Intelligent rehabilitation log text processing method and system using artificial intelligence
By obtaining basic sample rehabilitation log text from cloud database and combining rehabilitation activities and environmental factors for simulation analysis, we generate derivative sample rehabilitation log text, optimize the rehabilitation log text mining network, and solve the inefficient processing efficiency of rehabilitation log text in the existing technology and The lack of objectivity improves the accuracy and performance of the analysis.
Patent Information
- Application Number
- CN202410590388.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-13
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2044-05-13
AI Technical Summary
The existing text processing methods for rehabilitation logs are inefficient, lack objectivity and consistency, making it difficult to accurately extract and analyze key information in rehabilitation logs, and neglect environmental factors, resulting in insufficient comprehensiveness and accuracy of rehabilitation process analysis.
By obtaining basic sample rehabilitation log text from the cloud database, combining rehabilitation activity characteristic parameters and environmental impact factors to simulate it in the local database, generate text data of patient activity status and environmental impact, perform correlation analysis, and generate derivative sample rehabilitation logs Text and optimize rehabilitation log text mining network.
It improves the authenticity and usability of rehabilitation log text, enhances the diversity and representativeness of data, improves the accuracy and performance of rehabilitation log text processing and analysis, and provides more valuable rehabilitation information and suggestions.
Smart Images

Figure CN118280590B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of artificial intelligence technology, and more specifically, to an intelligent rehabilitation log text processing method and system using artificial intelligence. Background Art
[0002] In the field of rehabilitation medicine, rehabilitation diaries are a key tool for recording the patient's rehabilitation process and results. Their text content records important information such as the patient's rehabilitation activities, progress, and reactions. This information is extremely valuable for doctors to evaluate rehabilitation effects, adjust rehabilitation plans, and conduct scientific research.
[0003] However, in the current practice of processing rehabilitation log texts, traditional manual analysis methods have obvious problems. First, manual processing of large amounts of rehabilitation log texts is not only inefficient, but also easily limited by the subjective experience and knowledge level of the processor, making it difficult to ensure the objectivity and consistency of the analysis. Second, rehabilitation log texts often contain complex medical terms and professional rehabilitation descriptions, which are difficult for non-professionals to accurately interpret, further increasing the difficulty of manual analysis.
[0004] In recent years, although computer technology has been increasingly used in the medical field, there are still some challenges in processing rehabilitation log text. On the one hand, existing text processing systems often lack the ability to deeply understand and apply professional knowledge in the rehabilitation field, making it difficult to accurately extract and analyze key information in rehabilitation logs. On the other hand, due to the diversity and complexity of rehabilitation log text, general text processing methods often have difficulty adapting to changes in different scenarios and needs.
[0005] In addition, the rehabilitation environment has an important impact on the patient's rehabilitation effect, but in the traditional rehabilitation log text processing, environmental factors are often ignored or simplified. This leads to insufficient comprehensive and accurate analysis of the rehabilitation process, making it difficult to provide effective decision support for doctors. Summary of the invention
[0006] In order to at least overcome the above-mentioned deficiencies in the prior art, the purpose of the present application is to provide an intelligent rehabilitation log text processing method and system using artificial intelligence.
[0007] In a first aspect, the present application provides an intelligent rehabilitation log text processing method using artificial intelligence, which is applied to an intelligent rehabilitation guidance system, and the method includes:
[0008] Obtaining a basic sample rehabilitation log text from a cloud rehabilitation database, and adding the basic sample rehabilitation log text to a local rehabilitation database;
[0009] Acquire the initialized rehabilitation activity characteristic parameters and environmental influence factors, and simulate in the local rehabilitation database according to the rehabilitation activity characteristic parameters and the environmental influence factors to generate first rehabilitation behavior text data corresponding to the patient activity state area and second rehabilitation behavior text data corresponding to the environmental effect area;
[0010] Performing correlation analysis on the basic sample rehabilitation log text and the first rehabilitation behavior text data in the local rehabilitation database to generate first sample simulation text data, and acquiring environment change data, performing correlation analysis on the environment change data and the second rehabilitation behavior text data in the local rehabilitation database to generate second sample simulation text data;
[0011] In the local rehabilitation database, based on the first sample simulated text data and the second sample simulated text data, a derived sample rehabilitation log text is generated for the basic sample rehabilitation log text, and based on the basic sample rehabilitation log text and the derived sample rehabilitation log text, network parameters of the initialized rehabilitation log text mining network are optimized to generate an optimized rehabilitation log text mining network.
[0012] In a possible implementation of the first aspect, the rehabilitation activity characteristic parameters include spatial configuration information of a rehabilitation simulation scene, recovery process information, environmental quietness information, rehabilitation tracking device information, and rehabilitation patient personalized characteristic information, and the simulation is performed in the local rehabilitation database based on the rehabilitation activity characteristic parameters and the environmental influencing factors to generate first rehabilitation behavior text data corresponding to the patient activity state area and second rehabilitation behavior text data corresponding to the environmental action area, including:
[0013] Determining the three-dimensional layout of the rehabilitation simulation scene and the spatiotemporal information of the rehabilitation activities in the local rehabilitation database according to the spatial configuration information and the recovery process information of the rehabilitation simulation scene;
[0014] Determining a reference monitoring point and an optimized monitoring point based on the three-dimensional layout of the rehabilitation simulation scene and the rehabilitation tracking device information;
[0015] Determining the patient's activity state area based on the benchmark monitoring points and the rehabilitation patient's personalized characteristic information;
[0016] Determining an environmental action area based on the three-dimensional layout of the rehabilitation simulation scene and the environmental impact factors;
[0017] Based on the three-dimensional layout of the rehabilitation simulation scene, the optimized monitoring points, the spatiotemporal information of the rehabilitation activity and the patient activity state area, generating first rehabilitation behavior text data corresponding to the patient activity state area;
[0018] Based on the three-dimensional layout of the rehabilitation simulation scene, the optimized monitoring points, the spatiotemporal information of the rehabilitation activity and the environmental action area, second rehabilitation behavior text data corresponding to the environmental action area is generated.
[0019] In a possible implementation of the first aspect, determining the three-dimensional layout of the rehabilitation simulation scene and the spatiotemporal information of the rehabilitation activity in the local rehabilitation database according to the spatial configuration information and the recovery process information of the rehabilitation simulation scene includes:
[0020] Acquire a first horizontal extension parameter and a second horizontal extension parameter in the spatial configuration information of the rehabilitation simulation scene in the local rehabilitation database, wherein the first horizontal extension parameter and the second horizontal extension parameter are horizontal extension parameters in different vertical directions in the plane area where the rehabilitation simulation scene is located;
[0021] Acquire a height extension parameter of a height area in the spatial configuration information of the rehabilitation simulation scene;
[0022] Determining a target rehabilitation training time according to the recovery progress information;
[0023] The spatiotemporal information of the rehabilitation activity is determined according to the first horizontal extension parameter, the second horizontal extension parameter, the height extension parameter and the target rehabilitation training time of the rehabilitation simulation scene.
[0024] In a possible implementation of the first aspect, the rehabilitation tracking device information includes a first sensing range limit parameter, a device limit position parameter, a device label, a device quantity, and a device interval, and determining the reference monitoring point and the optimized monitoring point based on the three-dimensional layout of the rehabilitation simulation scene and the rehabilitation tracking device information includes:
[0025] Determining a reference monitoring point based on the three-dimensional layout of the rehabilitation simulation scene, the first sensing range limit parameter, and the device extreme position parameter;
[0026] Determining an optimized monitoring point based on the benchmark monitoring point, the device label, the number of devices, and the device interval;
[0027] In a possible implementation of the first aspect, the rehabilitation patient personalized characteristic information includes a second sensing range limit parameter, a patient dynamic remote ratio, a patient dynamic remote distance parameter limit information, a rehabilitation patient action proximity distance parameter limit information, and a rehabilitation patient action height limit information, and determining the patient activity state area based on the reference monitoring point and the rehabilitation patient personalized characteristic information includes:
[0028] Generate a first simulation value, and when the patient dynamic remote ratio is less than the first simulation value, determine the distance parameter between the rehabilitation patient and the reference monitoring point based on the patient dynamic remote distance parameter restriction information;
[0029] When the patient's dynamic remote ratio is not less than the first simulation value, the distance parameter between the rehabilitation patient and the reference monitoring point is determined based on the rehabilitation patient's action proximity distance parameter restriction information, the direction angle of the rehabilitation patient relative to the rehabilitation tracking device is randomly generated, and the patient's activity state area is determined according to the reference monitoring point, the distance parameter between the rehabilitation patient and the reference monitoring point, the direction angle, and the rehabilitation patient's action height restriction information;
[0030] In a possible implementation of the first aspect, the patient activity state area includes a horizontal coordinate of the rehabilitation patient, a longitudinal coordinate of the rehabilitation patient, and a vertical coordinate of the rehabilitation patient, and determining the patient activity state area based on the reference monitoring point, a spacing parameter between the rehabilitation patient and the reference monitoring point, the direction angle, and the rehabilitation patient's action height restriction information includes:
[0031] Determine, according to the spacing parameter and direction angle between the rehabilitation patient and the reference monitoring point, a first misalignment spacing parameter between the rehabilitation patient and the reference monitoring point in the transverse axis and a second misalignment spacing parameter between the rehabilitation patient and the reference monitoring point in the longitudinal axis;
[0032] Determining the initial transverse coordinate of the rehabilitation patient according to the transverse coordinate of the reference monitoring point and the first misalignment spacing parameter;
[0033] Determining the initial longitudinal coordinate of the rehabilitation patient according to the longitudinal coordinate of the reference monitoring point and the second offset spacing parameter;
[0034] Determine the vertical coordinates of the rehabilitation patient based on the rehabilitation patient's movement height restriction information; when it is determined that at least one of the initial horizontal coordinate and the initial longitudinal coordinate needs to be adjusted based on the initial horizontal coordinate, the initial longitudinal coordinate and the second sensing range restriction parameter, adjust at least one of the initial horizontal coordinate and the initial longitudinal coordinate that need to be adjusted to generate at least one of the adjusted horizontal coordinate and the adjusted longitudinal coordinate.
