Nursing scheme recommendation method for child behavior feature recognition

Abstract

The invention discloses a nursing scheme recommendation method for child behavior feature recognition, and mainly relates to the technical field of intelligent recommendation. Monitoring equipment is connected, monitoring stream data is acquired, a double-stream decision module is developed in a nursing recommendation system, interface connection with the monitoring equipment is established, and a three-dimensional corner point processor is introduced to pre-process the monitoring stream data, so that the nursing recommendation system is more intelligent. And obtaining effective stream data, executing child behavior feature recognition based on a double-stream decision module, outputting a behavior feature matrix, assisting a built-in nursing database to carry out matching decision based on the behavior feature matrix, determining a nursing recommendation strategy, and carrying out nursing guidance on a target child. The technical problem that in the prior art, an existing child nursing scheme recommendation technology is difficult to meet personalized and precise requirements is solved. The personalized degree and accuracy of recommendation can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent recommendation, and particularly relates to a method for recommending a nursing plan for identifying children's behavior characteristics. Background Art

[0002] In the prior art, traditional nursing depends on the experience of medical staff and parents. Medical staff handle common problems according to clinical practice, and parents provide daily care based on their parenting experience. With the progress of technology, monitoring devices have also been put into use, and some institutions have also tried to analyze children's movement and diet data to provide nursing suggestions.

[0003] However, there are many problems in the prior art. The data collection is not comprehensive, only limited basic physiological data can be obtained, and it is difficult to master children's behavior performance in different scenarios. There is a lack of personalized analysis, mostly based on group data without considering individual differences. The nursing plan lacks dynamic adjustment and cannot be changed in a timely manner with the change of children's behavior characteristics. The decision-making process is not intelligent enough, and manual judgment is prone to errors and low in efficiency.

[0004] In view of this, the existing technology for recommending children's nursing plans is difficult to meet the personalized and precise requirements, and it is urgent to develop an intelligent recommendation method based on the identification of children's behavior characteristics. Summary of the Invention

[0005] This application provides a method for recommending a nursing plan for identifying children's behavior characteristics, which is used to solve the technical problem that the existing technology for recommending children's nursing plans is difficult to meet the personalized and precise requirements.

[0006] In view of the above problems, this application provides a method for recommending a nursing plan for identifying children's behavior characteristics.

[0007] This application provides a method for recommending a nursing plan for identifying children's behavior characteristics. The method includes: connecting monitoring devices to collect behavior data of a target child to obtain monitoring stream data, where the monitoring devices include wearable devices and sensor arrays; developing a dual-stream decision module in a nursing recommendation system and establishing an interface connection between the dual-stream decision module and the monitoring devices, where dual-stream demarcation is performed in the time domain and the spatial domain; introducing a three-dimensional corner processor to preprocess the monitoring stream data, obtaining effective stream data and performing the identification of children's behavior characteristics based on the dual-stream decision module, outputting a behavior characteristic matrix, assisting an internal nursing database to perform a matching decision based on the behavior characteristic matrix, determining a nursing recommendation strategy, where the preprocessing standard is an effective frame and an effective area within the frame; and providing nursing guidance to the target child according to the nursing recommendation strategy.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] A method for recommending a nursing plan for identifying children's behavior characteristics provided by an embodiment of the present application connects a monitoring device to collect behavior data of a target child, obtains monitoring stream data, develops a two-stream decision-making module in a nursing recommendation system, and establishes an interface connection between the two-stream decision-making module and the monitoring device. By introducing a three-dimensional corner processor, the monitoring stream data is preprocessed to obtain effective stream data and perform the identification of children's behavior characteristics based on the two-stream decision-making module, output a behavior characteristic matrix, assist the built-in nursing database to perform a matching decision based on the behavior characteristic matrix, determine a nursing recommendation strategy, and provide nursing guidance for the target child. It is used to solve the technical problem in the prior art that the existing technology for recommending children's nursing plans is difficult to meet the personalized and precise requirements. The present application connects a monitoring device composed of a wearable device and a sensing array, collects comprehensive monitoring stream data, analyzes it from the time domain and the spatial domain using a two-stream decision-making module, and introduces a three-dimensional corner processor to preprocess the data, thereby realizing precise identification of children's behavior characteristics and determining an effective nursing recommendation strategy in combination with the nursing database, which can effectively improve the personalization degree and accuracy of the recommendation. Description of the Drawings

[0010] Figure 1 FIG. is a schematic flowchart of a method for recommending a nursing plan for identifying children's behavior characteristics provided by the present application;

[0011] Figure 2 FIG. is a schematic flowchart of nursing feedback management in a method for recommending a nursing plan for identifying children's behavior characteristics provided by the present application. Detailed Embodiments

[0012] The present application provides a method for recommending a nursing plan for identifying children's behavior characteristics, connects a monitoring device, obtains monitoring stream data, develops a two-stream decision-making module in a nursing recommendation system, establishes an interface connection with the monitoring device, preprocesses the monitoring stream data by introducing a three-dimensional corner processor, obtains effective stream data and performs the identification of children's behavior characteristics based on the two-stream decision-making module, outputs a behavior characteristic matrix, assists the built-in nursing database to perform a matching decision based on the behavior characteristic matrix, determines a nursing recommendation strategy, and provides nursing guidance for the target child. It is used to solve the technical problem in the prior art that the existing technology for recommending children's nursing plans is difficult to meet the personalized and precise requirements.

[0013] Embodiment: As Figure 1 shown, the present application provides a method for recommending a nursing plan for identifying children's behavior characteristics, and the method includes:

[0014] S1: Connect a monitoring device to collect behavior data of a target child, obtain monitoring stream data, and the monitoring device includes a wearable device and a sensing array.

