Detailed Description
Exemplary embodiments of the present application will now be described with reference to the accompanying drawings, in which various details of the embodiments of the present application are included to facilitate understanding, and are to be considered merely exemplary. Accordingly, one of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope of the application. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness. In the present application, the use of the terms "first," "second," etc. to describe various elements is not intended to limit the positional relationship, timing relationship, or importance relationship of these elements unless otherwise indicated, and such terms are merely used to distinguish one element from another element. In some examples, a first element and a second element may refer to the same instance of the element, and in some cases, they may also refer to different instances based on the description of the context. The terminology used in the description of the various illustrated examples in this disclosure is for the purpose of describing particular examples only and is not intended to be limiting. Unless the context clearly indicates otherwise, the elements may be one or more if the number of the elements is not specifically limited. Furthermore, the term "and/or" as used in the present application encompasses any and all possible combinations of the listed items.
As shown in fig. 1, in an application scenario of the method provided by the embodiment of the present application, one or more remote requesters 101 are connected to a data processing system 120 through a communication network 110.
In an embodiment of the application, the data processing system 120 may run one or more services or software applications that enable execution of smart care remote data processing methods based on smart medical treatment. In some embodiments, data processing system 120 may also provide other services or software applications, which may include non-virtual environments and virtual environments. In some embodiments, these services may be provided as web-based services or cloud services, for example, provided to users of remote requesters 101 under a software as a service (SaaS) model. In the configuration shown in FIG. 1, data processing system 120 may include one or more components that implement the functions performed by data processing system 120. These components may include software components, hardware components, or a combination thereof that are executable by one or more processors. A user operating remote requestor 101 may, in turn, utilize one or more applications to interact with data processing system 120 to take advantage of the services provided by these components. It should be appreciated that a variety of different scenario configurations are possible, which may differ from the architecture described above. Accordingly, FIG. 1 is one example of a system for implementing the various methods described herein and is not intended to be limiting.
In a care assessment scenario, the remote requesting party 101 may be any electronic device capable of uploading care data, such as a portable handheld device, a general purpose computer (such as a personal computer and a laptop computer), a workstation computer, a wearable device, a smart screen device, a self-service terminal device, a service robot, a gaming system, a thin client, various messaging devices, sensors or other sensing devices, etc., which may run various types and versions of software applications and operating systems, such as MICROSOFT Windows, application iOS, UNIX-like operating systems, linux, or Linux-like operating systems (e.g., GOOGLE Chrome OS); or include various mobile operating systems such as MICROSOFT Windows Mobile OS, iOS, windows Phone, android. Portable handheld devices may include cellular telephones, smart phones, tablet computers, personal Digital Assistants (PDAs), and the like. Wearable devices may include head mounted displays (such as smart glasses) and other devices. The gaming system may include various handheld gaming devices, internet-enabled gaming devices, and the like. The associated devices are capable of executing a variety of different applications, such as various Internet-related applications, communication applications (e.g., email applications), short Message Service (SMS) applications, and may use a variety of communication protocols. The assessment requestor may upload care data using the remote requestor 101. Network 110 may be any type of network known to those skilled in the art that may support data communications using any of a number of available protocols, including but not limited to TCP/IP, SNA, IPX, etc. For example only, the one or more networks 110 may be a Local Area Network (LAN), an ethernet-based network, a token ring, a Wide Area Network (WAN), the internet, a virtual network, a Virtual Private Network (VPN), an intranet, an extranet, a blockchain network, a Public Switched Telephone Network (PSTN), an infrared network, a wireless network (e.g., bluetooth, WIFI), and/or any combination of these and/or other networks. Data processing system 120 may include one or more general purpose computers, special purpose server computers (e.g., a PC (personal computer) server, a UNIX server, a middleend server), a blade server, a mainframe computer, a cluster of servers, or any other suitable arrangement and/or combination. Data processing system 120 may include one or more virtual machines running a virtual operating system, or other computing architecture that involves virtualization (e.g., one or more flexible pools of logical storage devices that may be virtualized to maintain virtual storage devices of servers). In various embodiments, data processing system 120 may run one or more services or software applications that provide the functionality described below.
The computing units in data processing system 120 may run one or more operating systems including any of the operating systems described above as well as any commercially available server operating systems. The data processing system 120 may also run any of a variety of additional server applications and/or middle tier applications, including HTTP servers, FTP servers, CGI servers, JAVA servers, database servers, etc. In some implementations, the data processing system 120 can include one or more applications to analyze and consolidate data feeds and/or event updates received from users of the remote requesters 101. The data processing system 120 can also include one or more applications to display data feeds and/or real-time events via one or more display devices of the remote requestor 101. In some implementations, the data processing system 120 can be a server of a distributed system or a server that incorporates a blockchain. The data processing system 120 may also be a cloud server, or an intelligent cloud computing server or intelligent cloud host with artificial intelligence techniques. The cloud server is a host product in a cloud computing service system, so as to solve the defects of large management difficulty and weak service expansibility in the traditional physical host and virtual private server (VPS, virtual Private Server) service.
Data processing system 100 also may include one or more databases 130. In some embodiments, these databases may be used to store data and other information. For example, one or more of the databases 130 may be used to store care data and care assessments. Database 130 may reside in various locations. For example, a database used by data processing system 120 may be local to data processing system 120 or may be remote from data processing system 120 and may communicate with data processing system 120 via a network-based or dedicated connection. Database 130 may be of different types. In some embodiments, the database used by data processing system 120 may be, for example, a relational database. One or more of these databases may store, update, and retrieve the databases and data from the databases in response to the commands.
