiment is included in one or more embodiments of the application. Thus, appearances of the phrases "in one embodiment," "in some embodiments," "in other embodiments," and the like in the specification are not necessarily all referring to the same embodiment, but mean "one or more but not all embodiments" unless expressly specified otherwise. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless expressly specified otherwise.
Aiming at the problem of low accuracy of the existing abnormal prediction, the embodiment of the application provides an abnormal prediction method of a stem cell hypoxia amplification stage, which comprises the steps of calculating the data characteristics of each variable data at each historical moment, calculating the global characteristics of each historical moment based on all the data characteristics, calculating the similarity relation between every two variable data at each historical moment based on the global characteristics of each historical moment, constructing a directed graph according to all the similarity relations, calculating the correlation attenuation rate between every two nodes in the directed graph according to all the historical moment, acquiring the final characteristics of each variable data based on all the correlation attenuation rates, respectively aggregating the final characteristics of all the variable data corresponding to the modes at each historical moment for each mode, obtaining the aggregate characteristics of the modes at each historical moment, updating all the aggregate characteristics, obtaining a plurality of final aggregate characteristics, carrying out abnormal prediction based on all the final aggregate characteristics, and obtaining an abnormal prediction result. The method comprises the steps of calculating global features based on data features of a plurality of variable data, comprehensively analyzing a plurality of variables of an environment where target stem cells are located, improving information richness of the global features, obtaining similarity relations between the variable data according to the global features, improving accuracy and global performance of the similarity relations, calculating final features of the variable data by using a correlation attenuation rate, considering correlation between historical moments, improving accuracy of the final features, further improving accuracy of final aggregated features of modes obtained according to final feature aggregation, carrying out anomaly prediction according to the accurate final aggregated features, and effectively improving accuracy of anomaly prediction.
The method for predicting abnormality in the low-oxygen expansion stage of stem cells according to the present application will be described in detail.
As shown in FIG. 1, the method for predicting the abnormality of the stem cell hypoxia amplification stage provided by the application comprises the following steps:
and 11, acquiring a plurality of variable data of the environment at a plurality of historical moments when the target stem cells are in a hypoxia amplification stage.
The variable data correspond to a plurality of modes, and each mode at least corresponds to one variable data. The target stem cells are stem cells which are growing and are in a hypoxia expansion stage. The plurality of variable data may be data affecting stem cell growth such as an ambient temperature value, a carbon dioxide concentration value, an oxygen concentration value, etc., and is divided into a plurality of modes such as a temperature (temperature value), a gas (carbon dioxide concentration value, oxygen concentration value), etc. The number of modalities is equal to or less than the number of variable data at each history time. The above-mentioned environment is an environment for stem cell preparation, such as the internal environment of a cell incubator.
In some embodiments of the present application, a plurality of variable data may be acquired using a sensor or the like.
Step 12, calculating the data characteristics of each variable data at each historical moment, and calculating the global characteristics of each historical moment based on all the data characteristics.
Specifically, the formula is as follows:
Calculate the first The first time of historyData characteristics of individual variable data。
Wherein, Which is indicative of the characteristics of the sensor,,,The convolution is represented by a representation of the convolution,Represent the firstA matrix of variable data for each historical moment,Represent the firstThe 1 st variable data of the historical moment,Represent the firstThe 2 nd variable data of the historical moment,Represent the firstThe first time of historyThe data of the individual variables are stored,,The number of variable data representing each historical moment,,The number of times of the history is indicated,Represent the firstThe first time of historyPosition coding of individual variable data:
wherein, Representing integer indices for determining sine and cosine functions of different frequencies,Representing the oscillation frequency of the sine and cosine functions in the position encoding.
By the formula:
Calculate the first Global features for individual historical moments。
Wherein, Represent the firstThe first time of historyThe degree of information center of the individual variable data,Represent the firstThe first time of historyThe data characteristics of the individual variable data,Represent the firstThe first time of historyVariable data and the firstThe first time of historyThe degree of information sharing between the individual variable data,Representation ofIs used as a reference to the entropy of (a),Representation ofIs used as a reference to the entropy of (a),Representation ofAnd (3) withThe information center degree is the sum of the information sharing degree between the variable data and each other variable data at the historical moment.