[0035] In a possible implementation of the first aspect, the environmental impact factor includes a third sensing range limit parameter and vertical boundary information of an environmental impact source, the environmental impact area includes a lateral coordinate of the environmental impact source, a longitudinal coordinate of the environmental impact source, and a vertical coordinate of the environmental impact source, and determining the environmental impact area based on the three-dimensional layout of the rehabilitation simulation scene and the environmental impact factor includes:
[0036] Determining the lateral coordinates of the environmental impact source according to the lateral extension parameter and the third sensing range limitation parameter in the three-dimensional layout of the rehabilitation simulation scene;
[0037] Determining the longitudinal coordinate of the environmental impact source according to the axial extension parameter and the third sensing range limit parameter in the three-dimensional layout of the rehabilitation simulation scene;
[0038] The vertical coordinates of the environmental impact source are determined based on the vertical definition information of the environmental impact source.
[0039] In a second aspect, an embodiment of the present application further provides an intelligent rehabilitation guidance system, which includes a processor and a machine-readable storage medium, wherein the machine-readable storage medium stores a computer program, and the computer program is loaded and executed according to the processor to implement the intelligent rehabilitation log text processing method using artificial intelligence according to the first aspect above.
[0040] Based on any one of the above aspects, the beneficial effects of this application are as follows:
[0041] By obtaining the basic sample rehabilitation log text from the cloud rehabilitation database and adding it to the local rehabilitation database, the sharing and local storage of rehabilitation log data are realized, which provides a basis for subsequent data processing and mining. Secondly, using the initialized rehabilitation activity characteristic parameters and environmental influencing factors, simulation is performed in the local rehabilitation database to generate the first rehabilitation behavior text data corresponding to the patient's activity state area and the second rehabilitation behavior text data corresponding to the environmental action area, enriching the content of the rehabilitation log text, making it more in line with the actual rehabilitation scene, and improving the authenticity and availability of the data. Thirdly, by performing correlation analysis on the basic sample rehabilitation log text and the first rehabilitation behavior text data, the first sample simulation text data is generated; at the same time, the second sample simulation text data is generated by combining the environmental change data with the second rehabilitation behavior text data, realizing the organic integration of rehabilitation log text and simulation data, and further enhancing the diversity and representativeness of the data. Finally, based on the generated first sample simulation text data and second sample simulation text data, a derivative sample rehabilitation log text for the basic sample rehabilitation log text is generated in the local rehabilitation database. These derived sample texts not only retain the characteristics and information of the original log text, but also incorporate new simulated data and content, providing richer data resources for the training and optimization of the rehabilitation log text mining network. By using these data to optimize network parameters, an optimized rehabilitation log text mining network can be generated, thereby improving the performance and accuracy of the network in rehabilitation log text processing and analysis. Thus, by simulating and generating rich and diverse rehabilitation log text data and combining network parameter optimization technology, the performance and accuracy of the rehabilitation log text mining network are improved. BRIEF DESCRIPTION OF THE DRAWINGS
[0042] Figure 1 A flowchart of an intelligent rehabilitation log text processing method using artificial intelligence provided in an embodiment of the present application;
[0043] Figure 2 A functional structure diagram of an intelligent rehabilitation guidance system for implementing the above-mentioned intelligent rehabilitation log text processing method using artificial intelligence provided in an embodiment of the present application. DETAILED DESCRIPTION
[0044] See also Figure 1 As shown, the present application provides an intelligent rehabilitation log text processing method using artificial intelligence, comprising the following steps.
[0045] Step S110, obtaining a basic sample rehabilitation log text from a cloud rehabilitation database, and adding the basic sample rehabilitation log text to a local rehabilitation database.
[0046] In detail, the cloud rehabilitation database is a large data set stored on the cloud intelligent rehabilitation guidance system, which contains various information related to rehabilitation, especially rehabilitation log text. These rehabilitation log text data can be accessed and analyzed remotely to support decision-making and research in the rehabilitation process. For example, a hospital uses cloud services to store the rehabilitation records of its patients, including patients' rehabilitation activities, rehabilitation progress, doctor's advice, etc. The cloud rehabilitation database allows these data to be accessed and analyzed by authorized medical staff at any time and any place.
[0047] The basic sample rehabilitation log text is a representative rehabilitation log text selected from the cloud rehabilitation database and is used as the basis for subsequent analysis and simulation. For example, there may be tens of thousands of rehabilitation log texts in the cloud rehabilitation database. In order to conduct research or optimize the model, researchers can select dozens or hundreds of representative texts as basic samples.
[0048] The local rehabilitation database is a database stored in a local intelligent rehabilitation guidance system or computer. It is used to store basic sample data downloaded from the cloud database, as well as subsequently generated simulation data and derivative data. For example, in order to ensure fast access and processing of data, a hospital may establish a local database on its own intelligent rehabilitation guidance system. This database will regularly synchronize data from the cloud and analyze and process it locally.
[0049] Therefore, in this embodiment, after the intelligent rehabilitation guidance system is started, it is first connected to the cloud rehabilitation database. This cloud rehabilitation database stores a large number of rehabilitation log texts, which are records of the rehabilitation process of previous patients. The intelligent rehabilitation guidance system uses its data access function to select some of these rehabilitation log texts as basic sample rehabilitation log texts. The basis for selection can be the integrity, representativeness or other specific standards of these texts, which are not specifically limited. Once the selection is completed, the intelligent rehabilitation guidance system downloads these basic sample rehabilitation log texts to the local and then adds them to the local rehabilitation database. The local database can be set up for faster data access speed, lower network latency or data security.
[0050] Step S120, obtaining the initialized rehabilitation activity characteristic parameters and environmental influencing factors, and simulating in the local rehabilitation database based on the rehabilitation activity characteristic parameters and the environmental influencing factors to generate first rehabilitation behavior text data corresponding to the patient activity state area and second rehabilitation behavior text data corresponding to the environmental effect area.
[0051] In detail, the rehabilitation activity characteristic parameters are quantitative indicators used to describe various characteristics of rehabilitation activities, such as spatial configuration information of rehabilitation simulation scenes, recovery process information, environmental quietness information, rehabilitation tracking device information, and personalized characteristic information of rehabilitation patients.
[0052] The environmental influencing factors refer to those interfering environmental factors that can affect the rehabilitation process, such as noise, light, temperature, humidity, etc.
[0053] The first rehabilitation behavior text data is text data related to the patient's activity status, describing the patient's specific behavior and status during the rehabilitation process; the second rehabilitation behavior text data is text data related to environmental effects, describing how environmental factors affect the patient's rehabilitation process.
[0054] Therefore, in this embodiment, the intelligent rehabilitation guidance system can obtain the initialized rehabilitation activity characteristic parameters and environmental influencing factors from other parts of the rehabilitation system or user input. Next, the intelligent rehabilitation guidance system uses these rehabilitation activity characteristic parameters and environmental influencing factors to simulate in the local rehabilitation database. In this process, the intelligent rehabilitation guidance system generates two sets of text data: the first rehabilitation behavior text data is text data related to the patient's activity status (such as the patient's movements, rehabilitation progress, etc.), and the second rehabilitation behavior text data is text data related to environmental effects (such as the impact of the environment on the patient's rehabilitation, recommended environmental adjustments, etc.).
[0055] Step S130, in the local rehabilitation database, the basic sample rehabilitation log text and the first rehabilitation behavior text data are correlated and analyzed to generate first sample simulation text data, and environmental change data are obtained, and in the local rehabilitation database, the environmental change data and the second rehabilitation behavior text data are correlated and analyzed to generate second sample simulation text data.
[0056] In detail, association analysis processing is a data mining technique used to discover interesting relationships between different items in a data set, such as the frequency of two events occurring at the same time. Here, it is used to analyze the association between the basic sample rehabilitation log text and the first and second rehabilitation behavior text data.