[0015] In the embodiments of the present application, the target child is the user for whom a nursing plan recommendation decision is to be made, the monitoring device is a hardware device for collecting the behavior data of the target child, and the monitoring device includes a wearable device and a sensor array.

[0016] Among them, the wearable device is a monitoring component worn by the target child and can record the behavior activity status of the child in real time and dynamically. Taking the wearable device in the form of a common smart bracelet as an example, it integrates a variety of high-precision sensors inside for collecting data on various behavior indicators of the target child. For example, acceleration sensing is used to accurately sense the acceleration changes of the target child's body in all directions in three-dimensional space; heart rate sensing is used to monitor the heart rate changes of the child in real time.

[0017] The sensor array is a set of sensor combinations arranged in the activity space of the target child for obtaining the behavior information of the child from the environmental level. The sensor array can be composed of various types of sensors to achieve the collection of different behavior data. Exemplarily, a pressure sensor array can be laid on the ground of the child's bedroom or activity room. When the child walks, stands or sits in this area, the pressure sensor will generate corresponding electrical signals according to the position and pressure magnitude of the child's contact with the ground. By analyzing these electrical signals, the activity trajectory and staying position of the child in the space can be accurately judged, and the activity preference area of the child can be understood.

[0018] In actual operation, the process of connecting the monitoring devices needs to ensure stable communication and accurate data transmission between the devices. Wireless communication technologies such as Bluetooth and Wi-Fi can be used to connect the wearable device and the sensor array to the data acquisition terminal. The monitored stream data is a continuous and real-time updated data stream, which contains various behavior information of the target child in a time series, and each node in the monitored stream data is marked with a collection timestamp, providing a rich data basis for subsequent child behavior feature recognition and nursing plan recommendation.

[0019] In summary, through the collaborative work of the devices, the behavior information of the target child can be comprehensively and accurately captured, providing strong support for subsequent data analysis and decision-making.

[0020] S2: Develop a two-stream decision module in the nursing recommendation system and establish an interface connection between the two-stream decision module and the monitoring device, where the two-stream is defined in the time domain and the spatial domain.

[0021] In the embodiments of the present application, the nursing recommendation system refers to an information system that provides personalized and intelligent suggestions or solutions for nursing practice. By developing a two-stream decision module, that is, processing and analyzing the child behavior data from two different dimensions of the time domain and the spatial domain.

[0022] Among them, the time domain dimension mainly focuses on the changes of data over time, which reflects the dynamic characteristics of children's behavior in the time series. Exemplarily, by recording the activities of children at different time periods in a day, the daily routine of children can be understood. For example, it is found that children wake up frequently during specific time periods at night, which may imply that children have sleep problems and need further attention and care.

[0023] The spatial domain dimension, on the other hand, focuses on the distribution characteristics of data in space, reflecting the performance of children's behavior in space. For example, by counting the staying time and behavior characteristics of children in different areas of the room, the preferences of children for different spaces can be understood.

[0024] Furthermore, an interface connection between the dual-stream decision module and the monitoring device is established. This interface connection is a key channel for data transmission, which ensures that the children's behavior data collected by the monitoring device can be accurately and timely transmitted to the dual-stream decision module for processing. In order to ensure the stability and reliability of the interface connection, a suitable data transmission protocol needs to be adopted.

[0025] Preferably, in order to protect the privacy and data security of children, the interface needs to be encrypted. For example, the SSL / TLS encryption protocol is used.

[0026] After the interface connection is established, the dual-stream decision module will receive the monitoring stream data from the monitoring device. At this time, it is necessary to conduct dual-stream boundary determination in the time domain and the spatial domain, that is, perform parallel analysis from the time domain dimension and the spatial domain dimension, and divide the received monitoring stream data into time domain stream data and spatial domain stream data.

[0027] The following is a feasible construction method for the dual-stream decision module:

[0028] Exemplarily, for time domain stream data, time series analysis methods are usually adopted for processing. Exemplarily, with behavioral characteristic elements, an ARIMA time series method can be constructed. The evolution analysis of each characteristic element under the time series can determine and predict the changing trend of children's behavior, thereby judging the stress state of children's bodies. For spatial domain stream data, spatial data analysis techniques can be used. For example, the GIS technology is used to draw the activity trajectory map of children in the activity space. By analyzing the density and distribution range of the trajectories in the trajectory map, the preferred activity areas of children can be understood.

[0029] Among them, supervised training requires collecting a large amount of labeled training data. These labeled data are samples of known child behavior characteristics and corresponding care plans. For example, multiple child samples are collected, each sample containing behavioral data in the time domain and spatial domain. Using the sample as the input, the marked behavioral characteristics as the output, and the extraction of behavioral characteristics based on the time domain and spatial domain as the training objective, supervise the training until convergence. If the preset accuracy is met, obtain the trained dual-stream decision module, which specifically includes a time-domain decision branch and a spatial-domain decision branch.

[0030] Preferably, the loss function is used to compare the difference between the output result and the behavioral characteristic label. Commonly used loss functions such as the cross-entropy loss function. If the prediction result is quite different from the true label, the value of the loss function will be high. At this time, the module will adjust its own parameters through optimization methods, such as stochastic gradient descent, to reduce the value of the loss function. Continuously repeat the above training process until the difference between the prediction result of the module and the true label is within an acceptable range, that is, the value of the loss function converges to a small value, indicating that the convergence condition is met, so that the dual-stream decision module can accurately identify the behavioral characteristics of children according to the input time-domain and spatial-domain flow data.