It should be noted that, in general industrial scenario, multiple monitored variables have symmetrical interdependence relationship between time and space, only features of adjacent variables are considered in related research of existing anomaly detection, and different types of sensors collect variable data of different modes, which have non-uniformity, complex dependency relationship and different sampling frequencies. Although professionals have some prior knowledge in the biological or chemical arts, most of the knowledge is limited only to the variables represented by adjacent variables, there may be hidden interactions between other micro-variables, and there may be unobserved variables that affect the relationship between two observed variables.
It is worth mentioning that the application considers a plurality of variable data, calculates the global feature based on the data features of the plurality of variable data, can comprehensively analyze a plurality of variables of the environment where the target stem cells are located, not only considers the relation of adjacent variables in the variable data, but also effectively improves the information richness of the global feature.
And step 13, calculating the similarity relation between every two variable data at each historical moment based on the global characteristic at each historical moment, and constructing a directed graph according to all the similarity relations.
The nodes of the directed graph correspond to the variable data one by one, and the edges between the two nodes are the similarity relationship between the two corresponding variable data.
In some embodiments of the present application, the step of calculating the similarity between each two variable data at each historical time based on the global feature at each historical time, and constructing the directed graph according to all the similarity specifically includes:
The first step, for each history time, is performed as follows:
First, the integrated similarity between every two variable data at the history time is calculated based on the global feature at the history time.
Specifically, the formula is as follows:
Calculate the first The first time of historyVariable data and the firstThe first time of historyComprehensive similarity between individual variable data;
Wherein, 、AndAll of which represent the saturation rate and,Represent the firstVariable data and the firstThe degree of information sharing between the individual variable data,Represent the firstVariable data and the firstThe degree of information sharing between global features at each historical moment,Represent the firstVariable data and the firstCosine similarity between the individual variable data,Represent the firstThe first time of historyThe data characteristics of the individual variable data,Represent the firstThe first time of historyThe data characteristics of the individual variable data,,,The number of variable data representing each historical moment,,Representing the number of historical moments;
And then, acquiring an initial similarity relation according to the comprehensive similarity between every two variable data, and updating the initial similarity relation to obtain the similarity relation between every two variable data.
Specifically, the formula is as follows:
Acquisition of the first Variable data and the firstInitial similarity relationship between individual variable data。
Wherein, Represents the value of the function of the index,Represent the firstVariable data and the firstNormalized similarity between the individual variable data,Representing a similarity threshold.
By the formula:
For initial similarity relationship Updating to obtain the firstVariable data and the firstSimilarity relationship between individual variable data。
Wherein, Representing the first priori knowledge in a directed graphVariable data and the firstThe adjacency between the individual variable data,Representing an activation function.
It should be noted that, in the prior knowledge directed graph, a plurality of nodes are in one-to-one correspondence with a plurality of variable data, edges between the nodes are prior knowledge relations between the two corresponding variable data, and the prior knowledge relations are used for describing association conditions between the variable data, and can be obtained by manually sorting experiences of professionals and combining expertise analysis in the fields of biology and chemistry. If the value of the prior knowledge relation is 1, the association relation between the two variable data is considered to exist, and if the value of the prior knowledge relation is 0, the association relation between the two variable data is considered to not exist.
And secondly, constructing a directed graph according to all the similarity relations.
Specifically, a plurality of nodes are generated, the nodes correspond to the variable data one by one, and edges between the nodes are corresponding similar relations.
For example, if the value of the similarity relationship is 0, it indicates that there is no similarity relationship between the two corresponding variable data, i.e. there is no connecting edge between the two corresponding nodes.
It is worth mentioning that the accuracy and the global property of the similarity relationship can be improved by acquiring the similarity relationship between the variable data according to the global characteristics, and the similarity relationship between the variable data can be expressed by constructing a directed graph according to the similarity relationship, so that the similarity relationship between the variable data can be analyzed and calculated in the subsequent steps.
And 14, calculating the correlation attenuation rate between every two nodes in the directed graph according to all the historical moments, and acquiring the final characteristics of each variable data based on all the correlation attenuation rates.
The above-mentioned correlation decay rate is used to describe the degree of influence between two history times corresponding to every two nodes.
In some embodiments of the present application, the calculating the correlation attenuation rate between every two nodes in the directed graph according to all the historical moments, and obtaining the final feature of each variable data based on all the correlation attenuation rates specifically includes:
And a first step of calculating the correlation attenuation rate between every two nodes in the directed graph according to all the historical moments.
Specifically, for each history time, the correlation decay rate between the node at the history time and the node at each other history time is defined as;The value of (c) is the difference between the historical time and the other historical time minus one,Is constant.