[0057] For example, suppose the basic sample rehabilitation log text mentions that a patient performed a certain rehabilitation activity in a specific environment. Association analysis can find that when the activity is performed in this specific environment, the patient's rehabilitation effect is particularly good. This association can provide valuable information for subsequent simulation and text generation.
[0058] The first sample simulation text data is a simulation text generated based on the association analysis of the basic sample rehabilitation log text and the first rehabilitation behavior text data; the second sample simulation text data is a simulation text generated based on the association analysis of the environmental change data and the second rehabilitation behavior text data. For example, if the association analysis finds that a certain rehabilitation activity is more effective in a quiet environment, then the first sample simulation text data may include "The patient performed rehabilitation activities for XX minutes in a quiet environment and felt good without obvious discomfort." The second sample simulation text data may describe "Due to the increase in environmental noise, the patient showed slight discomfort, and it is recommended to adjust the environmental noise level."
[0059] The environmental change data refers to data that changes in the rehabilitation environment in real time or periodically, such as the measured values of environmental parameters such as noise level, light intensity, temperature, etc. For example, in the rehabilitation room, there may be sensors that monitor the noise and light levels in real time. If the window of the rehabilitation room is opened one day, causing the light to suddenly increase, then this change will be recorded as environmental change data.
[0060] Therefore, in this embodiment, the intelligent rehabilitation guidance system starts to perform association analysis processing in the local rehabilitation database. First, the basic sample rehabilitation log text is compared and matched with the first rehabilitation behavior text data to find out the correlation and pattern between them. This process may involve natural language processing (NLP) technology, data mining algorithms, etc. Based on these correlations and patterns, the intelligent rehabilitation guidance system generates first sample simulation text data, which simulates the text records that may be generated in the actual rehabilitation process. At the same time, the intelligent rehabilitation guidance system also obtains environmental change data, which reflects the real-time changes in the rehabilitation environment. Then, the intelligent rehabilitation guidance system associates these environmental change data with the second rehabilitation behavior text data to understand how environmental changes affect the rehabilitation process. Finally, the intelligent rehabilitation guidance system generates second sample simulation text data, which simulates the text records of the rehabilitation process under different environmental conditions.
[0061] In another possible example, this embodiment can use text mining and natural language processing technology to perform correlation analysis on the basic sample rehabilitation log text and the first rehabilitation behavior text data. This can be achieved by identifying co-occurring keywords, phrases or patterns. The results of the analysis should reveal the potential relationship between the patient's activity status and the rehabilitation effect. In addition, the real-time collected environmental change data can be correlated with the second rehabilitation behavior text data. This can be achieved by comparing the rehabilitation effects under different environmental conditions, thereby identifying the specific impact of environmental factors on the rehabilitation process.
[0062] Based on the correlation analysis results of the basic sample rehabilitation log text and the first rehabilitation behavior text data, the first sample simulation text data is generated, which contains the description of the expected rehabilitation effect of the patient under different activity states. Combined with the correlation analysis results of the environmental change data and the second rehabilitation behavior text data, the second sample simulation text data is generated to describe the changes in the patient's rehabilitation effect under different environmental conditions.
[0063] The generated first sample simulated text data and second sample simulated text data are integrated into the local rehabilitation database for subsequent analysis and mining. When integrating the data, the use of machine learning algorithms can be considered to further optimize the simulated text data. For example, a classification algorithm can be used to classify and label the simulated text in order to more accurately identify the rehabilitation effects under different activity states and environmental conditions. At the same time, clustering algorithms can also be used to discover potential patterns and trends in simulated texts, providing more valuable reference information for the formulation of rehabilitation programs. In this way, the correlation analysis and processing of the basic sample rehabilitation log text and the rehabilitation behavior text data can be achieved, and the corresponding simulated text data can be generated. These data can provide strong support for decision-making and research in the rehabilitation process.
[0064] Step S140, in the local rehabilitation database, based on the first sample simulated text data and the second sample simulated text data, generate a derived sample rehabilitation log text for the basic sample rehabilitation log text, based on the basic sample rehabilitation log text and the derived sample rehabilitation log text, optimize the network parameters of the initialized rehabilitation log text mining network, and generate an optimized rehabilitation log text mining network.
[0065] In detail, the derived sample rehabilitation log text is a new simulation text generated based on the first sample simulation text data and the second sample simulation text data, combining the information of the basic sample text and the new findings in the simulation process. For example, by combining the description of the patient's rehabilitation activities in the basic sample text and the description of the environmental impact in the simulation text, a more complete derived sample text can be generated, such as "The patient performed rehabilitation activities for XX minutes in a quiet and moderately lit environment and felt good without obvious discomfort. However, when the environmental noise increased, the patient showed slight discomfort."
[0066] The rehabilitation log text mining network is a network model built using machine learning or deep learning technology, which is used to automatically analyze and extract useful information from rehabilitation log text. For example, this network model can be a deep learning neural network, which can be trained to identify key information in the text, such as the patient's rehabilitation progress, problems encountered, doctor's advice, etc. This information can be used to evaluate rehabilitation effects, improve treatment plans, etc. Therefore, its performance can be improved by adjusting the internal parameters of the network model (such as weights and biases), and the optimization is based on the basic sample text and the derived sample text. During the training process, the network model will continuously compare the differences between its predicted results and the actual text, and adjust its internal parameters based on these differences. Through multiple iterations and optimization, the model can gradually improve its prediction accuracy and generalization ability.
[0067] That is, in this embodiment, the intelligent rehabilitation guidance system now has two sets of simulated text data: the first sample simulated text data and the second sample simulated text data. These two sets of data respectively simulate the text records of the rehabilitation process under different conditions. Next, the intelligent rehabilitation guidance system uses these two sets of simulated data to generate derived sample rehabilitation log texts. These derived texts are extensions and supplements to the basic sample rehabilitation log texts, and they reflect possible text records under different rehabilitation activities and environmental conditions.
[0068] Then, the intelligent rehabilitation guidance system uses these basic sample rehabilitation log texts and derived sample texts to train and optimize the initialized rehabilitation log text mining network. This process involves deep learning technology, neural network models, etc. By continuously adjusting network parameters and comparing the differences between predicted results and actual texts, the intelligent rehabilitation guidance system gradually optimizes this rehabilitation log text mining network, enabling it to more accurately analyze and understand rehabilitation log text data.
[0069] Finally, the intelligent rehabilitation guidance system generated an optimized rehabilitation log text mining network, which can more effectively process and analyze text data in the rehabilitation process, and provide more accurate and valuable rehabilitation information and suggestions for medical staff and patients.
[0070] Based on the above steps, by obtaining the basic sample rehabilitation log text from the cloud rehabilitation database and adding it to the local rehabilitation database, the sharing and local storage of rehabilitation log data are realized, which provides a basis for subsequent data processing and mining. Secondly, using the initialized rehabilitation activity characteristic parameters and environmental influencing factors, simulation is performed in the local rehabilitation database to generate the first rehabilitation behavior text data corresponding to the patient's activity state area and the second rehabilitation behavior text data corresponding to the environmental action area, enriching the content of the rehabilitation log text, making it more in line with the actual rehabilitation scene, and improving the authenticity and availability of the data. Thirdly, by performing correlation analysis on the basic sample rehabilitation log text and the first rehabilitation behavior text data, the first sample simulation text data is generated; at the same time, the second sample simulation text data is generated by combining the environmental change data with the second rehabilitation behavior text data, realizing the organic integration of rehabilitation log text and simulation data, and further enhancing the diversity and representativeness of the data. Finally, based on the generated first sample simulation text data and second sample simulation text data, a derivative sample rehabilitation log text for the basic sample rehabilitation log text is generated in the local rehabilitation database. These derived sample texts not only retain the characteristics and information of the original log text, but also incorporate new simulated data and content, providing richer data resources for the training and optimization of the rehabilitation log text mining network. By using these data to optimize network parameters, an optimized rehabilitation log text mining network can be generated, thereby improving the performance and accuracy of the network in rehabilitation log text processing and analysis. Thus, by simulating and generating rich and diverse rehabilitation log text data and combining network parameter optimization technology, the performance and accuracy of the rehabilitation log text mining network are improved.
[0071] In a possible implementation, the rehabilitation activity characteristic parameters include spatial configuration information of the rehabilitation simulation scene, recovery process information, environmental quietness information, rehabilitation tracking device information, and rehabilitation patient personalized characteristic information. Step S120 may include:
[0072] Step S121, determining the three-dimensional layout of the rehabilitation simulation scene and the spatiotemporal information of the rehabilitation activities in the local rehabilitation database according to the spatial configuration information and the recovery process information of the rehabilitation simulation scene.
[0073] In this embodiment, the intelligent rehabilitation guidance system first constructs a three-dimensional virtual model in the local rehabilitation database based on the spatial configuration information of the rehabilitation simulation scene, such as room size, equipment placement, etc. This three-dimensional virtual model can accurately reflect the spatial layout of the actual rehabilitation environment. At the same time, the intelligent rehabilitation guidance system combines the recovery process information, such as the patient's rehabilitation stage, daily activity plan, etc., to assign a timeline to this three-dimensional layout, thereby determining the rehabilitation activities that the patient should perform in different time periods and their locations in space.