[0031] S3: Introduce a three-dimensional corner processor to preprocess the monitored flow data, obtain effective flow data and perform child behavior feature recognition based on the dual-stream decision module, output a behavior feature matrix, assist the built-in care database to perform matching decisions based on the behavior feature matrix, and determine a care recommendation strategy, where the effective frame and the effective area within the frame are used as the preprocessing criteria.

[0032] S4: Provide care guidance to the target child according to the care recommendation strategy.

[0033] In the embodiment of the present application, the three-dimensional corner processor is used to preprocess the monitored flow data. The monitored flow data is continuous data about the behavior of the target child collected by the connected monitoring device. However, these raw data may contain a large amount of noise, redundant information, and data irrelevant to the child behavior characteristics. The role of the three-dimensional corner processor is to remove this useless information and extract the data that is truly valuable for child behavior feature recognition.

[0034] Specifically, the three-dimensional corner processor is a component with special processing capabilities, analyzing and processing data in the two-dimensional space and the time dimension. Among them, corners represent key points where data changes significantly in data processing, and these points often contain important behavioral information. For example, when a child suddenly starts running from a stationary state, obvious corners will appear in the monitored flow data, and the three-dimensional corner processor can accurately capture this corner information.

[0035] During the preprocessing process, the valid frames and the in-frame valid regions are used as the criteria. Specifically, a valid frame refers to the frame data containing valuable behavior information. In the monitored stream data, not every frame data is meaningful for identifying children's behavior characteristics. Some frames may be just background noise or minor, featureless actions of children. Excluding these meaningless frames and only retaining the valid frames can greatly reduce the data processing volume and improve the processing efficiency.

[0036] The in-frame valid region refers to the region containing key behavior information in each valid frame. For example, in the monitored data frame of a child's hand movement, the region where the hand is located is the in-frame valid region, and the surrounding blank regions can be ignored. By determining the in-frame valid region, it is possible to more precisely focus on the key behavior parts of the child.

[0037] After the preprocessing by the three-dimensional corner processor, valid stream data can be obtained, that is, the data containing the real and valuable behavior information of children after screening and processing.

[0038] Next, the valid stream data is input into the dual-stream decision module for identifying children's behavior characteristics. That is, in the time domain, analyze the changing patterns of children's behavior over time, such as the duration and frequency of children's behavior; in the spatial domain, analyze the distribution and characteristics of children's behavior in space, such as the activity range and movement amplitude of children. Through the analysis of the valid stream data in these two dimensions, the dual-stream decision module can extract various behavior characteristics of children.

[0039] Then, the extracted behavior characteristics are integrated and output as a behavior characteristic matrix. The behavior characteristic matrix is a structured data containing various behavior characteristics of children presented in matrix form, which unifies and organizes the behavior characteristics in the time domain and the spatial domain. For example, the rows of the matrix can represent different behavior characteristic categories (such as movement speed, action frequency, activity range, etc.), the columns can represent different time points or spatial regions, and each element in the matrix represents the specific value of the corresponding behavior characteristic at that time point or spatial region.

[0040] Finally, based on the behavior characteristic matrix, assist the built-in nursing database for matching decisions. The nursing database stores a large number of nursing plans formulated for different children's behavior characteristics. The system will compare and match the behavior characteristic matrix with the data in the nursing database to find the nursing plan that best matches the current children's behavior characteristics, thereby determining the nursing recommendation strategy.

[0041] For example, if the behavior characteristic matrix shows that a child has too little activity and poor sleep quality recently, by matching the nursing database, the system may recommend nursing strategies such as increasing outdoor activity time and improving the sleep environment to ensure the healthy growth of the child. Then, according to the nursing recommendation strategy, provide nursing guidance for the target child.

[0042] Further, a three-dimensional corner processor is introduced to preprocess the monitored flow data and obtain valid flow data. Step S3 of this application includes:

[0043] According to the characteristics of the scene data, set the three-dimensional size of the flow data block, where the three-dimensional size includes two-dimensional space and time dimension; according to the three-dimensional size, segment the monitored flow data to determine the segmented flow data blocks; traverse the segmented flow data blocks, perform corner determination between data blocks and between frames within data blocks, and determine the valid flow data.

[0044] In the embodiment of this application, the three-dimensional size of the flow data block is set according to the characteristics of the scene data, such as the change law of the scene data. The three-dimensional size of the flow data block includes two-dimensional space and time dimension based on data frames. The two-dimensional space is used to describe the spatial distribution range of each frame frequency data, and the time dimension is used to measure the time span of the data. For example, it can be set as a time period in minutes.

[0045] Exemplarily, if the monitored scene is the daily activities of children indoors, the activity range is relatively fixed and the behavior changes are relatively slow. Then, the size of the two-dimensional space can be reasonably set according to the actual size of the room, such as one spatial unit per square meter, that is, the three-dimensional size; the time dimension can be set as one data block every 5 minutes, so as to capture the behavior changes of children within a period of time more carefully. If the monitored scene is the sports activities of children on the outdoor playground, the activity range is large and the behavior changes are rapid. The size of the two-dimensional space can be appropriately increased, such as one spatial unit per 10 square meters, and the time dimension can be set as one data block every 1 minute to track the rapid behavior changes of children in a timely manner.

[0046] Next, according to the set three-dimensional size, segment the monitored flow data into independent data blocks with specific spatial and time ranges. Each data block contains the behavior data of children in a specific spatial area within a specific time period, serving as the segmented flow data blocks.