And a second step of acquiring the final characteristics of each variable data based on all the correlation attenuation rates.
Specifically, the formula is as follows:
Calculate the first The first time of historyFinal characteristics of individual variable data;
Wherein, Representing the parameters that can be learned,Represent the firstThe first time of historyVariable data and the firstThe first time of historyThe comprehensive adjacency between the individual variable data,Represent the firstThe first time of historyThe data characteristics of the individual variable data,The number of times of the history is indicated,The number of variable data representing each historical moment,Represent the firstThe first time of historyVariable data and the firstThe first time of historyAdjacency between variable data ifThen,Represent the firstThe first time of historyVariable data and the firstThe first time of historyComprehensive similarity between variable data, ifThen,Represent the firstNode and No. of historical timeAnd the comprehensive adjacency relation is the adjacency relation between two variable data after the introduction of the correlation attenuation rate for adjustment.
It should be noted that in the context of intelligent stem cell preparation, there is a unidirectional dependency between complex variables that spans both time and space, e.g. in the firstAt a historical timeWill be to the firstAndAt a historical timeThe values respectively have different degrees of influence.
It is worth mentioning that by calculating the correlation attenuation rate and acquiring the final feature based on the correlation attenuation rate, the mutual influence of variable data on time sequence and the influence degree are considered, so that the information reflected by the final feature accords with the actual situation.
And 15, respectively aiming at each mode, aggregating the final characteristics of all variable data corresponding to the mode at each historical moment to obtain the aggregated characteristics of the mode at each historical moment.
Specifically, the formula is as follows:
Calculate the first The first mode is atAggregation features for individual historic moments。
Wherein, Represent the firstThe first mode is atAggregation of all variable data corresponding to historical momentsRepresents the firstThe first time of historyInformation-outputting centrality of the individual variable data.
And step 16, updating all the aggregation characteristics to obtain a plurality of final aggregation characteristics, and carrying out abnormal prediction based on all the final aggregation characteristics to obtain an abnormal prediction result.
The anomaly prediction result is used for describing the anomaly condition of a plurality of variable data of the environment in a future time period (such as the anomaly probability that the variable data is anomaly in all variable data).
In some embodiments of the present application, the updating of all the aggregation features to obtain a plurality of final aggregation features, and the performing of anomaly prediction based on all the final aggregation features to obtain an anomaly prediction result specifically includes:
in the first step, the degree of association between every two aggregated features is calculated.
Specifically, the formula is as follows:
Calculate the first Modality and the firstBetween the first modesDegree of correlation of individual historic moments。
Wherein, Represent the firstThe first mode is atA set of all variable data corresponding to each historical moment,Represent the firstThe first time of historyVariable data and the firstThe first time of historyThe association relationship between the individual variable data,,Represent the firstThe first time of historyVariable data and the firstThe first time of historyComprehensive similarity between the individual variable data.
And secondly, updating all the aggregation features according to all the association degrees to obtain a plurality of final aggregation features.
Specifically, the formula is as follows:
All the aggregation features are updated.
Wherein, Represent the firstThe feature matrix is aggregated in the middle of the layer,Representing a correlation degree matrix (i.e., a matrix composed of all correlation degrees, the same correlation degree matrix is used for each layer of operation, and when the variable data at the historic moment changes, the correlation degree is updated by the above formula for calculating the correlation degree),Represent the firstThe matrix of learnable weights of the layer,Represent the firstThe feature matrix is aggregated in the middle of the layer,The activation function is represented as a function of the activation,,Representing the update of the last layer whenIn the time-course of which the first and second contact surfaces,Representing the aggregate feature matrix (i.e., the matrix of all aggregate feature components obtained in the previous step), whenIn the time-course of which the first and second contact surfaces,Representing the final aggregate feature matrix.
It should be noted that, the learnable weight matrix may be optimized by using an optimization algorithm such as a gradient descent method, so as to improve the calculation performance of the above formula for updating all the aggregation features.
Thirdly, carrying out abnormal prediction based on all final aggregation characteristics to obtain an abnormal prediction result.
Specifically, a matrix formed by all final aggregation features can be input into a prediction model (such as a softmax activation function layer, etc.), the prediction model calculates the abnormal probability of all variable data in a future time window with the same size as a time window formed by all historical moments based on all final aggregation features, and an abnormal prediction result is obtained.