[0074] Step S122, determining a reference monitoring point and an optimized monitoring point based on the three-dimensional layout of the rehabilitation simulation scene and the rehabilitation tracking device information.
[0075] Based on the three-dimensional layout, the intelligent rehabilitation guidance system determines a series of benchmark monitoring points according to the rehabilitation tracking device information, such as the type, quantity, and placement of sensors. These benchmark monitoring points are used to collect basic data of patients during rehabilitation. In order to more accurately capture the patient's activity status and environmental impact, the intelligent rehabilitation guidance system will also optimize and adjust the benchmark monitoring points based on the patient's personalized characteristic information (such as height, weight, activity habits, etc.) and the characteristics of rehabilitation activities, and determine a series of optimized monitoring points.
[0076] Step S123, determining the patient's activity status area based on the benchmark monitoring points and the rehabilitation patient's personalized characteristic information.
[0077] The intelligent rehabilitation guidance system analyzes the data collected from the baseline monitoring points and the optimized monitoring points, and combines the patient's personalized characteristic information (such as the patient's mobility, preferences, etc.) to determine the patient's possible range of activity and patient activity status area during the rehabilitation process. These patient activity status areas can be static (such as the area where the patient performs a specific exercise) or dynamic (such as the path the patient walks).
[0078] Step S124: determining an environmental action area based on the three-dimensional layout of the rehabilitation simulation scene and the environmental impact factors.
[0079] Based on the three-dimensional layout of the rehabilitation simulation scene, the intelligent rehabilitation guidance system combines environmental influencing factors (such as temperature, humidity, noise level, etc.) to determine the areas where different environmental factors affect the patient's rehabilitation activities. For example, the noise level in a certain area may be too high, affecting the patient's concentration and rehabilitation effect; or the temperature in a certain area may be too low, which is not conducive to the patient's rehabilitation activities that require exposing certain body parts.
[0080] Step S125 , based on the three-dimensional layout of the rehabilitation simulation scene, the optimized monitoring points, the spatiotemporal information of the rehabilitation activity and the patient activity state area, generating first rehabilitation behavior text data corresponding to the patient activity state area.
[0081] The intelligent rehabilitation guidance system generates text data describing the patient's activity status based on the three-dimensional layout of the rehabilitation simulation scene, optimized monitoring points, the spatiotemporal information of the rehabilitation activities, and the information of the patient's activity status area. These data can include the patient's specific activities at different times and locations (such as what actions were performed, how long they lasted, how intense the activities were, etc.), as well as the potential impact of these activities on the patient's rehabilitation process.
[0082] Step S126, generating second rehabilitation behavior text data corresponding to the environmental action area based on the three-dimensional layout of the rehabilitation simulation scene, the optimized monitoring points, the spatiotemporal information of the rehabilitation activities and the environmental action area.
[0083] The intelligent rehabilitation guidance system is also based on the three-dimensional layout of the rehabilitation simulation scene, optimized monitoring points, and spatiotemporal information of rehabilitation activities, but this time it combines information on the environmental impact area to generate second rehabilitation behavior text data that describes the impact of the environment on the patient's rehabilitation activities. These second rehabilitation behavior text data may include specific values or levels of different environmental factors (such as noise, light, temperature, etc.) at different times and locations, and how these factors affect the patient's rehabilitation activities and rehabilitation effects. For example, the second rehabilitation behavior text data may indicate that during a certain period of time, due to excessive environmental noise, the patient showed irritability, resulting in a decrease in the completion and quality of the rehabilitation activities.
[0084] In a possible implementation, step S121 may include:
[0085] Step S1211, obtaining a first horizontal extension parameter and a second horizontal extension parameter in the spatial configuration information of the rehabilitation simulation scene in the local rehabilitation database, wherein the first horizontal extension parameter and the second horizontal extension parameter are horizontal extension parameters in different vertical directions in the plane area where the rehabilitation simulation scene is located.
[0086] In this embodiment, the intelligent rehabilitation guidance system first retrieves the spatial configuration information of the rehabilitation simulation scene from the local rehabilitation database. These spatial configuration information contain the key spatial parameters of the rehabilitation environment. Among them, the first horizontal extension parameter may represent the length or width of the rehabilitation scene in one direction (such as the east-west direction), while the second horizontal extension parameter represents the length or width in another direction perpendicular to the first direction (such as the north-south direction). The intelligent rehabilitation guidance system accurately obtains the first horizontal extension parameter and the second horizontal extension parameter, laying the foundation for constructing the three-dimensional layout of the rehabilitation simulation scene.
[0087] Step S1212, obtaining a height extension parameter of a height area in the spatial configuration information of the rehabilitation simulation scene.
[0088] After obtaining the horizontal extension parameters, the intelligent rehabilitation guidance system continues to extract the height extension parameters from the spatial configuration information. This height extension parameter defines the vertical range of the rehabilitation scene, which can be the height from the ground to the ceiling. This step is crucial for the complete construction of the three-dimensional rehabilitation environment because it ensures the accuracy of the model in the vertical dimension.
[0089] Step S1213, determining a target rehabilitation training time according to the recovery progress information.
[0090] The intelligent rehabilitation guidance system then determines the target rehabilitation training time based on the recovery progress information. The recovery progress information may include the patient's rehabilitation plan, the activity schedule recommended by the doctor, etc. By analyzing this information, the intelligent rehabilitation guidance system can identify the best time window for the patient to perform specific rehabilitation activities, which is the target rehabilitation training time. The determination of this time parameter helps the intelligent rehabilitation guidance system generate a personalized rehabilitation activity plan for the patient.
[0091] Step S1214, determining the spatiotemporal information of the rehabilitation activity according to the first horizontal extension parameter, the second horizontal extension parameter, the height extension parameter and the target rehabilitation training time of the rehabilitation simulation scene.
[0092] Finally, the intelligent rehabilitation guidance system combines the first level extension parameter, the second level extension parameter, the height extension parameter, and the target rehabilitation training time to determine the spatiotemporal information of the rehabilitation activities. This means that the intelligent rehabilitation guidance system now has a complete three-dimensional rehabilitation environment model and knows which activities the patient should perform in this environment within which time period. The determination of spatiotemporal information provides the necessary basis for subsequent simulation and text data generation, ensuring the accuracy and executability of the rehabilitation plan.
[0093] In a possible implementation, the rehabilitation tracking device information includes a first sensing range limit parameter, a device extreme position parameter, a device label, a device quantity, and a device interval. Step S122 may include:
[0094] Step S1221, determining a reference monitoring point based on the three-dimensional layout of the rehabilitation simulation scene, the first sensing range limit parameter, and the device extreme position parameter.
[0095] In this embodiment, the intelligent rehabilitation guidance system first analyzes the three-dimensional layout of the rehabilitation simulation scene, which is a virtual model that accurately reflects the spatial structure of the actual rehabilitation environment. On this basis, the intelligent rehabilitation guidance system considers the first sensing range limitation parameters, which define the range that the rehabilitation tracking device can effectively sense. For example, some sensors may only be able to accurately track the patient's activities within a specific distance. At the same time, the device extreme position parameters specify the farthest or closest position where the tracking device can be placed or installed. Combined with this information, the intelligent rehabilitation guidance system selects a series of positions in the three-dimensional layout as reference monitoring points. These reference monitoring points are within the effective sensing range of the sensor and meet the extreme position requirements of the device.
[0096] Step S1222: determining an optimized monitoring point based on the reference monitoring point, the device tag, the number of devices, and the device interval.
[0097] Once the benchmark monitoring points are determined, the intelligent rehabilitation guidance system then considers device tags, device quantity, and device spacing to further optimize the layout of these points. Device tags may contain information about the type, function, or calibration status of a specific tracking device. For example, the intelligent rehabilitation guidance system can use these device tags to ensure that different types of devices are deployed in the locations that best suit them. At the same time, the number of devices and device spacing parameters guide the intelligent rehabilitation guidance system on how to avoid unnecessary device overlap or waste while maintaining adequate coverage. By combining these factors, the intelligent rehabilitation guidance system makes adjustments based on the benchmark monitoring points and determines a series of optimized monitoring points that can more accurately capture the patient's activity status while effectively utilizing available tracking device resources.
[0098] In a possible implementation, the rehabilitation patient personalized characteristic information includes a second sensing range limit parameter, a patient dynamic remote ratio, a patient dynamic remote distance parameter limit information, a rehabilitation patient action proximity distance parameter limit information, and a rehabilitation patient action height limit information. Step S123 may include:
[0099] Step S1231, generating a first simulation value, when the patient's dynamic remote ratio is less than the first simulation value, determining the distance parameter between the rehabilitation patient and the benchmark monitoring point based on the patient's dynamic remote distance parameter restriction information.