[0047] Furthermore, perform corner determination between data blocks and between frames within data blocks for the segmented flow data blocks to determine the valid flow data. Specifically, corner determination is a key data processing method. Corners refer to the points where significant changes occur in the data, and these points often contain important behavior information. When performing corner determination between data blocks, compare the data differences between adjacent data blocks. If the data change between two adjacent data blocks exceeds a preset threshold, it indicates that within the time and space ranges corresponding to these two data blocks, the behavior of children has changed significantly. The boundary points of this data block may be corners, and the corresponding data block is regarded as a corner data block.

[0048] Further, for the selected corner data blocks, when determining corners between frames within the data blocks, within the corner data blocks, the data differences between adjacent frames are compared. Each frame of data represents the state of the monitoring data at a certain moment. If the data changes significantly between adjacent frames, it also indicates that there are obvious changes in the behavior of children within this data block, and the position of this frame may be the corner frame. For example, within a 5-minute data block, if a child's body posture suddenly changes from standing to squatting in a certain frame, this frame may be a corner.

[0049] By determining corners between data blocks and between frames within data blocks, the data blocks and frames containing corners are selected, and combining these data determines the effective flow data. These effective flow data contain the key information of significant changes in children's behavior, which is of great significance for subsequent identification of children's behavior characteristics and recommendation of nursing plans.

[0050] Further, perform corner determination between data blocks and between frames within data blocks to determine the effective flow data. Step S3 of this application includes:

[0051] Traverse the segmented flow data blocks, perform tangential screening based on behavior correlation, and perform axial screening based on behavior variables to determine corner data blocks, where the time series direction is the axial direction; traverse the corner data blocks, and perform frame screening based on the behavior variables between frames within the data blocks to determine the effective flow data; where the preset correlation and preset behavior variables are used as the screening criteria.

[0052] In the embodiment of this application, for all segmented flow data blocks, tangential screening is performed based on behavior correlation. Behavior correlation refers to the degree of correlation between the behavior data of different data blocks. Tangential screening examines the behavior correlation between data blocks from a horizontal perspective, that is, at the spatial level. The preset correlation is a standard value preset to measure the degree of behavior correlation.

[0053] Further, axial screening is performed based on behavior variables. Here, the time series direction is the axial direction, that is, the behavior variables of neighboring data blocks under the time series. Among them, behavior variables are various parameters describing children's behavior characteristics, such as movement speed, movement direction, body posture, etc. The preset behavior variables are some key behavior variables and their value ranges preset in advance.

[0054] Through this axial screening, it is possible to further focus on the data blocks containing key behavior information. Through tangential screening of behavior correlation and axial screening of behavior variables, corner data blocks are finally determined. Corner data blocks refer to those data blocks that show significant changes in terms of behavior correlation and behavior variables, and these data blocks often contain the key information of major changes in children's behavior.

[0055] Next, within each corner point data block, frame screening is performed using the inter-frame behavior variable within the data block. Each corner point data block contains multiple frame data, and each frame data represents the state of the monitoring data at a certain moment. The inter-frame behavior variable refers to the change in the behavior variable between adjacent frames. Similarly, using the preset behavior variable as the screening criterion, the change in the behavior variable between adjacent frames is compared. For example, if the preset change range of the child's body posture is within a certain range, when the change range of the child's body posture between adjacent frames exceeds this preset range, it indicates that there has been a significant change in the child's behavior between these two frames, and the data where these two frames are located may be key data. By screening the frames within each corner point data block, those frames with insignificant changes in the behavior variable are excluded, and finally the valid frames within the valid data block are determined as the said valid flow data.

[0056] In summary, the said valid flow data is data that has undergone layer-by-layer screening and contains key information on significant changes in children's behavior. These data are of great value for accurately identifying children's behavior characteristics and formulating targeted nursing plans subsequently.

[0057] Furthermore, obtain the valid flow data and perform the recognition of children's behavior characteristics based on the two-stream decision module, output the behavior feature matrix, assist the built-in nursing database to perform matching decisions based on the said behavior feature matrix, and determine the nursing strategy. Step S3 of this application includes:

[0058] Import the said valid flow data into the two-stream decision module, perform time-domain behavior feature extraction and spatial-domain behavior feature extraction in parallel, and jointly generate the behavior feature matrix; according to the said behavior feature matrix, traverse the nursing database to perform nursing matching decisions and determine the said nursing recommendation strategy.

[0059] In the embodiment of this application, the valid flow data after corner point processing is imported into the two-stream decision module, and the data is further processed and analyzed in parallel from two dimensions of time domain and spatial domain.

[0060] Specifically, time-domain behavior feature extraction focuses on the change law of data over time and reflects the dynamic characteristics of children's behavior in the time series.

[0061] Preferably, when performing time-domain behavior feature extraction, time series analysis methods such as autoregressive integrated moving average can be used to extract features such as the periodicity and trend of behavior.

[0062] Spatial-domain behavior feature extraction focuses on the distribution of data in space and reflects the characteristics of children's behavior in the spatial dimension. For example, by counting the staying time and activity range of children in different rooms, their preferences for different spaces can be understood.

[0063] Preferably, when extracting spatial behavior features, spatial data analysis methods can be used to visualize and analyze the activity trajectories of children in space, and to extract features such as concentrated areas of activity and moving distances.

[0064] After completing the extraction of temporal and spatial behavioral features, a behavioral feature matrix is ​​jointly generated, that is, structured data containing multiple behavioral features of children in the form of a matrix, which unifies the behavioral features of the temporal and spatial domains. For example, the rows of the matrix can represent different behavioral feature categories (such as movement speed, movement frequency, activity range, etc.), the columns can represent different time points or spatial regions, and each element in the matrix represents the specific value of the corresponding behavioral feature at that time point or spatial region. In this way, the behavioral feature matrix can comprehensively and intuitively display the behavioral characteristics of children.