After the abnormal probability is obtained, early warning can be performed according to the value of the abnormal probability, corresponding measures are timely taken to keep the values of a plurality of variables of the environment where the target stem cells are located in the low-oxygen amplification stage, if the value of the abnormal probability is larger than the abnormal probability threshold value, early warning is performed, if the value of the abnormal probability is smaller than the abnormal probability threshold value, early warning is not needed, and if early warning occurs, the temperature, the gas concentration and the like of the environment need to be managed and regulated so as to keep the environment state in the state where the stem cells are needed in the low-oxygen amplification stage. The future time is a time after the current time, if the current time is 9 points, the future time may be 10 points, and the plurality of historical times may be 7 points, half 7 points, and 8 points.
In order to ensure the accuracy of the prediction model for advanced prediction, the prediction model needs to be trained before the steps are carried out, the data abnormal conditions of each mode at a plurality of times after each historical time are obtained, the prediction model is trained based on all the data abnormal conditions, and specifically, the prediction model is utilized for the first stepPredicting historical time to obtain the data including the firstFuture of individual historical momentsWindow anomaly probability for each instantDynamic windowWill beAs the fixed firstIndividual historical time prediction windowsWherein. Obviously, the greater the anomaly probability, the closer the anomaly occurrence is, so a dynamic window mechanism is adopted for the windowIs of the dynamic size ofThenAnd (3) withThe relationship of (2) is as follows:
wherein, To adjust the parameters, the window size is adjusted empirically. Future ofTrue tag value at each instant,. In the first placeAt each historical time, if setIf there is an abnormal label, the abnormal probability is describedIs accurately predicted if inIf no abnormal label exists in the list, the abnormal probability is describedThe calculation is inaccurate.
Probability of abnormalityToo high to be in factNo anomalies occur in the collection, or anomaly probabilitiesLow but atAnomalies in the collection occur, both of which should result in higher losses. Therefore, constructing a loss function to train the prediction model, wherein the loss function is as follows:
wherein, The value of the loss function is represented,Represent the firstThe weight of the individual historical moments in time,Represent the firstThe deviation loss at the time of the history,Indicating an actual abnormal condition of the device,,,Representing the deviation value.
And training the prediction model according to the loss function, if the value of the loss function is smaller than the preset value of the loss function, considering that the training of the prediction model is completed, carrying out the step of carrying out abnormal prediction based on all final aggregation characteristics to obtain an abnormal prediction result, and if the value of the loss function is larger than or equal to the preset value of the loss function, adjusting the parameters of the prediction model, and continuing training.
It is worth mentioning that the global feature is calculated based on the data features of the variable data, multiple variables of the environment where the target stem cells are located can be comprehensively analyzed, the information richness of the global feature is improved, the similarity relationship between the variable data is obtained according to the global feature, the accuracy and the global property of the similarity relationship can be improved, the final feature of the variable data is calculated by utilizing the correlation attenuation rate, the correlation between historical moments is considered, the accuracy of the final feature is improved, the accuracy of the final aggregated feature of the mode obtained according to the final feature aggregation is further improved, the abnormal prediction is carried out according to the accurate final aggregated feature, and the accuracy of the abnormal prediction can be effectively improved.
The following is an exemplary description of the device for predicting abnormality in the hypoxia expansion phase of stem cells according to the present application.
As shown in fig. 2, an embodiment of the present application provides an abnormality prediction apparatus for a stem cell hypoxia amplification stage, wherein the abnormality prediction apparatus 200 for a stem cell hypoxia amplification stage includes:
The acquisition module 201 acquires a plurality of variable data of the environment where the target stem cells are located at a plurality of historical moments in a hypoxia amplification stage, wherein the plurality of variable data correspond to a plurality of modes, and each mode at least corresponds to one variable data;
A first calculation module 202 that calculates a data feature of each variable data at each history time, and calculates a global feature at each history time based on all the data features;
the construction module 203 calculates the similarity relation between every two variable data at each historical moment based on the global feature at each historical moment, and constructs a directed graph according to all the similarity relations;
The second calculation module 204 calculates the correlation attenuation rate between every two nodes in the directed graph according to all the historical moments, and obtains the final feature of each variable data based on all the correlation attenuation rates;
The aggregation module 205 aggregates the final characteristics of all variable data corresponding to the modes at each historical moment for each mode respectively to obtain the aggregated characteristics of the modes at each historical moment;
The anomaly prediction module 206 updates all the aggregation features to obtain a plurality of final aggregation features, and performs anomaly prediction based on all the final aggregation features to obtain an anomaly prediction result, where the anomaly prediction result is used to describe the anomaly condition of a plurality of variable data of the environment in a future time period.