[0100] In this embodiment, the intelligent rehabilitation guidance system first generates a first simulation value, which can be calculated based on factors such as the average size of the rehabilitation environment and the general activity ability of the patient. Then, the intelligent rehabilitation guidance system obtains the patient's dynamic remote ratio in the personalized characteristic information of the rehabilitation patient, and the patient's dynamic remote ratio reflects the range of possible activities of the patient during the rehabilitation process. The intelligent rehabilitation guidance system compares the patient's dynamic remote ratio with the first simulation value.
[0101] If the patient's dynamic remote ratio is less than the first simulation value, it means that the patient's range of activity is relatively small and may need to be monitored at a closer distance. In this case, the intelligent rehabilitation guidance system can determine the spacing parameters between the rehabilitation patient and the reference monitoring point based on the patient's dynamic spacing parameter restriction information. These parameters can include minimum and maximum spacing to ensure that the sensor can accurately capture the patient's activities while avoiding interference caused by being too close.
[0102] Step S1232: When the patient's dynamic remote ratio is not less than the first simulation value, the distance parameter between the rehabilitation patient and the benchmark monitoring point is determined based on the rehabilitation patient's action proximity distance parameter restriction information, and the direction angle of the rehabilitation patient relative to the rehabilitation tracking device is randomly generated. The patient's activity status area is determined based on the benchmark monitoring point, the distance parameter between the rehabilitation patient and the benchmark monitoring point, the direction angle, and the rehabilitation patient's action height restriction information.
[0103] If the patient's dynamic remote ratio is not less than the first simulation value, it means that the patient's range of activity is wide and a more flexible monitoring strategy may be needed. In this case, the intelligent rehabilitation guidance system can determine the distance parameters between the patient and the reference monitoring point based on the rehabilitation patient's action proximity distance parameter limit information. These parameters may allow a larger activity space while still ensuring the accuracy of monitoring.
[0104] At the same time, the intelligent rehabilitation guidance system can randomly generate the direction angle of the rehabilitation patient relative to the rehabilitation tracking device. This angle represents the possible activity direction of the patient in the rehabilitation environment, which helps to more comprehensively simulate the patient's activity state.
[0105] Finally, the intelligent rehabilitation guidance system can comprehensively consider the location of the reference monitoring point, the distance parameters between the patient and the reference monitoring point, the direction angle, and the height restriction information of the rehabilitation patient's movement (such as the highest and lowest positions that the patient can reach) to determine the patient's activity state area. This area represents the three-dimensional spatial range of the patient's possible activities during the rehabilitation process, providing an important basis for the subsequent generation of rehabilitation behavior text data.
[0106] In a possible implementation, the patient activity state area includes a horizontal coordinate of the rehabilitation patient, a longitudinal coordinate of the rehabilitation patient, and a vertical coordinate of the rehabilitation patient. Step S1232 may include:
[0107] Step S1232-1, based on the spacing parameters and direction angles between the rehabilitation patient and the reference monitoring point, determine the first misalignment spacing parameter between the rehabilitation patient and the reference monitoring point in the transverse axis and the second misalignment spacing parameter between the rehabilitation patient and the reference monitoring point in the longitudinal axis.
[0108] In this embodiment, the intelligent rehabilitation guidance system first calculates based on the spacing parameter and direction angle between the rehabilitation patient and the reference monitoring point. The spacing parameter can be a specific value, which represents the straight-line distance between the patient and the monitoring point; and the direction angle indicates the direction of the patient's activity relative to the monitoring point. The intelligent rehabilitation guidance system uses this information and mathematical methods such as trigonometric functions to calculate the first misalignment spacing parameter between the rehabilitation patient and the reference monitoring point in the horizontal axis, and the second misalignment spacing parameter in the longitudinal axis. These two parameters represent the patient's offset relative to the monitoring point in the horizontal and vertical directions, respectively.
[0109] Step S1232-2, determining the initial lateral coordinate of the rehabilitation patient based on the lateral coordinate of the reference monitoring point and the first offset spacing parameter.
[0110] Step S1232-3, determining the initial longitudinal coordinate of the rehabilitation patient based on the longitudinal coordinate of the reference monitoring point and the second offset spacing parameter.
[0111] The intelligent rehabilitation guidance system will then determine the initial horizontal coordinates of the rehabilitation patient based on the horizontal coordinates of the reference monitoring point and the first misalignment spacing parameter through simple addition or subtraction operations. Similarly, the intelligent rehabilitation guidance system will also calculate the initial vertical coordinates of the rehabilitation patient based on the vertical coordinates of the reference monitoring point and the second misalignment spacing parameter. These two coordinates together constitute the initial position of the patient in the rehabilitation environment.
[0112] Step S1232-4, determining the vertical coordinates of the rehabilitation patient based on the rehabilitation patient's movement height restriction information.
[0113] The intelligent rehabilitation guidance system can determine the patient's vertical coordinates based on the rehabilitation patient's height limit information. This information may include the minimum and maximum height limits for the patient's activities. The intelligent rehabilitation guidance system can select a suitable vertical coordinate within this range to ensure that the patient's activities can be effectively monitored.
[0114] Step S1232-5, when it is determined that at least one of the initial horizontal coordinate and the initial longitudinal coordinate needs to be adjusted based on the initial horizontal coordinate, the initial longitudinal coordinate and the second sensing range limit parameter, at least one of the initial horizontal coordinate and the initial longitudinal coordinate that need to be adjusted is adjusted to generate at least one of the adjusted horizontal coordinate and the adjusted longitudinal coordinate.
[0115] After determining the patient's initial transverse coordinates, initial longitudinal coordinates, and vertical coordinates, the intelligent rehabilitation guidance system can further check whether these coordinates meet the requirements of the second sensing range limit parameters. These parameters define the area range that the sensor can effectively monitor. If at least one of the initial transverse coordinates and the initial longitudinal coordinates is outside the effective monitoring range of the sensor, the intelligent rehabilitation guidance system will make corresponding adjustments to the coordinates that need to be adjusted. The adjustment method can be to move the coordinates a certain distance within the effective monitoring range, or to directly select a new coordinate within the effective monitoring range. The adjusted coordinates will be used as the final coordinates of the patient's activity status area.
[0116] In a possible implementation, the environmental impact factor includes a third sensing range limit parameter and vertical boundary information of an environmental impact source, and the environmental action area includes a lateral coordinate of the environmental impact source, a longitudinal coordinate of the environmental impact source, and a vertical coordinate of the environmental impact source. Step S124 may include:
[0117] Step S1241 : determining the lateral coordinate of the environmental impact source according to the lateral extension parameter and the third sensing range limitation parameter in the three-dimensional layout of the rehabilitation simulation scene.
[0118] In this embodiment, the intelligent rehabilitation guidance system first obtains the three-dimensional layout information of the rehabilitation simulation scene, which includes the lateral extension parameter of the scene, which defines the width or range of the scene in the horizontal direction. At the same time, the intelligent rehabilitation guidance system also obtains the third sensing range limitation parameter in the environmental impact factor, and this third sensing range limitation parameter limits the lateral range of the environmental impact source. The intelligent rehabilitation guidance system combines these two parameters and determines the lateral coordinate position of the environmental impact source through calculation or logical judgment. This position can be a specific point in the scene or the center point of a range. It represents the starting position where the environmental impact source affects the rehabilitation activity in the horizontal direction.
[0119] Step S1242: determining the longitudinal coordinate of the environmental impact source according to the axial extension parameter and the third sensing range limit parameter in the three-dimensional layout of the rehabilitation simulation scene.
[0120] Next, the intelligent rehabilitation guidance system continues to use the axial extension parameters (here can refer to the parameters in the longitudinal or depth direction) in the three-dimensional layout information of the rehabilitation simulation scene, and the third sensing range limitation parameter to determine the coordinate position of the environmental influence source in the longitudinal direction. The axial extension parameter defines the length or range of the scene in the longitudinal direction, while the third sensing range limitation parameter limits the scope of the environmental influence source in the longitudinal direction. The intelligent rehabilitation guidance system uses a similar method to combine these two parameters for calculation or logical judgment, and finally determines the coordinate position of the environmental influence source in the longitudinal direction. This position represents the starting position of the environmental influence source in the vertical direction to affect the rehabilitation activity.
[0121] Step S1243, determining the vertical coordinates of the environmental impact source according to the vertical boundary information of the environmental impact source.
[0122] Finally, the intelligent rehabilitation guidance system determines the vertical coordinate position of the environmental impact source based on the vertical definition information of the environmental impact source in the environmental impact factor. The vertical definition information may include parameters such as the height range, starting height, and ending height of the environmental impact source in the vertical direction. The intelligent rehabilitation guidance system uses these parameters for calculation or direct assignment to determine the specific vertical coordinate position of the environmental impact source. This position can be a fixed height point in the scene or a position point within a height range. It represents the starting height of the environmental impact source in the vertical direction that affects the rehabilitation activity.
[0123] Through the above three steps, the intelligent rehabilitation guidance system can accurately determine the three-dimensional coordinate position (horizontal coordinate, longitudinal coordinate and vertical coordinate) of the environmental impact source in the rehabilitation simulation scene, thereby further defining the scope and boundaries of the environmental impact area. This environmental impact area will serve as one of the basic data for subsequent simulation and analysis of the impact of the environment on rehabilitation activities.