[0065] Next, based on the generated behavior feature matrix, the system traverses the care database to make care matching decisions. The care database is a database that stores a large number of care plans developed for different children's behavior features. When making matching decisions, the behavior feature matrix is ​​compared with the data in the care database one by one. For example, if the behavior feature matrix shows that the child has too little activity and poor sleep quality recently, the system will search the care database for a care plan that matches this combination of behavior features.

[0066] By comparing the features in the behavior feature matrix with the behavior feature descriptions corresponding to each care plan in the database, the care plan with the highest similarity and the most consistent user behavior is found.

[0067] Ultimately, the recommended nursing strategy is determined based on the matching results, that is, a personalized nursing plan tailored to the specific behavioral characteristics of the current child, which can provide caregivers with accurate and effective nursing guidance.

[0068] Furthermore, step S3 of the present application includes: the dual-stream decision module includes a time domain decision branch and a spatial domain decision branch; traversing the valid stream data, performing stream data attribution based on the time domain and the spatial domain, and determining the time domain stream data and the spatial domain stream data; importing the time domain stream data into the time domain decision branch to extract time domain features; importing the spatial domain stream data into the spatial domain decision branch to extract spatial domain features; interactively integrating the time domain features and the spatial domain features to determine the behavior feature matrix.

[0069] In the embodiments of the present application, the dual-stream decision-making module includes a time-domain decision-making branch and a space-domain decision-making branch. The time-domain decision-making branch is mainly responsible for processing and analyzing behavior data related to time, and mining the changing rules and characteristics of children's behaviors in the time dimension; the space-domain decision-making branch focuses on processing and analyzing behavior data related to space, and exploring the distribution and characteristics of children's behaviors in the space dimension. These two branches are independent of each other and work together to complete a comprehensive analysis of children's behavior characteristics.

[0070] Specifically, traverse the valid flow data, and classify the valid flow data into two categories, namely the time domain and the space domain, according to their characteristics, so as to determine the time-domain flow data and the space-domain flow data.

[0071] Among them, the time-domain flow data refers to the data that mainly reflects the changes in children's behaviors over time, that is, the data that presents different states over time. The space-domain flow data is the data that reflects the distribution of children's behaviors in space, that is, the data related to spatial positions is the space-domain flow data.

[0072] Further, import the time-domain flow data into the time-domain decision-making branch, and extract time-domain characteristics such as duration, frequency change, periodicity, etc. based on behavior characteristics according to the trained time-domain decision-making branch.

[0073] At the same time, import the space-domain flow data into the space-domain decision-making branch for space-domain feature extraction. Exemplarily, according to spatial cluster analysis, cluster the activity positions of children in space to find out the hot spots and cold spots of their activities; or use spatial distance analysis to calculate the moving distances between different spatial positions of children to understand their activity ranges and activity paths. Extract space-domain characteristics such as the concentrated area of activity behaviors, the uniformity of spatial distribution, the moving distance and direction.

[0074] Finally, interactively integrate the time-domain characteristics and the space-domain characteristics, and arrange and combine the extracted time-domain characteristics and space-domain characteristics according to certain rules. For example, the rows of the matrix can represent different types of behavior characteristics, including the behavior duration, frequency in the time-domain characteristics, and the activity concentrated area, moving distance, etc. in the space-domain characteristics; the columns of the matrix can represent different time points or spatial regions. In this way, the characteristic information in the time domain and the space domain is unified into a matrix, forming a behavior characteristic matrix that comprehensively and accurately reflects children's behavior characteristics, providing a solid data basis for subsequent matching and recommendation of nursing plans based on behavior characteristics.

[0075] Further, construct a nursing database. Step S3 of the present application includes:

[0076] Call the nursing records, cluster them based on age stages, compare and determine the personalized nursing needs, where the personalized nursing needs are the distinctive nursing needs in the children's stage, and the children's stage is subdivided into multiple sub-stages; cluster the nursing records by nursing methods, and integrate and determine N nursing standards; construct a nursing database according to the personalized nursing needs and the N nursing standards.

[0077] In the embodiment of the present application, the nursing records are detailed records of the children's nursing process and related information, including data in many aspects such as the children's health status, behavior performance, nursing measures and effects. These records come from multiple channels such as medical databases, community health service centers, and home nursing logs.

[0078] Next, cluster the nursing records based on age stages. Age stage is an important factor affecting the physical and mental development of children. Children in different age stages have different growth and development characteristics and nursing needs. The children's stage is subdivided into multiple sub-stages, such as infancy (0 - 1 year old), toddlerhood (1 - 3 years old), preschool age (3 - 6 years old), school age (6 - 12 years old), etc. When clustering the nursing records, group the children's nursing records belonging to the same sub-stage into one category.

[0079] After completing the clustering based on age stages, compare various nursing records to determine the personalized nursing needs. The personalized nursing needs refer to the nursing needs that are unique to different children's stages and are different from other stages. By analyzing the nursing records of the same sub-stage in detail, find out the common health problems and nursing key points of children in this stage.

[0080] At the same time, cluster the nursing records by nursing methods. Nursing methods refer to the specific measures taken for different health problems and nursing needs, such as physical therapy, psychological counseling, nutritional intervention, etc. Group the nursing records that use the same or similar nursing methods into one category, and count N nursing standards corresponding to each nursing method. Nursing standards are the standardized and normalized descriptions of various nursing methods, which clarify the operation procedures, application scopes, precautions, etc. of nursing. The determined N nursing standards cover the nursing methods for various common children's health problems, providing a scientific basis for the subsequent formulation of nursing plans.