It should be noted that, because the content of information interaction and execution process between the above devices/units is based on the same concept as the method embodiment of the present application, specific functions and technical effects thereof may be referred to in the method embodiment section, and will not be described herein.
It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-described division of the functional units and modules is illustrated, and in practical application, the above-described functional distribution may be performed by different functional units and modules according to needs, i.e. the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-described functions. The functional units and modules in the embodiment may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit, where the integrated units may be implemented in a form of hardware or a form of a software functional unit. In addition, the specific names of the functional units and modules are only for distinguishing from each other, and are not used for limiting the protection scope of the present application. The specific working process of the units and modules in the above system may refer to the corresponding process in the foregoing method embodiment, which is not described herein again.
As shown in fig. 3, an embodiment of the present application provides a terminal device D10 of the embodiment comprising at least one processor D100 (only one processor is shown in fig. 3), a memory D101 and a computer program D102 stored in the memory D101 and executable on the at least one processor D100, the processor D100 implementing the steps of any of the respective method embodiments described above when executing the computer program D102.
Specifically, when the processor D100 executes the computer program D102, by calculating the data features of each variable data at each historical moment, calculating the global feature at each historical moment based on all the data features, calculating the similarity relationship between every two variable data at each historical moment based on the global feature at each historical moment, constructing a directed graph according to all the similarity relationships, calculating the correlation attenuation rate between every two nodes in the directed graph according to all the historical moments, acquiring the final feature of each variable data based on all the correlation attenuation rates, aggregating the final feature of all the variable data corresponding to each mode at each historical moment for each mode, obtaining the aggregate feature of the mode at each historical moment, updating all the aggregate features, obtaining a plurality of final aggregate features, performing anomaly prediction based on all the final aggregate features, and obtaining the anomaly prediction result. The method comprises the steps of calculating global features based on data features of a plurality of variable data, comprehensively analyzing a plurality of variables of an environment where target stem cells are located, improving information richness of the global features, obtaining similarity relations between the variable data according to the global features, improving accuracy and global performance of the similarity relations, calculating final features of the variable data by using a correlation attenuation rate, considering correlation between historical moments, improving accuracy of the final features, further improving accuracy of final aggregated features of modes obtained according to final feature aggregation, carrying out anomaly prediction according to the accurate final aggregated features, and effectively improving accuracy of anomaly prediction.
The Processor D100 may be a central processing unit (CPU, central Processing Unit), the Processor D100 may also be other general purpose processors, digital signal processors (DSP, digital Signal processors), application SPECIFIC INTEGRATED integrated circuits (ASICs), off-the-shelf Programmable gate arrays (FPGAs) or other Programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The memory D101 may in some embodiments be an internal storage unit of the terminal device D10, for example a hard disk or a memory of the terminal device D10. The memory D101 may also be an external storage device of the terminal device D10 in other embodiments, for example, a plug-in hard disk, a smart memory card (SMC, smart Media Card), a Secure Digital (SD) card, a flash memory card (FLASH CARD) or the like, which are provided on the terminal device D10. Further, the memory D101 may also include both an internal storage unit and an external storage device of the terminal device D10. The memory D101 is used for storing an operating system, an application program, a boot loader (BootLoader), data, other programs, etc., such as program codes of the computer program. The memory D101 may also be used to temporarily store data that has been output or is to be output.
Embodiments of the present application also provide a computer readable storage medium storing a computer program which, when executed by a processor, implements steps for implementing the various method embodiments described above.
Embodiments of the present application provide a computer program product enabling a terminal device to carry out the steps of the method embodiments described above when the computer program product is run on the terminal device.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the present application may implement all or part of the flow of the method of the above embodiments, and may be implemented by a computer program to instruct related hardware, where the computer program may be stored in a computer readable storage medium, and when the computer program is executed by a processor, the computer program may implement the steps of each of the method embodiments described above. Wherein the computer program comprises computer program code which may be in source code form, object code form, executable file or some intermediate form etc. The computer readable medium may include at least any entity or device, a recording medium, a computer Memory, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), an electrical carrier signal, a telecommunications signal, and a software distribution medium that is capable of carrying computer program code to the abnormality prediction method device/terminal device of the hypoxia amplification stage. Such as a U-disk, removable hard disk, magnetic or optical disk, etc.
In the foregoing embodiments, the descriptions of the embodiments are emphasized, and in part, not described or illustrated in any particular embodiment, reference is made to the related descriptions of other embodiments.
Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
While the foregoing is directed to the preferred embodiments of the present application, it will be appreciated by those skilled in the art that various modifications and adaptations can be made without departing from the principles of the present application, and such modifications and adaptations are intended to be comprehended within the scope of the present application.
Claims (6)
1. A method for predicting abnormalities in a stem cell hypoxia-expansion phase, comprising:
Acquiring a plurality of variable data of the environment where the target stem cells are located at a plurality of historical moments in a hypoxia amplification stage, wherein the plurality of variable data correspond to a plurality of modes, and each mode at least corresponds to one variable data;
calculating the data characteristics of each variable data of each historical moment, and calculating the global characteristics of each historical moment based on all the data characteristics;
calculating the similarity relation between every two variable data at each historical moment based on the global feature at each historical moment, and constructing a directed graph according to all the similarity relations, wherein a plurality of nodes of the directed graph are in one-to-one correspondence with the variable data, and edges between the two nodes are the similarity relations between the two corresponding variable data;
Calculating the correlation attenuation rate between every two nodes in the directed graph according to all the historical moments, and acquiring the final characteristic of each variable data based on all the correlation attenuation rates, wherein the correlation attenuation rate is used for describing the influence degree between the two historical moments corresponding to every two nodes;
respectively aiming at each mode, aggregating the final characteristics of all variable data corresponding to the mode at each historical moment to obtain the aggregated characteristics of the mode at each historical moment;
updating all the aggregation features to obtain a plurality of final aggregation features, and carrying out exception prediction based on all the final aggregation features to obtain an exception prediction result, wherein the exception prediction result is used for describing the exception condition of a plurality of variable data of the environment in a future time period;
the calculating the data characteristic of each variable data of each historical moment comprises the following steps:
by the formula:
Calculate the first The first time of historyData characteristics of individual variable data;
Wherein, Which is indicative of the characteristics of the sensor,,,The convolution is represented by a representation of the convolution,Represents the firstA matrix of variable data for each historical moment,Represents the firstThe 1 st variable data of the historical moment,Represents the firstThe 2 nd variable data of the historical moment,Represents the firstThe first time of historyThe data of the individual variables are stored,,The number of variable data representing each historical moment,,The number of times of the history is indicated,Represents the firstThe first time of historyPosition coding of individual variable data:
wherein, Represents an index of an integer number,Representing the oscillation frequency;
the calculating the global feature of each historical moment based on all the data features comprises:
by the formula:
Calculate the first Global features for individual historical moments;
Wherein, Represents the firstThe first time of historyThe degree of information center of the individual variable data,Represents the firstThe first time of historyThe data characteristics of the individual variable data,Represents the firstThe first time of historyVariable data and the firstThe first time of historyThe degree of information sharing between the individual variable data,Representation ofIs used as a reference to the entropy of (a),Representation ofIs used as a reference to the entropy of (a),Representation ofAnd (3) withThe information center degree is the sum of the information sharing degree between the variable data and each other variable data at the historical moment;
the calculating the correlation attenuation rate between every two nodes in the directed graph according to all the historical moments comprises the following steps:
For each history time, defining the correlation attenuation rate between the node of the history time and the node of each other history time as ;The value of (2) is the difference between the historical time and the other historical time minus one,Is a constant;
The obtaining the final feature of each variable data based on all the correlation attenuation rates includes:
by the formula:
Calculate the first The first time of historyFinal characteristics of individual variable data;
Wherein, Representing the parameters that can be learned,Represent the firstThe first time of historyVariable data and the firstThe first time of historyThe comprehensive adjacency between the individual variable data,Represents the firstThe first time of historyThe data characteristics of the individual variable data,The number of times of the history is indicated,The number of variable data representing each historical moment,Represent the firstThe first time of historyVariable data and the firstThe first time of historyAdjacency between variable data ifThen,Represents the firstThe first time of historyVariable data and the firstThe first time of historyComprehensive similarity between variable data, ifThen,Represent the firstNode and No. of historical timeThe comprehensive adjacency relation is the adjacency relation between two variable data after the adjustment of the introduced correlation attenuation rate;
The step of aggregating the final characteristics of all variable data corresponding to the mode at each historical moment to obtain the aggregate characteristics of the mode at each historical moment comprises the following steps:
by the formula:
Calculate the first The first mode is atAggregation features for individual historic moments;
Wherein, Represents the firstThe first mode is atA set of all variable data corresponding to each historical moment,Represents the firstThe first time of historyInformation-outputting centrality of the individual variable data.