In some embodiments, one or more of databases 130 may also be used by applications to store application data. The databases used by the application may be different types of databases, such as key value stores, object stores, or conventional stores supported by the file system.
Referring to fig. 2, the smart care remote data processing method based on smart medical treatment according to an embodiment of the present application includes the following steps 100 to 700:
100: and receiving the nursing evaluation instruction, and responding to the nursing evaluation instruction to acquire the number information carried by the nursing evaluation instruction.
In the embodiment of the application, the nursing evaluation instruction is sent by a remote requesting party or is generated regularly by a server, and the remote requesting party is a nursing area of a hospital, a home nursing guide and the like. The care evaluation instruction may be automatically transmitted by an electronic device provided at the remote requester according to a preset period or manually transmitted by a person of the remote requester. The number information carried in the nursing evaluation instruction is used for indicating the corresponding remote requesting party so as to distinguish, the same remote requesting party can send the nursing evaluation instruction according to a preset period, and the number information in the nursing evaluation instruction sent in each period can be the same.
200: and retrieving a target object nursing data set corresponding to the number information from the nursing database, wherein the target object nursing data set comprises nursing record data of the target object in the current evaluation period.
A plurality of subject care data sets are stored in the care database, along with a historical care data set for each subject. It will be appreciated that the care data set for each subject corresponds to a unique number. The corresponding target object nursing data set can be accurately called through the unique number. The care record data may include process data recorded by a care person for nursing the target object in a history period, including but not limited to administration frequency, administration dosage, massage frequency, cleaning frequency, content, physical index detection items, etc., and the electronic device of the remote requesting party stores the data recorded by the care person after quantifying and periodically uploads the data to the server for saving in a care database, for example, the care record data may be text information. The specific cycle range of the evaluation cycle may be adapted according to practical situations, for example, one week, which is not limited by the embodiment of the present application.
300: a core care element included in the target subject care dataset is acquired.
The core nursing elements are nursing data corresponding to the key nursing dimensions marked in the target object nursing data set, such as nursing data generated by dimensions of wound cleaning, vital sign monitoring and the like.
400: mining vector representations of data blocks included in a target object care dataset based on a care data embedding mapping network, and mining vector representations of a plurality of comparative care evaluations previously deployed based on a care evaluation embedding mapping network.
The nursing data embedding mapping network and the nursing evaluation embedding mapping network are respectively obtained through mixed debugging according to output results of repeated debugging and optimization (namely rolling debugging or iterative debugging), learning of mutual information is carried out through alternate debugging, and the target object nursing data set comprises a plurality of data blocks. The comparison nursing evaluation is stored in a database of the server in advance, and the comparison nursing evaluation is taken as a corresponding nursing evaluation which comprises evaluation content of a nursing process aiming at a target object and matched nursing reference data, wherein the evaluation content can be reflected through scores, the nursing reference data is conventional nursing data obtained through induction statistics, and it is understood that the more the number of comparison nursing evaluation is, the more accurate the comparison between the core nursing elements and the comparison nursing evaluation is.
The nursing data embedding mapping network and the nursing evaluation embedding mapping network can be a transducer model based on an attention mechanism, for example, a BERT model can be specifically adopted, and the nursing data embedding mapping network obtained through debugging can mine characteristic information of more accurate nursing data.
For example, a target object care data set is loaded into a care data embedding mapping network (a network encoded by embedding mapping), the target object care data set is split into a plurality of data blocks by a data decomposition module in the care data embedding mapping network, and then the care data embedding mapping network respectively mines vector representations of each data block to obtain vector representations of the data blocks of the target object care data set, wherein the vector representations contain characteristic information of the data blocks. Of course, the care data embedding mapping network and the care evaluation embedding mapping network can also be obtained by debugging by adopting other machine learning model architectures, such as LSTM, CNN and the like.
The embodiment of the application is carried out in a mixed mode, in other words is carried out alternately in the debugging process of the nursing data embedding mapping network and the nursing evaluation embedding mapping network, specifically, the rolling real-time nursing data embedding mapping network is obtained through the nursing data-nursing evaluation comparison error and is used for outputting the data block vector representation in the nursing evaluation-nursing data comparison error, the real-time nursing data embedding mapping network is a Momentum Encoder, the rolling real-time nursing data embedding mapping network is the Momentum Encoder after iteration, and the follow-up real-time nursing evaluation embedding mapping network is understood similarly. The real-time nursing data is introduced to be embedded into the mapping network, so that the output at the current moment is not completely determined by the network parameter at the current moment, and the updating of the network is slow and the consistency is stronger due to the fact that the network parameter at the last moment is also dependent on the network parameter at the current moment. And obtaining a rolling real-time nursing evaluation embedded mapping network through nursing evaluation-nursing data comparison errors, wherein the rolling real-time nursing evaluation embedded mapping network is used for next-round debugging and outputting vector representation of a first nursing evaluation example, and the processes are alternately executed in sequence. Then, the nursing data embedding mapping network and the nursing evaluation embedding mapping network are respectively obtained through mixed debugging according to the output results of the repeated debugging optimization.
500: a first correlation coefficient between the core care element and each of the comparative care evaluations is determined based on the vector characterization of the data chunk and the vector characterization of the plurality of comparative care evaluations.
For example, the correlation coefficient between the vector representation of each data block and the vector representation of each comparative care evaluation (the degree of matching between the two representations) is determined respectively, the core care element to which each data block belongs is determined, and then the first correlation coefficient between the core care element and each comparative care evaluation is obtained based on the correlation coefficient between the data block included in the core care element and the vector representation of each comparative care evaluation.