[0081] Finally, construct a nursing database according to the personalized nursing needs and the N nursing standards. The nursing database is a structured information library that integrates and stores the personalized nursing needs and nursing standards. In the database, it can be classified and indexed according to age stages and nursing methods, which is convenient for quick query and matching. Exemplarily, when formulating a nursing plan for a 5-year-old child, first determine that he is in the preschool age according to his age, find the personalized nursing needs of this stage, and then select appropriate nursing measures from the nursing standards in combination with specific health problems.

[0082] In this way, the nursing database can provide comprehensive and accurate information support for child care, and realize the recommendation of personalized and scientific nursing plans.

[0083] Furthermore, as Figure 2 shown, there is also step S5 in this application. After providing nursing guidance to the target child, step S5 of this application includes:

[0084] According to the nursing recommendation strategy, determine the expected nursing energy efficiency; according to the expected nursing energy efficiency, construct a standard nursing portrait, where the standard nursing portrait includes the entire nursing cycle; interact with the target child's behavior habits, conduct bias deduction based on the standard nursing portrait, and construct a multiple nursing bias portrait; establish a mapping between the standard nursing portrait and the multiple nursing bias portraits, and conduct nursing feedback management.

[0085] In the embodiment of this application, the expected nursing energy efficiency is determined according to the nursing recommendation strategy. The nursing recommendation strategy is obtained by matching the behavior feature matrix of the child with the nursing database, and includes a series of nursing measures for the specific situation of this child. The expected nursing energy efficiency refers to the nursing effects and goals that are expected to be achieved after implementing these nursing measures. Preferably, by determining reasonable expected nursing energy efficiency indicators, a basis is provided for subsequent nursing evaluations.

[0086] Next, a standard nursing portrait is constructed according to the expected nursing energy efficiency. The standard nursing portrait is a set of labels that comprehensively describe the nursing process and expected effects, and includes information on the entire nursing cycle. The entire nursing cycle refers to the entire time period from the start of implementing nursing measures to achieving the expected nursing effects, including each stage of nursing, the goals and tasks of each stage, and the expected changes in effects, etc.

[0087] For example, for a child with sleep problems, the standard nursing portrait may detail the specific steps to adjust the sleep schedule in the first week and the expected improvement in the sleep onset speed, the specific measures to improve the sleep environment in the second week and the expected increase in the sleep duration, etc. By constructing the standard nursing portrait, the overall picture of the nursing process and the expected results can be clearly shown, providing clear guidance for nursing staff.

[0088] Then, interact with the target child's behavior habits, conduct bias deduction based on the standard nursing portrait, and construct a multiple nursing bias portrait. The behavior habits of the target child refer to the unique behavior patterns and preferences shown by this child in daily life, and these behavior habits may affect the nursing effects. Bias deduction refers to adjusting and modifying the standard nursing portrait considering the differences in the behavior habits of children.

[0089] For example, if the target child has the habit of using electronic devices before going to bed, this may affect the falling asleep speed and deviate from the expected effect in the standard care portrait. By analyzing the impact of this behavior habit on the care effect, adjustments are made based on the standard care portrait to construct multiple care bias portraits that may exist for the target child. The multiple care bias portraits reflect the expected deviation effect in the care process considering the individual differences of children.

[0090] Exemplarily, based on the behavior habit, big data mining and statistics are carried out to determine the care impact direction and degree of influence, and the standard care portrait is adjusted for bias to be used as the care bias portrait. Among them, the multiple care bias portraits are the bias energy efficiency based on one or more behavior habits of the target child and represent the actual state that may exist in the care process of the target child.

[0091] Finally, a mapping between the standard care portrait and the multiple care bias portraits is established for care feedback management. Mapping means associating and corresponding each element in the standard care portrait with the corresponding element in the multiple care bias portraits for comparison and analysis during the care process. At the same time, the behavior habit causing the bias is used for mapping identification.

[0092] By establishing this mapping relationship, the changes in the care process and expected effect due to the differences in children's behavior habits can be clearly seen. Care feedback management refers to adjusting the care strategy in a timely manner according to the actual care situation and the comparison results with the standard care portrait and the multiple care bias portraits.

[0093] Through continuous care feedback management, the care plan can be continuously optimized, the care effect can be improved, and it can be ensured that children receive the most suitable care services.

[0094] Furthermore, in establishing the mapping between the standard care portrait and the multiple care bias portraits, step S5 of the present application includes:

[0095] Based on the standard care portrait and the first care bias portrait, determine the first feedback recommendation strategy; based on the standard care portrait and the Mth care bias portrait, determine the Mth feedback recommendation strategy; according to the first feedback recommendation strategy to the Mth feedback recommendation strategy, perform portrait mapping identification.

[0096] In the embodiment of the present application, the standard care portrait contains information on the entire care cycle and is an ideal standard constructed based on the expected care energy efficiency, providing a comprehensive and standardized reference framework for the care process. The care bias portrait is a portrait obtained by performing bias deduction on the standard care portrait considering individual differences such as the unique behavior habits of the target child, and represents the care response that the target child may cause based on personal habits in actual care.

[0097] The first nursing bias portrait is a random one among M nursing bias portraits and the last one among the M nursing bias portraits, which respectively represent specific nursing portraits obtained by considering the individual differences of children in different situations. M is an integer greater than 1. Different bias portraits reflect the representations of different personalized habits of the target child.

[0098] Based on the standard nursing portrait and the first nursing bias portrait, determine the first feedback recommendation strategy. When comparing the standard nursing portrait and the first nursing bias portrait, some differences will be found. These differences are caused by factors such as the individual behavior habits of the target child. By deeply analyzing these differences, the areas that need to be adjusted and optimized in the nursing process can be identified.