2. The anomaly prediction method according to claim 1, wherein the calculating of the similarity relationship between every two variable data at each of the historical moments based on the global feature at each of the historical moments includes:
the following steps are respectively carried out for each historical moment:
calculating the comprehensive similarity between every two variable data of the historical moment based on the global characteristic of the historical moment;
And acquiring an initial similarity relation according to the comprehensive similarity between every two variable data, and updating the initial similarity relation to obtain the similarity relation between every two variable data.
3. The anomaly prediction method of claim 2, wherein the calculating the integrated similarity between every two variable data at the historical moment based on the global feature at the historical moment comprises:
by the formula:
Calculate the first The first time of historyVariable data and the firstThe first time of historyComprehensive similarity between individual variable data;
Wherein, 、AndAll of which represent the saturation rate and,Represents the firstVariable data and the firstThe degree of information sharing between the individual variable data,Represents the firstVariable data and the firstThe degree of information sharing between global features at each historical moment,Represents the firstVariable data and the firstCosine similarity between the individual variable data,Represents the firstThe first time of historyThe data characteristics of the individual variable data,Represents the firstThe first time of historyThe data characteristics of the individual variable data,,,The number of variable data representing each historical moment,,Representing the number of historical moments;
the obtaining the initial similarity relation according to the comprehensive similarity between every two variable data comprises the following steps:
by the formula:
Acquiring the first Variable data and the firstInitial similarity relationship between individual variable data;
Wherein, Represents the value of the function of the index,Represents the firstVariable data and the firstNormalized similarity between the individual variable data,Representing a similarity threshold;
The updating the initial similarity relationship to obtain a similarity relationship between every two variable data comprises the following steps:
by the formula:
For initial similarity relationship Updating to obtain the firstVariable data and the firstSimilarity relationship between individual variable data;
Wherein, Representing the first described in the prior knowledge directed graphVariable data and the firstThe adjacency between the individual variable data,Representing an activation function.
4. The anomaly prediction method of claim 1, wherein updating all the aggregated features to obtain a plurality of final aggregated features comprises:
calculating the association degree between every two aggregation features;
and updating all the aggregation features according to all the association degrees to obtain a plurality of final aggregation features.
5. The anomaly prediction method of claim 4, wherein the calculating the degree of association between each two aggregated features comprises:
by the formula:
Calculate the first Modality and the firstBetween the first modesDegree of correlation of individual historic moments;
Wherein, Represents the firstThe first mode is atA set of all variable data corresponding to each historical moment,Represents the firstThe first time of historyVariable data and the firstThe first time of historyThe association relationship between the individual variable data,,Represents the firstThe first time of historyVariable data and the firstThe first time of historyComprehensive similarity between the individual variable data;
Updating all the aggregation features according to all the association degrees to obtain a plurality of final aggregation features, wherein the updating comprises the following steps:
by the formula:
updating all the aggregation characteristics;
wherein, Represent the firstThe feature matrix is aggregated in the middle of the layer,The degree of association matrix is represented by a matrix of degrees of association,Represent the firstThe matrix of learnable weights of the layer,Represent the firstThe feature matrix is aggregated in the middle of the layer,The activation function is represented as a function of the activation,,Representing the update of the last layer whenIn the time-course of which the first and second contact surfaces,Representing an aggregate feature matrix whenIn the time-course of which the first and second contact surfaces,Representing the final aggregate feature matrix.