600: a second correlation coefficient between the target subject care dataset and the comparative care assessment is determined based on the first correlation coefficient.
The target subject care dataset is comprised of a plurality of core care elements, a first correlation coefficient of each core care element with the comparative care assessment determining a second correlation coefficient of the target subject care dataset with the comparative care assessment. Since the influence coefficients of the respective core care elements in the target object care data set are different, a second correlation coefficient between the target object care data set and the comparative care evaluation may be determined together based on the influence coefficients of the core care elements and the first correlation coefficient, so that the second correlation coefficient between the target object care data set and the respective comparative care evaluation is obtained.
700: based on the second correlation coefficient, a target care evaluation is determined from the plurality of comparative care evaluations, the target care evaluation being for an evaluation result as a target subject care dataset.
The second correlation coefficient characterizes a commonality metric, i.e., similarity, between the vector characterization of the target subject care dataset and the vector characterization of the comparative care evaluation. Then, the greater the second correlation coefficient, the greater the degree of matching of the target subject care dataset and the comparative care assessment. Then, a plurality of comparative care evaluations whose second correlation coefficient is forward may be selected from the second correlation coefficients corresponding to the respective comparative care evaluations as target care evaluations of the evaluation results of the target subject care dataset, while the target care evaluation is returned to the remote requester.
According to the method provided by the embodiment of the application, the nursing evaluation instruction is received, the number information carried by the nursing evaluation instruction is obtained in response to the nursing evaluation instruction, the target object nursing data set corresponding to the number information is obtained from the nursing database, the target object nursing data set comprises nursing record data of a target object in a current evaluation period, then core nursing elements included in the target object nursing data set are obtained, vector representations of data blocks included in the target object nursing data set are mined based on the nursing data embedding mapping network, vector representations of a plurality of contrast nursing evaluations deployed in advance are mined based on the nursing evaluation embedding mapping network, the nursing data embedding mapping network and the nursing evaluation embedding mapping network are obtained through mixed debugging according to output results of repeated debugging optimization, a first association coefficient between the core nursing elements and each contrast nursing evaluation is determined based on the vector representations of the data blocks, a second association coefficient between the target object nursing data set and each contrast nursing evaluation is determined based on the first association coefficient, and the target nursing evaluation is determined from the plurality of contrast nursing evaluations based on the second association coefficient, and the target evaluation is used as the evaluation result of the target object nursing data set. In the process, the core nursing elements in the nursing data set based on the target object are associated with the contrast nursing evaluation, so that the nursing requirement information of the core nursing elements is matched with the nursing evaluation content, the nursing data and the nursing evaluation are associated more finely, in addition, the nursing data for excavating vector characterization are embedded into the mapping network and the nursing evaluation is embedded into the mapping network, the nursing data for excavating vector characterization are obtained through mixed debugging according to the output results of the nursing data for repeated debugging and optimization, the feature vector characterization performance of the network is improved through the unsupervised debugging process of contrast learning, the nursing information matching effect of the nursing data and the nursing evaluation is further improved, and the accuracy is high.
As another embodiment of the present application, an intelligent care remote data processing method based on intelligent medical treatment includes: .
101: and receiving the nursing evaluation instruction, and responding to the nursing evaluation instruction to acquire the number information carried by the nursing evaluation instruction.
102: and retrieving a target object nursing data set corresponding to the number information from the nursing database, wherein the target object nursing data set comprises nursing record data of the target object in the current evaluation period.
103: a core care element included in the target subject care dataset is acquired.
104: mining vector representations of data blocks included in the target object care dataset based on a care data embedding mapping network, and mining vector representations of a plurality of comparative care evaluations deployed in advance based on a care evaluation embedding mapping network.
The nursing data embedding mapping network and the nursing evaluation embedding mapping network are respectively obtained through mixed debugging according to the output results of the nursing data embedding mapping network and the nursing evaluation embedding mapping network when the repeated debugging and optimization are carried out.
In the embodiment of the present application, the steps 101 to 104 may refer to the steps 100 to 400.
105: and respectively carrying out association calculation on the vector characterization of the data block and the vector characterization of the plurality of contrast nursing evaluations deployed in advance to obtain a plurality of first component association coefficients.
The vector representation of the data block and the vector representation of the comparative care evaluation are feature vectors of the corresponding data, and the correlation coefficient can be determined by calculating a common metric result (which can be determined based on calculating a distance between two vectors, such as euclidean distance, the smaller the distance, the larger the common metric result), the larger the common metric result, and the larger the correlation coefficient, by calculating the vector representation of the data block and the vector representation of a plurality of comparative care evaluations deployed in advance. In the embodiment of the application, a common measurement result between the vector characterization of the data block and the vector characterization of each comparative care evaluation is used as a first component association coefficient.
In an alternative embodiment, performing the association calculation on the vector representation of the data block and the vector representations of the plurality of contrast care evaluations deployed in advance respectively may include:
1051: a set of vector characterizations constructed from vector characterizations of a plurality of comparative care evaluations is determined.
1052: and obtaining a number product between the vector representation and the vector representation set of each data block to obtain a plurality of first component association coefficients.
The vector characterization of the multiple comparative care evaluations constructs a vector characterization set, and the correlation coefficient between the target object care data set and the comparative care evaluation is calculated based on the number product similarity, for example, the number product between the vector characterization of each calculated data block and the vector characterization set is calculated, so as to obtain the first component correlation coefficient corresponding to the data block of the target object care data set and each comparative care evaluation. For example, the first component association coefficients for determining the vector representations of data chunks 1, 2, 3, 4 and the vector representation of comparative care evaluation a are 0.21, 0.33, 0.56, 0.71, respectively, and the first component association coefficients for vector representations of comparative care evaluation b are 0.43, 0.65, 0.89, 0.52, respectively.