[0124] In a possible implementation, step S140 may include:
[0125] Step S141, obtaining initialized text ratio parameters and regularization parameters, and determining a target mixing ratio according to the text ratio parameters, and determining target regularization mapping information according to the regularization parameters.
[0126] In this embodiment, the intelligent rehabilitation guidance system first accesses the internal storage or configuration file to obtain the initialized text ratio parameter and regularization parameter. The text ratio parameter can be a specific value or ratio, which is used to determine the weight distribution of different sample texts in the mixing process. The regularization parameter can be a set of rules or mapping relationships, which is used to regularize the results after the text is mixed to ensure that the generated text meets specific format, grammar or semantic requirements.
[0127] Thus, the intelligent rehabilitation guidance system performs calculation or logical judgment based on the obtained text ratio parameter to determine the target mixing ratio. This target mixing ratio will be used to guide the weight distribution of the first sample simulated text data and the second sample simulated text data in the mixing process. For example, if the text ratio parameter is set to 0.7, the intelligent rehabilitation guidance system can determine the target mixing ratio to be 70% of the first sample text and 30% of the second sample text.
[0128] The intelligent rehabilitation guidance system will then parse or process the obtained regularization parameters to generate target regularization mapping information. These target regularization mapping information may include a series of replacement rules, insertion rules, deletion rules, etc., which are used to regularize the results after the text is mixed. For example, the regularization parameters may specify the replacement rules of certain specific words, or the adjustment rules of certain sentence structures, etc.
[0129] Step S142, determining a mixing weight of the second sample simulated text data according to the target mixing ratio and the first sample simulated text data.
[0130] The intelligent rehabilitation guidance system calculates the mixing weight that should be assigned to the second sample simulated text data during the mixing process based on the target mixing ratio and the length and importance of the first sample simulated text data. This mixing weight will be used in the subsequent calculation of the mixing result to ensure that the two sample texts can be mixed according to the specified ratio.
[0131] Step S143, determining a mixing result according to the first sample simulated text data, the second sample simulated text data and the mixing weight.
[0132] The intelligent rehabilitation guidance system mixes the first sample simulated text data and the second sample simulated text data according to the calculated mixing weight. The mixing process may include operations such as vocabulary replacement, sentence reorganization, and paragraph merging to generate a mixed text result that contains both the first sample features and the second sample features.
[0133] Step S144, performing regularized mapping on the mixed result according to the target regularized mapping information to generate a derived sample rehabilitation log text.
[0134] Finally, the intelligent rehabilitation guidance system performs regularization processing on the mixed results according to the target regularization mapping information. The processing process may include replacing certain specific words in the mixed results, adjusting the sentence structure to conform to grammatical rules, adding necessary punctuation marks, and other operations. After regularization processing, the intelligent rehabilitation guidance system finally generates a derivative sample rehabilitation log text that meets the requirements. This derivative sample rehabilitation log text not only retains the characteristics of the original sample text, but also conforms to specific format and grammatical requirements, and can be used for subsequent rehabilitation log analysis and application.
[0135] In a possible implementation, the basic sample rehabilitation log text includes at least the basic sample rehabilitation log text of the first training stage, and correspondingly, the derived sample rehabilitation log text includes at least the derived sample rehabilitation log text of the first training stage, and step S140 further includes:
[0136] Step S145 , caching the derived sample rehabilitation log text of the first training phase into a sample sharing database.
[0137] In this embodiment, after the intelligent rehabilitation guidance system completes the generation of the derived sample rehabilitation log text of the first training stage, it will store these derived sample rehabilitation log text data in the sample sharing database. This sample sharing database can be a distributed storage system specifically used to store and manage sample text data of various training stages. The intelligent rehabilitation guidance system uploads the derived sample rehabilitation log text to the database through a database interface or API, and assigns a unique identifier or index to it, so that the data can be quickly accessed and used in the subsequent network learning and optimization process.
[0138] Step S146, during the network learning process, the network parameters of the initialized rehabilitation log text mining network are optimized based on the basic sample rehabilitation log text and the derived sample rehabilitation log text in the first training phase.
[0139] In this embodiment, the intelligent rehabilitation guidance system starts the network learning process of the rehabilitation log text mining network, and uses the basic sample rehabilitation log text and the derived sample rehabilitation log text in the first training phase as training data. The intelligent rehabilitation guidance system inputs these basic sample rehabilitation log texts and the derived sample rehabilitation log texts into the initialized rehabilitation log text mining network, and obtains the output result through the forward propagation calculation of the network. Then, the intelligent rehabilitation guidance system calculates the error between the output result and the true label according to the preset loss function, and updates the parameters of the network through the back propagation algorithm to reduce the error and improve the performance of the network. This process will be repeated until the network reaches a certain convergence condition or performance requirement.
[0140] Step S147, obtaining the basic sample rehabilitation log text of the subsequent training stage, simulating the basic sample rehabilitation log text of the subsequent training stage in parallel in the simulation program, generating the derived sample rehabilitation log text of the subsequent training stage accordingly, and loading the derived sample rehabilitation log text of the subsequent training stage into the sample sharing database.
[0141] After completing the network parameter optimization in the first training phase, the intelligent rehabilitation guidance system can obtain the basic sample rehabilitation log texts for the subsequent training phase. These basic sample rehabilitation log texts may come from different data sources or acquisition devices and have different characteristics and distributions. The intelligent rehabilitation guidance system inputs these basic sample rehabilitation log text data into the simulation program, and uses the parallel processing capabilities of the simulation program to simulate and transform the basic sample rehabilitation log texts to generate derivative sample rehabilitation log texts corresponding to the subsequent training phase. Then, the intelligent rehabilitation guidance system loads these newly generated derivative sample text data into the sample sharing database for subsequent network learning and optimization use.
[0142] Step S148, in the network learning process, based on the basic sample rehabilitation log text and the derived sample rehabilitation log text in the subsequent training phase, the network parameters of the rehabilitation log text mining network are continuously optimized until an optimized rehabilitation log text mining network is generated.
[0143] The intelligent rehabilitation guidance system continues the network learning process and uses the basic sample rehabilitation log text and the derived sample rehabilitation log text in the subsequent training phase as new training data. Similar to the first training phase, the intelligent rehabilitation guidance system inputs these text data into the current rehabilitation log text mining network and updates and optimizes the parameters through the network's forward propagation and back propagation. This process will continue to iterate until the network achieves better performance on the new training data or meets the preset termination conditions. Finally, the intelligent rehabilitation guidance system generates an optimized rehabilitation log text mining network that can better process and analyze various rehabilitation log text data.
[0144] In a possible implementation, this embodiment may establish multiple simulation programs.
[0145] Step S147 may include:
[0146] Step S1471, loading the basic sample rehabilitation log texts of the same training phase into the multiple simulation programs.
[0147] Step S1472, based on the multiple simulation programs, the basic sample rehabilitation log texts loaded into the corresponding ones are simulated in parallel to generate derived sample rehabilitation log texts of the basic sample rehabilitation log texts for the same training stage.
[0148] In this embodiment, the intelligent rehabilitation guidance system creates or configures multiple simulation programs during the initialization phase. These simulation programs can be based on different algorithms, models, or parameter settings, and are used to perform diversified simulations and transformations on the basic sample rehabilitation log text. Each simulation program has independent processing capabilities and specific simulation rules to ensure that the generated derived sample rehabilitation log text is rich and diverse. The intelligent rehabilitation guidance system may determine the number and type of simulation programs based on hardware resources, computing power, and demand scale, and perform corresponding resource allocation and scheduling.
[0149] When the intelligent rehabilitation guidance system obtains the basic sample rehabilitation log text of a certain training stage, it can distribute these basic sample rehabilitation log text data to multiple simulation programs established previously. This process can be implemented through data pipelines, message queues, or distributed file systems to ensure that the data can be efficiently and accurately transmitted to each simulation program. Each simulation program will receive the same basic sample text data as input for subsequent simulations and transformations.
[0150] Once the basic sample rehabilitation log text is loaded into each simulator, the intelligent rehabilitation guidance system can trigger a parallel simulation process. Each simulator simulates and transforms the input basic sample text according to its own algorithms and rules. This can include operations such as vocabulary replacement, sentence reorganization, grammatical transformation, style transfer, etc. to generate derived sample texts that are similar to the original text but have differences. Since this process is carried out in parallel in multiple simulators, the speed and efficiency of generating derived sample texts can be significantly improved. Ultimately, each simulator will output derived sample rehabilitation log texts for the basic sample rehabilitation log text in the same training phase, which will be used in subsequent network learning and optimization processes.
[0151] In a possible implementation, the method further includes:
[0152] Step A110, outputs the three-dimensional layout of the rehabilitation simulation scene when generating the derived sample rehabilitation log text, the optimized monitoring points, the patient activity status area, the first rehabilitation behavior text data and the second rehabilitation behavior text data as simulated event data, and adds the simulated event data to the simulated event repository.