6. A terminal device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the method for abnormality prediction of the stem cell hypoxia-expansion phase according to any of claims 1 to 5 when executing the computer program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202411852226.4A CN119312114B (en) | 2024-12-16 | 2024-12-16 | A method for predicting abnormalities during the hypoxia expansion phase of stem cells and related equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202411852226.4A CN119312114B (en) | 2024-12-16 | 2024-12-16 | A method for predicting abnormalities during the hypoxia expansion phase of stem cells and related equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN119312114A CN119312114A (en) | 2025-01-14 |
| CN119312114B true CN119312114B (en) | 2025-03-25 |
Family
ID=94187880
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202411852226.4A Active CN119312114B (en) | 2024-12-16 | 2024-12-16 | A method for predicting abnormalities during the hypoxia expansion phase of stem cells and related equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN119312114B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108320040A (en) * | 2017-01-17 | 2018-07-24 | 国网重庆市电力公司 | Acquisition terminal failure prediction method and system based on Bayesian network optimization algorithm |
| WO2024097315A2 (en) * | 2022-11-02 | 2024-05-10 | Sana Biotechnology, Inc. | Cell therapy products and methods for producing same |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| AU785425B2 (en) * | 2001-03-30 | 2007-05-17 | Genetic Technologies Limited | Methods of genomic analysis |
| US8751867B2 (en) * | 2011-10-12 | 2014-06-10 | Vmware, Inc. | Method and apparatus for root cause and critical pattern prediction using virtual directed graphs |
| CN104928235A (en) * | 2015-07-10 | 2015-09-23 | 奥思达干细胞有限公司 | Composition based on stem cells and application thereof in preparing preparation for coronary heart disease |
| CN107190101A (en) * | 2017-08-01 | 2017-09-22 | 安诺优达基因科技(北京)有限公司 | A kind of method and kit for Embryonic limb bud cell prochromosome abnormality detection |
| CN110175085B (en) * | 2019-04-19 | 2020-11-27 | 同济大学 | A method and device for diagnosing abnormal causes of Hadoop system using graph analysis |
| IL305678A (en) * | 2021-03-08 | 2023-11-01 | Hemanext Inc | Methods for storing hematopoietic stem cells |
| CN117708738A (en) * | 2023-12-16 | 2024-03-15 | 郑州大学 | Sensor time sequence anomaly detection method and system based on multi-modal variable correlation |
-
2024
- 2024-12-16 CN CN202411852226.4A patent/CN119312114B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108320040A (en) * | 2017-01-17 | 2018-07-24 | 国网重庆市电力公司 | Acquisition terminal failure prediction method and system based on Bayesian network optimization algorithm |
| WO2024097315A2 (en) * | 2022-11-02 | 2024-05-10 | Sana Biotechnology, Inc. | Cell therapy products and methods for producing same |
Also Published As
| Publication number | Publication date |
|---|---|
| CN119312114A (en) | 2025-01-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Grbić et al. | Adaptive soft sensor for online prediction and process monitoring based on a mixture of Gaussian process models | |
| CN102819772B (en) | Power matching network builds material requirements Forecasting Methodology and device | |
| CN112529301A (en) | Power consumption prediction method, equipment and storage medium | |
| Zheng et al. | A deep learning approach for optimizing monoclonal antibody production process parameters | |
| Huang et al. | An experimental method to estimate running time of evolutionary algorithms for continuous optimization | |
| CN106295121A (en) | Landscape impoundments Bayes's water quality grade Forecasting Methodology | |
| Yao et al. | Nonlinear Gaussian mixture regression for multimode quality prediction with partially labeled data | |
| CN118536410B (en) | Big data driven modeling-based energy consumption optimization decision analysis method and system | |
| CN116050674A (en) | Hydraulic engineering operation trend prediction method and device | |
| CN117875037A (en) | BOPP film production line digital simulation modeling method and system | |
| Gangadharan et al. | Metaheuristic approaches in biopharmaceutical process development data analysis | |
| CN111738530B (en) | River water quality prediction method, device and computer readable storage medium | |
| Bae et al. | Construction of a valid domain for a hybrid model and its application to dynamic optimization with controlled exploration | |
| CN110046377A (en) | A kind of selective ensemble instant learning soft-measuring modeling method based on isomery similarity | |
| Zhang et al. | Implementable stability guaranteed Lyapunov-based data-driven model predictive control with evolving Gaussian process | |
| Li et al. | Data cleaning method for the process of acid production with flue gas based on improved random forest | |
| Garcia-Ceja et al. | Towards the automation of a chemical sulphonation process with machine learning | |
| CN116628444A (en) | Water quality early warning method based on improved meta-learning | |
| Karkaria et al. | A Digital Twin Framework Utilizing Machine Learning for Robust Predictive Maintenance: Enhancing Tire Health Monitoring | |
| CN119312114B (en) | A method for predicting abnormalities during the hypoxia expansion phase of stem cells and related equipment | |
| CN114726751A (en) | Intelligent early warning method, system, equipment and storage medium for resource quality monitoring | |
| Vaitkus et al. | Soft-sensors based on black-box models for bioreactors monitoring and state estimation | |
| EP4053661B1 (en) | Measurement data recording apparatus, method, and program | |
| Li et al. | Online Monitoring of Heterogeneous Partially Observable Data Streams Based on Q-Learning | |
| CN116629918A (en) | User consumption prediction method and system based on cross-border electronic commerce |
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 |