106: a first correlation coefficient between the core care element and the plurality of comparative care evaluations is determined based on the plurality of first component correlation coefficients.
The core care element consists of a plurality of data blocks, each data block and one contrast care evaluation correspond to a first component association coefficient respectively, the plurality of data blocks of the core care element and one contrast care evaluation correspond to a plurality of first component association coefficients, and the first association coefficients between the core care element and the contrast care evaluation are obtained through means of average calculation or addition of the first component association coefficients of the data blocks.
For example, based on the first component association coefficient, the first association coefficient of the core care element a (consisting of the determination data blocks 1 and 2) and the comparative care evaluation a is (0.21+0.33)/(2=0.27), and the first association coefficient of the core care element a and the comparative care evaluation b is (0.43+0.65)/(2=0.54. Then, the first correlation coefficients of the core care elements of the target subject care data set and the comparative care evaluation a are 0.27, 0.56, 0.71, and the first correlation coefficients of the core care elements of the target subject care data set and the comparative care evaluation 2 are 0.54, 0.89, 0.52.
In an alternative embodiment, determining a first association coefficient between the core care element and the plurality of comparative care evaluations based on the plurality of first component association coefficients may include:
1061: a first distribution of data chunks in a target subject care data set and a second distribution of core care elements in the target subject care data set are obtained.
1062: based on the first distribution instance and the second distribution instance, a plurality of target data chunks included by the core care element are determined.
1063: determining a first preset calculation result of a first component association coefficient corresponding to the plurality of target data blocks, and determining the first preset calculation result as the first association coefficient between the core care element and the comparative care evaluation.
The first distribution situation represents the distribution position of the data block in the target object nursing data set, and the second distribution situation represents the distribution position of the core nursing element in the target object nursing data set. The care data embedding mapping network outputs a first distribution condition of the data blocks in the target object care data set together when outputting the data block vector representation of the target object care data set. And in the process of mining the core nursing elements, outputting the second distribution condition of the core nursing elements together. Based on the two distribution conditions, the data block belonging to the same core care element can be determined, and the core care element and the data block are matched. After the target data blocks included by the core care elements are obtained, the first component association coefficients corresponding to the target data blocks are fused, and the first association coefficients between the core care elements and the comparison care evaluation are obtained. For example, the process of fusing the first component association coefficients corresponding to the target data blocks is to obtain a first preset calculation result of the first component association coefficients, such as averaging or summing.
After step 106, the method provided by the embodiment of the present application may further include 107 and 108, or 109.
107: and determining the weight corresponding to the core care element based on the influence coefficient of the core care element in the target object care data set.
For example, the influence coefficient of the core care elements in the target object care data set is calculated based on the normalized exponential function to obtain probability distribution, so as to obtain the weight of each core care element in the target object care data set.
108: and obtaining a second association coefficient between the target object nursing data set and the comparative nursing evaluation according to the weight corresponding to each core nursing element and the first association coefficient corresponding to the core nursing element.
And on the premise of measuring the influence coefficients (importance) of different core care elements, proper weights are distributed for the vector characterization of the core care elements, and the second association coefficient is more reliable. In other embodiments, the weight calculation is not necessary, and the first correlation coefficient corresponding to the core care element may be summed to obtain the second correlation coefficient between the target object care data set and the comparative care evaluation.
109: from the first correlation coefficients corresponding to the respective core care elements of the target subject care dataset, the largest first correlation coefficient is determined as the second correlation coefficient between the target subject care dataset and the comparative care assessment.
The core nursing element corresponding to the largest first association coefficient is the core nursing element which is most associated with the comparison nursing evaluation and has the highest matching degree, and the core nursing element is determined to be the second association coefficient between the target object nursing data set and the comparison nursing evaluation, so that the second association coefficient can be obtained more efficiently. In other embodiments, the average value of the first association coefficients corresponding to the core care elements in the target object care data set may be further taken, and determined as the second association coefficient.
110: a target care assessment is determined from the plurality of comparative care assessments based on the second correlation coefficient. The target care evaluation is used as an evaluation result of the target subject care data set.
The second correlation coefficient characterizes a degree of matching of the vector characterization of the target subject care dataset and the vector characterization of the comparative care assessment, the greater the second correlation coefficient, the more approximate the target subject care dataset and the comparative care assessment. A plurality of comparative care evaluations preceding the second correlation coefficient may be determined from the second correlation coefficients corresponding to the respective comparative care evaluations as target care evaluations.
Compared with the first embodiment (steps 100-700), in the embodiment (101-110), vector representations of the data blocks and vector representations of a plurality of contrast care evaluations deployed in advance are respectively subjected to association calculation to obtain a plurality of first component association coefficients, the first association coefficients between the core care elements and the plurality of contrast care evaluations are determined based on the plurality of first component association coefficients, the vector representations of the target object care data set are thinned to the data blocks to improve matching capability, and in addition, weighting summation is performed based on the weights corresponding to the core care elements and the first association coefficients, so that not only are the influence coefficients of different core care elements measured, but also reasonable weights are allocated to the vector representations of the core care elements, so that the second association coefficients of the target object care data set and the contrast care evaluations are more reliable, or the largest first association coefficient is determined to be the second association coefficients of the target object care data set and the contrast care evaluations, and the second association coefficients are obtained efficiently.