[0153] In this embodiment, in the process of generating the derived sample rehabilitation log text by the intelligent rehabilitation guidance system, various data related to the simulation process can be recorded and collected at the same time. These data include the three-dimensional layout information of the rehabilitation simulation scene (such as the size of the room, the placement of furniture, etc.), the location and data of the optimized monitoring points (such as the key activity area of the patient, the body parts that need special attention, etc.), the patient's activity state area (such as the movement trajectory of the patient during the simulation process, the stay area, etc.), and the first rehabilitation behavior text data and the second rehabilitation behavior text data (these text data describe the specific rehabilitation behavior and activities of the patient during the simulation process). The intelligent rehabilitation guidance system integrates these data into simulation event data and adds them to the simulation event repository. This repository may be a database or file system specifically used to store and manage simulation event data, which can easily store, retrieve and update data.
[0154] Step A120, when it is necessary to generate a derived sample rehabilitation log text from the derived basic sample rehabilitation log text, obtain a randomly generated new event simulation rate.
[0155] During the operation of the intelligent rehabilitation guidance system, when it is necessary to generate a derived sample rehabilitation log text based on the derived basic sample rehabilitation log text, a randomly generated new event simulation rate can be obtained first. This new event simulation rate is a parameter used to control the probability of generating new events during the simulation process, and can be dynamically adjusted according to actual needs or algorithm settings. The intelligent rehabilitation guidance system may use a random number generator or other algorithms to generate this new event simulation rate to ensure that each simulation process has a certain degree of difference and randomness.
[0156] Step A130: When the new event simulation rate is less than the initialized preset simulation rate, randomly obtain target simulation event data from the simulation event storage library.
[0157] After obtaining the new event simulation rate, the intelligent rehabilitation guidance system will compare it with the preset simulation rate set during initialization. If the new event simulation rate is less than the preset simulation rate, it means that more existing simulation event data needs to be used to generate derived sample rehabilitation log texts in the current simulation process. Therefore, the intelligent rehabilitation guidance system can randomly select some target simulation event data from the simulation event repository as a reference. These target simulation event data may include information such as rehabilitation scene layout, patient activity status, rehabilitation behavior text data recorded in the previous simulation process, which can provide valuable reference and basis for the current simulation process.
[0158] Step A140, generating a derived sample rehabilitation log text for the derived basic sample rehabilitation log text in the local rehabilitation database based on the derived basic sample rehabilitation log text, the first rehabilitation behavior text data and the second rehabilitation behavior text data in the target simulation event data.
[0159] Finally, the intelligent rehabilitation guidance system will use the first rehabilitation behavior text data and the second rehabilitation behavior text data in the selected target simulation event data, combined with the derived basic sample rehabilitation log text, to generate derived sample rehabilitation log text in the local rehabilitation database. This process may include text merging, replacement, insertion and other operations to ensure that the generated derived sample text retains the characteristics and information of the original text and incorporates the elements and content in the new simulation event data. The generated derived sample rehabilitation log text will be used in the subsequent network learning and optimization process to improve the performance and accuracy of the rehabilitation log text mining network.
[0160] In a possible implementation, the method further includes:
[0161] Step A150, when the new event simulation rate is not less than the initialized preset simulation rate, simulation is performed in the local rehabilitation database based on the rehabilitation activity characteristic parameters and the environmental influencing factors to generate derived first rehabilitation behavior text data corresponding to the patient activity status area and derived second rehabilitation behavior text data corresponding to the environmental action area.
[0162] When the intelligent rehabilitation guidance system determines that the new event simulation rate is not less than the preset simulation rate, it means that the current simulation process needs more new data to increase diversity and innovation. Therefore, the intelligent rehabilitation guidance system can simulate in the local rehabilitation database based on the characteristic parameters of rehabilitation activities (such as the patient's age, gender, physical condition, rehabilitation stage, etc.) and environmental influencing factors (such as the type, layout, lighting, noise, etc. of rehabilitation facilities). By simulating the patient's behavior and reaction in different activity states, the intelligent rehabilitation guidance system generates derived first rehabilitation behavior text data corresponding to the patient's activity state area. These derived first rehabilitation behavior text data describe the patient's specific activities and performance in the simulation scene. At the same time, the intelligent rehabilitation guidance system also simulates the impact of the environment on the patient's rehabilitation behavior and generates derived second rehabilitation behavior text data corresponding to the environmental action area. These derived second rehabilitation behavior text data reflect how environmental factors interact with the patient's rehabilitation behavior.
[0163] Step A160, performing correlation analysis on the derived basic sample rehabilitation log text and the derived first rehabilitation behavior text data in the local rehabilitation database to generate derived first sample simulation text data.
[0164] After generating the derived first rehabilitation behavior text data, the intelligent rehabilitation guidance system can associate and analyze these derived first rehabilitation behavior text data with the derived basic sample rehabilitation log text. This process may include operations such as text alignment, matching, and fusion to ensure that the derived first rehabilitation behavior text data can be accurately inserted into the corresponding position of the derived basic sample rehabilitation log text. Through association analysis and processing, the intelligent rehabilitation guidance system generates derived first sample simulation text data, which contains both the information of the original log text and the content of the new simulation behavior data.
[0165] Step A170, obtaining derived environment change data, and performing correlation analysis on the derived environment change data and the derived second rehabilitation behavior text data in the local rehabilitation database to generate derived second sample simulation text data.
[0166] Next, the intelligent rehabilitation guidance system can obtain derived environmental change data, which reflect the dynamic changes of environmental factors in the simulation scene (such as changes in light intensity, fluctuations in noise levels, etc.). The intelligent rehabilitation guidance system can associate these derived environmental change data with the derived second rehabilitation behavior text data generated previously. This process is similar to the association analysis process in step two, but the focus is on combining the changes in environmental factors with the patient's rehabilitation behavior text data. Through association analysis processing, the intelligent rehabilitation guidance system generates derived second sample simulation text data, which reflects the specific impact of changes in environmental factors on the patient's rehabilitation behavior.
[0167] Step A180: generating, in the local rehabilitation database, a derived derivative sample rehabilitation log text for the derived basic sample rehabilitation log text based on the derived first sample simulation text data and the derived second sample simulation text data.
[0168] Finally, the intelligent rehabilitation guidance system will use the generated derived first sample simulation text data and derived second sample simulation text data to generate derived sample rehabilitation log text for the derived basic sample rehabilitation log text in the local rehabilitation database. This process may include text merging, replacement, adjustment and other operations to ensure that the generated derived sample text not only meets the structure and format requirements of the original log text, but also incorporates new simulation data and content. The generated derived sample rehabilitation log text will be used in the subsequent network learning and optimization process to improve the performance and accuracy of the rehabilitation log text mining network.
[0169] Figure 2The intelligent rehabilitation guidance system 100 shown includes: a processor 1001 and a memory 1003. The processor 1001 and the memory 1003 are connected, such as through a bus 1002. Optionally, the intelligent rehabilitation guidance system 100 may also include a transceiver 1004, which can be used for data interaction between the intelligent rehabilitation guidance system 100 and other intelligent rehabilitation guidance systems 100, such as data transmission and / or data reception. It should be noted that in actual scheduling, the transceiver 1004 is not limited to one, and the structure of the intelligent rehabilitation guidance system 100 does not constitute a limitation on the embodiments of the present application.
[0170] Processor 1001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of this application. Processor 1001 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.
[0171] The bus 1002 may include a path to transmit information between the above components. The bus 1002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus 1002 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 2 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.
[0172] The memory 1003 may be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), magnetic disk storage medium, other magnetic storage devices, or any other medium that can be used to carry or store program code and can be read by a computer, without limitation herein.
[0173] The memory 1003 is used to store program codes for executing the embodiments of the present application, and the execution is controlled by the processor 1001. The processor 1001 is used to execute the program codes stored in the memory 1003 to implement the steps shown in the above method embodiments.
[0174] An embodiment of the present application provides a computer-readable storage medium having program code stored thereon. When the program code is executed by a processor, the steps and corresponding contents of the aforementioned method embodiment can be implemented.
[0175] The above is only an optional implementation method for some implementation scenarios of the present application. It should be pointed out that for ordinary technicians in this technical field, without departing from the technical concept of the scheme of the present application, other similar implementation methods based on the technical ideas of the present application also belong to the protection scope of the embodiments of the present application.