[0099] For example, if it is expected in the standard nursing portrait that the child can fall asleep quickly after 30 minutes of relaxation activities before going to bed, but the first nursing bias portrait shows that the child still takes a long time to fall asleep even after relaxation activities. After analysis, it may be found that the child has a special preference for the way of relaxation activities. Then the first feedback recommendation strategy may be to adjust the way of relaxation activities. Through this comparison and analysis, considering various factors comprehensively, the first feedback recommendation strategy suitable for this situation is finally determined. This strategy aims to make up for the deviation of nursing effect caused by individual differences and make the nursing process more in line with the actual needs of children.

[0100] Similarly, analyze each nursing bias portrait in turn until the analysis based on the standard nursing portrait and the Mth nursing bias portrait is completed, and determine the Mth feedback recommendation strategy.

[0101] Finally, according to the first feedback recommendation strategy to the Mth feedback recommendation strategy, perform portrait mapping identification. Portrait mapping identification is the process of associating and marking each feedback recommendation strategy with the corresponding nursing bias portrait. Through this mapping identification, the corresponding relationship between different individual difference situations (reflected by the nursing bias portrait) and the corresponding adjustment strategies (feedback recommendation strategies) can be clearly seen.

[0102] So that in the actual nursing process of the target child, when encountering similar individual difference situations, the corresponding feedback recommendation strategy can be quickly referred to, providing accurate and convenient decision-making basis for the nursing staff, improving the implementation efficiency and accuracy of the nursing plan, and better meeting the personalized nursing needs of children.

[0103] Furthermore, step S5 of the present application includes: executing the nursing recommendation strategy, and combining the monitoring device to track the nursing energy efficiency, performing nursing energy efficiency bias determination and feedback recommendation strategy matching based on the mapping of the standard nursing portrait and the multiple nursing bias portraits, and providing nursing feedback guidance for the target child.

[0104] Specifically, implement and carry out the nursing recommendation strategy to guide the nursing process of the target child. While implementing the nursing recommendation strategy, combine it with monitoring devices for nursing energy efficiency tracking. The monitoring devices, namely the aforementioned wearable device and the sensing array, can collect the behavioral data of the target child in real time. Through these devices, various physiological and behavioral information of the child during the implementation of the nursing recommendation strategy can be obtained, and the implementation effect of the nursing recommendation strategy, that is, the nursing energy efficiency, can be intuitively understood.

[0105] For example, if it is found that the child's sleep time gradually shortens and the sleep duration increases after implementing the sleep nursing strategy, it indicates that the nursing strategy has played a positive role; on the contrary, if the data does not improve significantly or even deteriorates, the reasons need to be further analyzed.

[0106] Next, perform nursing energy efficiency bias determination based on the mapping of the standard nursing portrait and the multiple nursing bias portraits. The standard nursing portrait is constructed based on the expected nursing energy efficiency, which describes the goals and effects that should be achieved at each stage of the nursing process under ideal circumstances. The multiple nursing bias portraits are obtained by adjusting the standard nursing portrait considering factors such as the individual behavioral habits of the target child. By comparing the actual nursing energy efficiency data with the standard nursing portrait and the multiple nursing bias portraits, it is possible to determine whether there is a nursing energy efficiency bias.

[0107] After determining that there is a nursing energy efficiency bias, perform feedback recommendation strategy matching. That is, according to the situation of the nursing energy efficiency bias, match and determine the current nursing bias portrait. Based on the portrait mapping, identify the corresponding strategy as the feedback recommendation strategy.

[0108] Finally, provide nursing feedback guidance to the target child. Convey the matched feedback recommendation strategy to the nursing staff or parents and guide them to correct the nursing process. At the same time, continuously pay attention to the changes in the nursing energy efficiency of the child after implementing the new strategy, and make further adjustments and optimizations according to the actual situation, forming a closed-loop nursing feedback mechanism to ensure that the nursing plan can meet the needs of the target child to the greatest extent and promote the healthy growth of the child.

[0109] A nursing plan recommendation method for identifying children's behavioral characteristics provided by this application has the following technical effects:

[0110] 1. Collect the behavioral data of the target child using monitoring equipment, introduce a three-dimensional corner processor, preprocess the monitored stream data with effective frames and the effective area within the frame as the criteria, set the three-dimensional size of the stream data block according to the characteristics of the scene data, divide it including two-dimensional space and time dimensions, determine the corner data block through tangential screening of behavioral correlation and axial screening of behavioral variables, and then perform frame screening on the inter-frame behavioral variables within the data block to obtain effective stream data, which can effectively improve the data quality, focus on key behavioral information, and provide a high-quality data basis for subsequent accurate behavioral feature recognition.

[0111] 2. The auxiliary dual-stream decision module, including a time-domain decision branch and a space-domain decision branch, performs time-domain and space-domain stream data attribution and parallel decision on the effective stream data to determine the behavioral feature matrix. Analyze the child's behavioral data comprehensively from two dimensions of time and space, mine the dynamic change rules and spatial distribution characteristics of behaviors, and the comprehensively generated behavioral feature matrix can more accurately and comprehensively reflect the child's behavioral characteristics, providing a strong basis for nursing plan recommendation.

[0112] 3. Cluster to determine personalized nursing needs based on age stages. The child stage is divided into multiple sub-stages, and N nursing standards are determined by clustering and integrating nursing methods. Combine the two to construct a nursing database. Fully consider the growth and development characteristics and nursing need differences of children in different age stages, as well as the standardization of common nursing methods.