The embodiment of the application also provides a nursing data embedding mapping network and a debugging process of the nursing evaluation embedding mapping network, which specifically comprise the following steps:
11: based on the vector representation of the first nursing evaluation example, the data block vector representation of the first nursing data example and the annotated nursing evaluation deployed in advance, the basic nursing data embedded mapping network and the real-time nursing data embedded mapping network are debugged for the first time, and the rolling nursing data embedded mapping network and the rolling real-time nursing data embedded mapping network are obtained.
12: and based on the data block vector representation of the second nursing data example and the vector representation of the second nursing evaluation example and the annotated nursing data (marked by marks for example) which are deployed in advance, the basic nursing evaluation embedded mapping network and the real-time nursing evaluation embedded mapping network are debugged for the first time, and the rolling nursing evaluation embedded mapping network and the rolling real-time nursing evaluation embedded mapping network are obtained.
13: and on the basis of the rolling real-time nursing evaluation embedded mapping network and the rolling nursing data embedded mapping network, performing multi-round debugging on the rolling nursing data embedded mapping network until a first error result obtained by debugging is smaller than a first preset error value, and stopping debugging to obtain the nursing data embedded mapping network.
14: and on the basis of the rolling real-time nursing data embedded mapping network and the rolling nursing evaluation embedded mapping network, performing multi-round debugging on the rolling nursing evaluation embedded mapping network until a second error result obtained by debugging is smaller than a second preset error value, and stopping debugging to obtain the nursing evaluation embedded mapping network.
In the above steps, the first care data example and the first care evaluation example are, for example, data summarized from the history data, and when the first debug is performed, the first care data example is one care data example to be corresponding, and the first care evaluation example is a care evaluation example to be adapted to the first care data example. Loading a plurality of first care evaluation examples into a real-time care evaluation embedded mapping network, loading first care data examples into a basic care data embedded mapping network, processing a target care evaluation example with network processing output matched with the first care data examples, and comparing the target care evaluation example with annotated care evaluation to obtain a first error result. And optimizing and adjusting the basic nursing data embedded mapping network and the real-time nursing data embedded mapping network based on the first error result to obtain a rolling nursing data embedded mapping network and a rolling real-time nursing data embedded mapping network. And then, based on the rolling real-time nursing data embedded mapping network after optimization and adjustment, debugging the basic nursing evaluation embedded mapping network to obtain a second error result. And debugging the basic nursing evaluation embedded mapping network and the real-time nursing evaluation embedded mapping network based on the second error result, and debugging the rolling real-time nursing data embedded mapping network based on the debugged and optimized rolling nursing data embedded mapping network until the first error result and the second error result are smaller than the corresponding first preset error value and second preset error value, wherein the network converges, and at the moment, the debugging can be ended, so that the nursing data embedded mapping network and the nursing evaluation embedded mapping network are obtained.
Optionally, loading a plurality of first care evaluation examples into the real-time care evaluation embedding mapping network, saving the first care evaluation examples passing through the real-time care evaluation embedding mapping network into a care evaluation database, loading the first care data examples into the basic care data embedding mapping network, obtaining target care evaluation examples matched with the first care data examples through network processing, and comparing the target care evaluation examples with annotated care evaluation to obtain the first error. The base care data embedding mapping network and the real-time care data embedding mapping network are optimized based on the first error debugging. And then, based on the rolling real-time nursing data embedded mapping network after debugging, debugging the basic nursing evaluation embedded mapping network to obtain a second error. And debugging the basic nursing evaluation embedded mapping network and the real-time nursing evaluation embedded mapping network based on the second error, and debugging the rolling nursing data embedded mapping network based on the rolling real-time nursing data embedded mapping network after the debugging and optimization. And stopping when the first error and the second error are smaller than the corresponding first preset error value and second preset error value, and completing debugging at the moment.
In the process, rolling adjustment is completed through the basic nursing evaluation embedded mapping network, the real-time nursing evaluation embedded mapping network, the basic nursing data embedded mapping network and the real-time nursing data embedded mapping network to obtain the nursing data embedded mapping network and the nursing evaluation embedded mapping network, the nursing data embedded mapping network and the nursing evaluation embedded mapping network are obtained through mixed adjustment according to the output results of the nursing data embedded mapping network and the nursing evaluation embedded mapping network when repeated adjustment and optimization are carried out, the feature vector representation performance of the network is improved through the unsupervised adjustment process of contrast learning, the nursing information matching effect of nursing data and nursing evaluation is further improved, the accuracy is high, and the real-time performance is excellent.
Optionally, step 11 may specifically include:
111: determining a first error result of the current underlying care data embedded mapping network commissioning based on the vector characterization of the first care evaluation example and the data chunked vector characterization of the first care data example, and the previously deployed annotated care evaluation; the vector representation of the first care assessment example is output through the real-time care assessment embedded mapping network, and the data chunked vector representation of the first care data example is output through the base care data embedded mapping network.
112: based on the first error result, network parameter values of the basic nursing data embedded mapping network and the real-time nursing data embedded mapping network are respectively adjusted to obtain a rolling nursing data embedded mapping network and a rolling real-time nursing data embedded mapping network.