Claims
1. An intelligent rehabilitation log text processing method using artificial intelligence, characterized in that: The method comprises: Obtaining a basic sample rehabilitation log text from a cloud rehabilitation database, and adding the basic sample rehabilitation log text to a local rehabilitation database; Acquire the initialized rehabilitation activity characteristic parameters and environmental influence factors, and simulate in the local rehabilitation database according to the rehabilitation activity characteristic parameters and the environmental influence factors to generate first rehabilitation behavior text data corresponding to the patient activity state area and second rehabilitation behavior text data corresponding to the environmental effect area; Performing correlation analysis on the basic sample rehabilitation log text and the first rehabilitation behavior text data in the local rehabilitation database to generate first sample simulation text data, and acquiring environment change data, performing correlation analysis on the environment change data and the second rehabilitation behavior text data in the local rehabilitation database to generate second sample simulation text data; In the local rehabilitation database, based on the first sample simulated text data and the second sample simulated text data, a derived sample rehabilitation log text for the basic sample rehabilitation log text is generated, and based on the basic sample rehabilitation log text and the derived sample rehabilitation log text, network parameters of an initialized rehabilitation log text mining network are optimized to generate an optimized rehabilitation log text mining network; The rehabilitation activity characteristic parameters include spatial configuration information of the rehabilitation simulation scene, recovery process information, environmental quietness information, rehabilitation tracking device information, and rehabilitation patient personalized characteristic information. The simulation is performed in the local rehabilitation database based on the rehabilitation activity characteristic parameters and the environmental influencing factors to generate first rehabilitation behavior text data corresponding to the patient activity state area and second rehabilitation behavior text data corresponding to the environmental effect area, including: Determining the three-dimensional layout of the rehabilitation simulation scene and the spatiotemporal information of the rehabilitation activities in the local rehabilitation database according to the spatial configuration information and the recovery process information of the rehabilitation simulation scene; Determining a reference monitoring point and an optimized monitoring point based on the three-dimensional layout of the rehabilitation simulation scene and the rehabilitation tracking device information; Determining the patient's activity state area based on the benchmark monitoring points and the rehabilitation patient's personalized characteristic information; Determining an environmental action area based on the three-dimensional layout of the rehabilitation simulation scene and the environmental impact factors; Based on the three-dimensional layout of the rehabilitation simulation scene, the optimized monitoring points, the spatiotemporal information of the rehabilitation activity and the patient activity state area, generating first rehabilitation behavior text data corresponding to the patient activity state area; Based on the three-dimensional layout of the rehabilitation simulation scene, the optimized monitoring points, the spatiotemporal information of the rehabilitation activity and the environmental action area, generating second rehabilitation behavior text data corresponding to the environmental action area; Determining the three-dimensional layout of the rehabilitation simulation scene and the spatiotemporal information of the rehabilitation activities in the local rehabilitation database according to the spatial configuration information and the recovery process information of the rehabilitation simulation scene includes: Acquire a first horizontal extension parameter and a second horizontal extension parameter in the spatial configuration information of the rehabilitation simulation scene in the local rehabilitation database, wherein the first horizontal extension parameter and the second horizontal extension parameter are horizontal extension parameters in different vertical directions in the plane area where the rehabilitation simulation scene is located; Acquire a height extension parameter of a height area in the spatial configuration information of the rehabilitation simulation scene; Determining a target rehabilitation training time according to the recovery progress information; Determining the spatiotemporal information of the rehabilitation activity according to the first horizontal extension parameter, the second horizontal extension parameter, the height extension parameter and the target rehabilitation training time of the rehabilitation simulation scene; The generating, in the local rehabilitation database, a derivative sample rehabilitation log text for the basic sample rehabilitation log text based on the first sample simulation text data and the second sample simulation text data comprises: Obtaining initialized text ratio parameters and regularization parameters, and determining a target mixing ratio according to the text ratio parameters, and determining target regularization mapping information according to the regularization parameters; Determining a mixing weight of the second sample simulated text data according to the target mixing ratio and the first sample simulated text data; Determine a mixing result according to the first sample simulated text data, the second sample simulated text data and the mixing weight; The mixed result is regularized and mapped according to the target regularized mapping information to generate a derived sample rehabilitation log text.
2. The intelligent rehabilitation log text processing method using artificial intelligence according to claim 1 is characterized in that: The rehabilitation tracking device information includes a first sensing range limit parameter, a device limit position parameter, a device label, a device quantity, and a device interval. The determining of the reference monitoring point and the optimized monitoring point based on the three-dimensional layout of the rehabilitation simulation scene and the rehabilitation tracking device information includes: Determining a reference monitoring point based on the three-dimensional layout of the rehabilitation simulation scene, the first sensing range limit parameter, and the device extreme position parameter; An optimized monitoring point is determined based on the reference monitoring point, the device tags, the number of devices, and the device intervals.
3. The intelligent rehabilitation log text processing method using artificial intelligence according to claim 1 is characterized in that: The environmental impact factor includes a third sensing range limit parameter and vertical boundary information of the environmental impact source; the environmental impact area includes a lateral coordinate of the environmental impact source, a longitudinal coordinate of the environmental impact source, and a vertical coordinate of the environmental impact source; and determining the environmental impact area based on the three-dimensional layout of the rehabilitation simulation scene and the environmental impact factor includes: Determining the lateral coordinates of the environmental impact source according to the lateral extension parameter and the third sensing range limit parameter in the three-dimensional layout of the rehabilitation simulation scene; Determining the longitudinal coordinate of the environmental impact source according to the axial extension parameter and the third sensing range limit parameter in the three-dimensional layout of the rehabilitation simulation scene; The vertical coordinates of the environmental impact source are determined based on the vertical definition information of the environmental impact source.
4. The intelligent rehabilitation log text processing method using artificial intelligence according to any one of claims 1 to 3, characterized in that: The basic sample rehabilitation log text at least includes the basic sample rehabilitation log text of the first training stage, and correspondingly, the derived sample rehabilitation log text at least includes the derived sample rehabilitation log text of the first training stage, and the network parameter optimization of the initialized rehabilitation log text mining network based on the basic sample rehabilitation log text and the derived sample rehabilitation log text is performed to generate an optimized rehabilitation log text mining network, including: caching the derived sample rehabilitation log texts of the first training phase into a sample sharing database; In the network learning process, optimizing the network parameters of the initialized rehabilitation log text mining network based on the basic sample rehabilitation log text and the derived sample rehabilitation log text in the first training phase; Obtaining a basic sample rehabilitation log text of a subsequent training phase, simulating the basic sample rehabilitation log text of the subsequent training phase in parallel in a simulation program, correspondingly generating a derived sample rehabilitation log text of the subsequent training phase, and loading the derived sample rehabilitation log text of the subsequent training phase into the sample sharing database; In the network learning process, based on the basic sample rehabilitation log text and the derived sample rehabilitation log text in the subsequent training phase, the network parameters of the rehabilitation log text mining network are continuously optimized until an optimized rehabilitation log text mining network is generated.
5. The intelligent rehabilitation log text processing method using artificial intelligence according to claim 4 is characterized in that: The method further comprises: Establish multiple simulation programs; The basic sample rehabilitation log text of the subsequent training stage is simulated in parallel in the simulation program to generate the derived sample rehabilitation log text of the subsequent training stage, including: loading basic sample rehabilitation log texts of the same training phase into the plurality of simulation programs; Based on the multiple simulation programs, the basic sample rehabilitation log texts loaded into the corresponding ones are simulated in parallel to generate derived sample rehabilitation log texts of the basic sample rehabilitation log texts for the same training stage.
6. The intelligent rehabilitation log text processing method using artificial intelligence according to any one of claims 1 to 3, characterized in that: The method further comprises: Outputting the three-dimensional layout of the rehabilitation simulation scene, the optimized monitoring points, the patient activity state area, the first rehabilitation behavior text data and the second rehabilitation behavior text data when generating the derived sample rehabilitation log text as simulation event data, and adding the simulation event data to the simulation event repository; When it is necessary to generate a derived sample rehabilitation log text for a derived base sample rehabilitation log text, obtaining a randomly generated new event simulation rate; When the new event simulation rate is less than the initialized preset simulation rate, randomly acquiring target simulation event data from the simulation event storage repository; Generate, in the local rehabilitation database, a derived sample rehabilitation log text for the derived basic sample rehabilitation log text based on the derived basic sample rehabilitation log text, the first rehabilitation behavior text data and the second rehabilitation behavior text data in the target simulation event data; And, when the new event simulation rate is not less than the initialized preset simulation rate, simulation is performed in the local rehabilitation database according to the rehabilitation activity characteristic parameters and the environmental influencing factors to generate derived first rehabilitation behavior text data corresponding to the patient activity state area and derived second rehabilitation behavior text data corresponding to the environmental effect area; Performing correlation analysis on the derived basic sample rehabilitation log text and the derived first rehabilitation behavior text data in the local rehabilitation database to generate derived first sample simulation text data; Acquire derived environment change data, and perform correlation analysis on the derived environment change data and the derived second rehabilitation behavior text data in the local rehabilitation database to generate derived second sample simulation text data; A derived derivative sample rehabilitation log text for the derived basic sample rehabilitation log text is generated in the local rehabilitation database based on the derived first sample simulation text data and the derived second sample simulation text data.
7. An intelligent rehabilitation guidance system, characterized in that: The intelligent rehabilitation guidance system includes a processor and a machine-readable storage medium, in which machine-executable instructions are stored. The machine-executable instructions are loaded and executed by the processor to implement the intelligent rehabilitation log text processing method using artificial intelligence as described in any one of claims 1-6.