[0113] 4. Construct a standard nursing portrait according to the expected nursing energy efficiency, construct multiple nursing bias portraits considering the target child's behavior habits, and establish the mapping relationship between the two. Determine the feedback recommendation strategy based on the standard nursing portrait and different nursing bias portraits and perform portrait mapping identification. Through the portrait mapping identification, it is convenient to quickly search for and apply feedback recommendation strategies suitable for different situations, form a closed-loop nursing feedback mechanism, and improve the nursing decision-making efficiency and personalization degree.

[0114] Through the foregoing detailed description of a method for recommending a nursing plan for child behavior feature recognition in this specification, those skilled in the art can clearly know a method for recommending a nursing plan for child behavior feature recognition in this embodiment. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0115] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for recommending a nursing plan for identifying children's behavioral characteristics, characterized in that, The method includes: Connect a monitoring device to collect behavioral data of a target child and obtain monitoring stream data. The monitoring device includes a wearable device and a sensing array; Develop a two-stream decision module in the nursing recommendation system and establish an interface connection between the two-stream decision module and the monitoring device, where two-stream definition is performed in the time domain and the spatial domain; Introduce a three-dimensional corner processor to preprocess the monitoring stream data, obtain effective stream data, perform recognition of children's behavioral characteristics based on the two-stream decision module, output a behavioral characteristic matrix, and assist the built-in nursing database to make a matching decision based on the behavioral characteristic matrix to determine a nursing recommendation strategy, where the effective frame and the effective area within the frame are used as the preprocessing criteria; According to the nursing recommendation strategy, provide nursing guidance to the target child.

2. The method for recommending a nursing plan for identifying children's behavioral characteristics according to claim 1, characterized in that, Introduce a three-dimensional corner processor to preprocess the monitoring stream data and obtain effective stream data, including: Set the three-dimensional size of the stream data block according to the characteristics of the scene data, where it includes two-dimensional space and time dimension; Segment the monitoring stream data according to the three-dimensional size to determine segmented stream data blocks; Traverse the segmented stream data blocks, perform corner determination between data blocks and between frames within data blocks to determine the effective stream data.

3. The nursing plan recommendation method for identifying children's behavioral characteristics according to claim 2, characterized in that, Perform corner determination between data blocks and between frames within data blocks to determine the effective stream data, including: Traverse the segmented stream data blocks, perform tangential screening based on behavioral correlation and axial screening based on behavioral variables to determine corner data blocks, where the time series direction is the axial direction; Traverse the corner data blocks, perform frame screening based on the inter-frame behavioral variables within the data blocks to determine the effective stream data; Among them, the preset correlation and the preset behavioral variables are used as the screening criteria.

4. The nursing plan recommendation method for identifying children's behavioral characteristics according to claim 1, wherein, Obtain effective stream data and perform recognition of children's behavioral characteristics based on the two-stream decision module, output a behavioral characteristic matrix, and assist the built-in nursing database to make a matching decision based on the behavioral characteristic matrix to determine a nursing strategy, including: Import the effective stream data into the two-stream decision module, perform time-domain behavioral feature extraction and spatial-domain behavioral feature extraction in parallel, and jointly generate a behavioral characteristic matrix; According to the behavioral characteristic matrix, traverse the nursing database to make a nursing matching decision to determine the nursing recommendation strategy.

5. The method for recommending a nursing plan for identifying children's behavioral characteristics according to claim 4, wherein The two-stream decision module includes a time-domain decision branch and a spatial-domain decision branch; Traverse the effective stream data, perform attribution of the stream data in the time domain and the spatial domain to determine time-domain stream data and spatial-domain stream data; Import the time-domain stream data into the time-domain decision branch to extract time-domain features; Import the spatial-domain stream data into the spatial-domain decision branch to extract spatial-domain features; Interactively integrate the time-domain features and the spatial-domain features to determine the behavioral characteristic matrix.

6. The method for recommending a nursing plan for identifying children's behavior characteristics according to claim 4, wherein, Construct a nursing database, including: Call nursing records, perform clustering based on age stages, compare and determine personalized nursing needs, where the personalized nursing needs are the differentiated nursing needs in children's stages, and the children's stages are subdivided into multiple sub-stages; Cluster the nursing records by nursing methods and integrate to determine N nursing standards; Construct a nursing database according to the personalized nursing needs and the N nursing standards.

7. The nursing plan recommendation method for identifying children's behavioral characteristics according to claim 1, characterized in that, After providing nursing guidance to the target child, including: Determine the expected nursing energy efficiency according to the nursing recommendation strategy; Construct a standard nursing portrait according to the expected nursing energy efficiency, where the standard nursing portrait includes the entire nursing cycle; Interact with the behavior habits of the target child, conduct a bias deduction based on the standard nursing portrait, and construct multiple nursing bias portraits; Establish a mapping between the standard nursing portrait and the multiple nursing bias portraits, and conduct nursing feedback management.

8. The method for recommending a nursing plan for identifying children's behavior characteristics according to claim 7, wherein, Establishing the mapping between the standard nursing portrait and the multiple nursing bias portraits includes: Determine the first feedback recommendation strategy based on the standard nursing portrait and the first nursing bias portrait; Determine the Mth feedback recommendation strategy based on the standard nursing portrait and the Mth nursing bias portrait; Perform portrait mapping identification according to the first feedback recommendation strategy to the Mth feedback recommendation strategy.

9. The method for recommending a nursing plan for identifying children's behavioral characteristics according to claim 8, wherein, Execute the nursing recommendation strategy, track the nursing energy efficiency in combination with the monitoring device, and match the nursing energy efficiency bias determination and the feedback recommendation strategy with the mapping between the standard nursing portrait and the multiple nursing bias portraits to provide nursing feedback guidance to the target child.

Images