In the above steps, the first nursing data example serving as the target object nursing data set is loaded to the basic nursing data embedding mapping network, and the data block vector representation of the first nursing data example is obtained. The underlying care data is embedded into the mapping network as a neural network to be commissioned. In addition, a plurality of first nursing evaluation examples are loaded to a real-time nursing evaluation embedded mapping network, vector characterization of the first nursing evaluation examples is obtained, and the vector characterization of the first nursing evaluation examples is stored in a nursing evaluation database, so that subsequent application is facilitated. Corresponding to the foregoing embodiment, core care elements of the first care data example are acquired, association coefficients of the data blocks and the care evaluation example are determined based on the data block vector representation of the first care data example and the vector representation of the first care evaluation example, then the association coefficients between the core care elements and the first care evaluation example are determined based on the distribution condition of the data blocks in the first care data example, so as to obtain first example association coefficients between the first care data example and a plurality of first care evaluation examples, and finally a target care evaluation example is determined based on the first example association coefficients.
After the target care evaluation example is acquired, the target care evaluation example is compared with the annotated care evaluation which is most similar to the first care data example and realizes the marking, for example, the vector representation of the target care evaluation example is compared with the vector representation of the annotated care evaluation, and the first error result of current debugging is obtained based on the comparison result. And judging whether the first error result is smaller than a first preset error value deployed in advance, if not, adjusting a network parameter of the basic nursing data embedded mapping network based on the first error result to obtain a rolling nursing data embedded mapping network, and adjusting a network parameter of the real-time nursing data embedded mapping network according to the first error result to obtain the rolling real-time nursing data embedded mapping network.
Optionally, step 12 may specifically include:
121: a second error result of the current base care evaluation embedded map network commissioning is determined based on the data chunking vector representation of the second care data instance and the vector representation of the second care evaluation instance, and the previously deployed annotated care data.
The data block vector representation is output through the rolling real-time nursing data embedding mapping network, and the vector representation of the second nursing evaluation example is output through the basic nursing evaluation embedding mapping network.
122: based on the second error result, network parameter values of the basic nursing evaluation embedded mapping network and the real-time nursing evaluation embedded mapping network are respectively adjusted to obtain a rolling nursing evaluation embedded mapping network and a rolling real-time nursing evaluation embedded mapping network.
In the above step, a second error result of the current base care evaluation embedded map network commissioning is determined based on the data chunking vector representation of the second care data instance and the vector representation of the second care evaluation instance, and the previously deployed annotated care data. The data block vector representation is output through the rolling real-time nursing data embedding mapping network, and the vector representation of the second nursing evaluation example is output through the basic nursing evaluation embedding mapping network.
And determining whether the second error result is smaller than a second preset error value deployed in advance, if not, adjusting a network parameter of the basic care evaluation embedded mapping network based on the second error result to obtain a rolling care evaluation embedded mapping network, and simultaneously adjusting a network parameter of the real-time care evaluation embedded mapping network based on a second preset calculation result (a result obtained by moving average calculation) obtained by the second error result to obtain the rolling real-time care evaluation embedded mapping network.
Optionally, step 121 may specifically include:
1211: core care elements of the second care data instance are acquired, and a core care element vector representation of the second care data instance is determined based on the data chunks included by the core care elements.
1212: the core care element vector representations of the plurality of second care data examples are saved to a care database, respectively.
1213: and loading the second nursing evaluation example into a basic nursing evaluation embedded mapping network to be debugged, and obtaining the vector representation of the second nursing evaluation example.
1214: a second example association coefficient between the second care evaluation example and the plurality of second care data examples is determined based on the core care element vector representation of the second care data example and the vector representation of the second care evaluation example.
1215: a target care data instance is determined from a plurality of second care data instances based on the second instance correlation coefficients, and a second error result of the current commissioning is determined based on the target care data instance and the annotated care data that was previously deployed.
The objective of the above process is to determine a second error result of the current basic care evaluation embedded mapping network debug, and correspondingly to the foregoing manner, obtain a core care element of the second care data example, and obtain a vector representation of the core care element based on the data block included in the core care element. And performing association calculation on the vector representation of the core nursing element and the vector representation of the second nursing evaluation example to obtain a second example association coefficient, determining a second nursing data example with the largest second example association coefficient as a target nursing data example, and determining a second error based on the target nursing data example and the annotated nursing data.
In step 1212 of the embodiment of the present application, the core care element vector characterizations of the plurality of second care data examples are respectively stored in the care database, so that the vector characterizations of the core care elements can be stored more surprise, and the preprocessing is performed for later comparison, which is more convenient and efficient.
Optionally, based on the first error result, network parameters of the basic care data embedding mapping network and the real-time care data embedding mapping network are respectively adjusted to obtain a rolling care data embedding mapping network and a rolling real-time care data embedding mapping network, which specifically may include: adjusting network parameter values of the basic nursing data embedded mapping network based on the first error result to obtain a rolling nursing data embedded mapping network; acquiring parameter values P1 (P1 for short and P2 to P4 for follow-up understanding as well) before the rolling nursing data are embedded into the mapping network rolling repetition and parameter values P2 after the rolling repetition; and adjusting network parameters of the real-time nursing data embedded mapping network based on the second preset calculation results of the parameters p1 and the parameters p2 to obtain the rolling real-time nursing data embedded mapping network.
Based on the second error result, respectively adjusting network parameters of the basic care evaluation embedded mapping network and the real-time care evaluation embedded mapping network to obtain a rolling care evaluation embedded mapping network and a rolling real-time care evaluation embedded mapping network, which specifically can include: adjusting network parameter values of the basic nursing evaluation embedded mapping network based on the second error result to obtain a rolling nursing evaluation embedded mapping network; acquiring a parameter p3 before rolling and repeating and a parameter p4 after rolling and repeating of the rolling nursing evaluation embedded mapping network; and adjusting network parameters of the real-time care evaluation embedded mapping network based on the second preset calculation results of the parameter p3 and the parameter p4 to obtain the rolling real-time care evaluation embedded mapping network.
In the embodiment of the application, for the purpose of optimizing learning ability in disturbance supervision, disturbance data is cleaned up when the disturbance data is debugged on the care evaluation embedded mapping network based on a real-time care evaluation embedded mapping network (Momentum Encoder architecture), wherein the network architecture of the real-time care evaluation embedded mapping network and the network architecture of the basic care evaluation embedded mapping network are the same, but the debugging optimization of the real-time care evaluation embedded mapping network is performed through moving average calculation. Correspondingly, disturbance data is cleared on the basis of the real-time care data embedded mapping network to disturbance when the care data embedded mapping network is debugged, wherein the network composition of the real-time care data embedded mapping network is the same as the network composition of the basic care data embedded mapping network, and the debugging optimization of the real-time care data embedded mapping network is performed through moving average calculation. For example, when the network is debugged, the real-time nursing evaluation embedding mapping network and the basic nursing evaluation embedding mapping network are debugged and updated simultaneously, wherein the updating process of the real-time nursing evaluation embedding mapping network and the real-time nursing data embedding mapping network is performed through moving average calculation, and the updating speed of network parameter values is slow, unlike the updating speed of the basic nursing data embedding mapping network is directly updated based on the current loading data, and the connection with the past data is not tight.
Specifically, the real-time care data embedding mapping network completes updating through a weighted first preset calculation result of a parameter p1 before rolling and repeating and a parameter p2 after rolling and repeating of the rolling care data embedding mapping network. The real-time nursing evaluation embedded mapping network completes updating through a weighted first preset calculation result of a parameter p3 before rolling repetition and a parameter p4 after rolling repetition of the rolling nursing evaluation embedded mapping network. Based on the above, once the disturbance data is loaded, a general embedded mapping network is disturbed, and the debugging result is not ideal, but for a real-time embedded mapping network, because the reference constraint is carried out by depending on the past network parameters, the influence is smaller, and the robustness is higher. The two types of embedded mapping networks are used for debugging together, so that the two types of embedded mapping networks complement each other, and the network debugging efficiency is ensured on the premise of improving the parameter results.
Optionally, the saving the core care element vector characterizations of the plurality of second care data examples into the care database respectively may specifically include:
(1) Setting the vector capacity of the core care element vector representation corresponding to each second care data example as a target capacity, complementing the position smaller than the target capacity with a preset value (for example, 0), and determining the preset value as an invalid vector representation.
(2) And encoding the properties of the core care elements in the core care element vector characterization set according to the core care element encoding group.
In the above process, the vector capacity of the core care element vector characterization of each second care data example is set to be the target capacity, in other words, the capacity of the core care element vector characterization of each second care data example is the same, so that data can be conveniently mined from the vector set once. Because the data volume of the second care data instance is not determined, nor is the corresponding core care element, valid vector representations and invalid vector representations (e.g., 0, 1,0 representing invalid, 1 representing valid) in the core care element vector representation set are recorded based on the core care element code set (e.g., a set of various tokens).
For a care database, for example, the number of core care elements is at most 10, if the number of core care elements of the care data set is 7, 7 valid vector characterizations from the real-time care data embedded mapping network are stored in the core care element vector characterizations of the care data set, and the rest positions are complemented.
In other application scenarios, the embodiment of the present application may further provide a process for performing correspondence between the fitting care evaluation and the reference care data set, where the concept of determining the target care evaluation based on the care data set may be similar to the foregoing concept, and specifically includes:
901: a vector characterization of the fit pair of care evaluations is obtained.
902: determining a third correlation coefficient between the vector representation of the proposed pair of care assessments and the vector representations of the plurality of reference care datasets deployed in advance; the vector characterization of the reference care data set is determined based on the vector characterization of the core care elements of the reference care data set and the influence coefficients of the core care elements in the reference care data set.
903: a target care data set is determined from the plurality of reference care data sets based on the third correlation coefficient.
The above procedure can still be based on the commissioned care data embedding mapping network and care assessment embedding mapping network provided above. For example, a vector representation of the fit pair of care evaluations is obtained based on the care evaluation embedding mapping network, and a vector representation of the data chunks of the reference care dataset is obtained based on the care data embedding mapping network. And then based on the association coefficient between the vector representation of the data block and the vector representation of the reference nursing data set, obtaining the association coefficient between the vector representation of the core nursing element and the vector representation of the reference nursing data set, and then combining the influence coefficient of the core nursing element in the reference nursing data set to obtain a third association coefficient between the vector representation of the fitting pair nursing evaluation and the vector representations of the plurality of reference nursing data sets. A target care data set that most closely matches the proposed pair of care evaluations is determined from the reference care data set having the higher third correlation coefficient.
According to the embodiment of the application, the vector representation of the fit pair nursing evaluation is obtained, the third association coefficient between the vector representation of the fit pair nursing evaluation and the vector representations of the plurality of reference nursing data sets deployed in advance is determined, the vector representation of the reference nursing data set is determined based on the vector representation of the core nursing element of the reference nursing data set and the influence coefficient of the core nursing element in the reference nursing data set, and the target nursing data set is determined from the plurality of reference nursing data sets based on the third association coefficient. And associating the nursing data with the nursing evaluation more finely by matching the nursing demand information of the core nursing elements with the content of the nursing evaluation according to the association of the core nursing elements in the reference nursing data set with the nursing evaluation.
According to another aspect of the present application, there is also provided a data processing apparatus, referring to fig. 3, an apparatus 900 includes:
the instruction acquisition module 910 is configured to receive a care evaluation instruction, and acquire number information carried by the care evaluation instruction in response to the care evaluation instruction;
the data retrieving module 920 is configured to retrieve, from the care database, a target object care data set corresponding to the number information, where the target object care data set includes care record data of the target object in the current evaluation period;
An element determination module 930 configured to obtain a core care element included in the target object care data set;
the feature mining module 940 is configured to mine vector characterizations of data blocks included in the target object care data set based on the care data embedding mapping network, and mine vector characterizations of a plurality of comparative care evaluations deployed in advance based on the care evaluation embedding mapping network; the nursing data embedding mapping network and the nursing evaluation embedding mapping network are respectively obtained through mixed debugging according to the output results of the nursing data embedding mapping network and the nursing evaluation embedding mapping network when the repeated debugging optimization is carried out;
an association determination module 950 for determining a first association coefficient between the core care element and each of the comparative care evaluations based on the vector representation of the data chunk and the vector representations of the plurality of comparative care evaluations; and determining a second correlation coefficient between the target subject care dataset and the comparative care assessment based on the first correlation coefficient;
the evaluation determination module 960 is configured to determine a target care evaluation from the plurality of comparative care evaluation based on the second correlation coefficient, the target care evaluation being used as an evaluation result of the target subject care dataset.
According to an embodiment of the present application, there is also provided an electronic device (data processing system), a readable storage medium, and a computer program product.
Referring to fig. 4, which is a block diagram of the electronic device 1000 (i.e., a data processing system) of the present application, the electronic device 1000 includes a computing unit 1001 that can perform various appropriate actions and processes according to a computer program stored in a Read Only Memory (ROM) 1002 or a computer program loaded from a storage unit 1008 into a Random Access Memory (RAM) 1003. In the RAM 1003, various programs and data required for the operation of the electronic apparatus 1000 can also be stored. The computing unit 1001, the ROM 1002, and the RAM 1003 are connected to each other by a bus 1004. An input/output (I/O) interface 1005 is also connected to bus 1004. Various components in the electronic device 1000 are connected to the I/O interface 1005, including: an input unit 1006, an output unit 1007, a storage unit 1008, and a communication unit 1009. The input unit 1006 may be any type of device capable of inputting information to the electronic device 1000, the input unit 1006 may receive input numeric or character information and generate key signal inputs related to user settings and/or function control of the electronic device, and may include, but is not limited to, a mouse, a keyboard, a touch screen, a trackpad, a trackball, a joystick, a microphone, and/or a remote control. The output unit 1007 may be any type of device capable of presenting information and may include, but is not limited to, a display, speakers, video/audio output terminals, vibrators, and/or printers. Storage unit 1008 may include, but is not limited to, magnetic disks, optical disks. Communication unit 1009 allows electronic device 1000 to exchange information/data with other devices via a computer network, such as the internet, and/or various telecommunications networks, and may include, but is not limited to, modems, network cards, infrared communication devices, wireless communication transceivers and/or chipsets, such as bluetooth (TM) devices, 802.11 devices, wiFi devices, wiMax devices, cellular communication devices, and/or the like.
The computing unit 1001 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of computing unit 1001 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various specialized Artificial Intelligence (AI) computing chips, various computing units running machine learning model algorithms, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 1001 performs the various methods and processes described above, such as method 200. For example, in some embodiments, the method 200 may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as the storage unit 1008. In some embodiments, part or all of the computer program may be loaded and/or installed onto the electronic device 1000 via the ROM 1002 and/or the communication unit 1009. One or more of the steps of the method 200 described above may be performed when the computer program is loaded into RAM 1003 and executed by the computing unit 1001. Alternatively, in other embodiments, the computing unit 1001 may be configured to perform the method 200 in any other suitable way (e.g., by means of firmware).
Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuit systems, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), systems On Chip (SOCs), complex Programmable Logic Devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs, the one or more computer programs may be executed and/or interpreted on a programmable system including at least one programmable processor, which may be a special purpose or general-purpose programmable processor, that may receive data and instructions from, and transmit data and instructions to, a storage system, at least one input device, and at least one output device. Program code for carrying out methods of the present application may be written in any combination of one or more programming languages. These program code may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus such that the program code, when executed by the processor or controller, causes the functions/operations specified in the flowchart and/or block diagram to be implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.
In the context of the present application, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and pointing device (e.g., a mouse or trackball) by which a user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic input, speech input, or tactile input.
The systems and techniques described here can be implemented in a computing system that includes a background component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such background, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), wide Area Networks (WANs), and the internet. The computer system may include a client and a server. The client and server are typically remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server may be a cloud server, a server of a distributed system, or a server incorporating a blockchain. It should be appreciated that various forms of the flows shown above may be used to reorder, add, or delete steps. For example, the steps recited in the present application may be performed in parallel, sequentially or in a different order, provided that the desired results of the disclosed embodiments are achieved, and are not limited herein. Although embodiments or examples of the present application have been described with reference to the accompanying drawings, it is to be understood that the above-described methods, systems, and apparatus are merely illustrative embodiments or examples, and that the scope of the present application is not limited by these embodiments or examples but only by the claims following the grant and their equivalents. Various elements of the embodiments or examples may be omitted or replaced with equivalent elements thereof. Furthermore, the steps may be performed in a different order than described in the present application. Further, various elements of the embodiments or examples may be combined in various ways. It is important that as technology evolves, many of the elements described herein may be replaced by equivalent elements that appear after the application.