Postoperative rehabilitation nursing system for old people based on artificial intelligence

Abstract

The invention provides an old people postoperative rehabilitation nursing system based on artificial intelligence, and relates to the field of intelligent nursing, and the old people postoperative rehabilitation nursing system comprises the steps: firstly, carrying out the encryption processing and data normalization of a collected sentiment log data set of a target old people object, and obtaining the time sequence distribution of an encrypted sentiment log; and then performing content summarization and semantic embedding on the time sequence distribution of the encrypted sentiment logs by adopting a data analysis mode based on artificial intelligence to obtain time sequence distribution of sentiment log summary semantic features, and obtaining an emotion recognition result based on context causal association coding features of the time sequence distribution of the sentiment log summary semantic features. And finally, judging whether a psychological intervention prompt signal is generated based on an emotion recognition result. Therefore, by continuously monitoring and analyzing the emotion fluctuation and the law of the senior citizens, the emotion state can be more accurately captured, and then the timeliness, the effectiveness and the intelligent degree of psychological intervention are improved.

Description

Technical Field

[0001] This application relates to the field of intelligent nursing, and more specifically, to a postoperative rehabilitation nursing system for the elderly based on artificial intelligence. Background Art

[0002] In the process of postoperative rehabilitation nursing for the elderly, the importance of psychological intervention cannot be ignored. Surgery and its recovery period often bring physical discomfort and psychological pressure to the elderly, such as negative emotions like worry about the surgical outcome and anxiety about future self-care ability. These emotions not only affect the patient's sleep quality and appetite, but may also lead to a decline in the immune system function, delay the wound healing speed, and increase the risk of postoperative complications. Therefore, implementing effective psychological intervention measures in the postoperative rehabilitation nursing of the elderly is of great significance for improving the patient's mental health status, promoting their positive attitude towards the rehabilitation process, and enhancing the quality of life.

[0003] However, traditional psychological intervention in postoperative rehabilitation nursing for the elderly faces many challenges. First, traditional psychological assessments mainly rely on the observations of medical staff and the self-reports of patients. This method may have deviations due to individual differences and it is difficult to comprehensively and accurately capture the real emotional changes of each patient. Second, due to the limitation of human resources, it is difficult to continuously monitor and analyze the emotional states of each elderly patient, thus missing some of the best opportunities for early psychological intervention. In addition, previous sentiment analyses mostly stay at the surface level and cannot deeply understand the underlying emotional fluctuations and their evolution rules behind the patient's emotional logs. The existence of these problems greatly reduces the effectiveness of traditional psychological intervention in postoperative rehabilitation nursing for the elderly.

[0004] Therefore, an optimized postoperative rehabilitation nursing plan is expected. Summary of the Invention

[0005] To solve the above technical problems, this application is proposed. An embodiment of this application provides a postoperative rehabilitation nursing system for the elderly based on artificial intelligence.

[0006] According to one aspect of this application, there is provided a postoperative rehabilitation nursing system for the elderly based on artificial intelligence, which includes:

[0007] A data acquisition and preprocessing module, configured to collect the emotional logs of a target elderly object to obtain a data set of emotional logs, and perform encryption processing and data regularization on the data set of emotional logs to obtain the time series distribution of the encrypted emotional logs;

[0008] An emotional state recognition module for performing emotion analysis based on text semantics on the time series distribution of the encrypted emotional logs to obtain an emotional state recognition result, where the emotional state recognition module includes: a log time series summary embedding unit for performing content summary and semantic embedding on the time series distribution of the encrypted emotional logs to obtain the time series distribution of the summary semantic embedding encoding features of the emotional logs; a log time series encoding unit for performing emotional log context causal association encoding on the time series distribution of the summary semantic embedding encoding features of the emotional logs to obtain the time series evolution semantic encoding features of the emotional state; a recognition result generation unit for obtaining the emotional state recognition result based on the time series evolution semantic encoding features of the emotional state;

[0009] An intervention signal generation module for determining whether to generate a psychological intervention prompt signal according to the emotional state recognition result.

[0010] Compared with the prior art, the artificial intelligence-based postoperative rehabilitation nursing system for the elderly provided by the present application first performs encryption processing and data regularization on the collected emotional log data set of the target elderly object to obtain the time series distribution of the encrypted emotional logs, then uses an artificial intelligence-based data analysis method to perform content summary and semantic embedding on the time series distribution of the encrypted emotional logs to obtain the time series distribution of the summary semantic features of the emotional logs, and obtains an emotion recognition result based on the context causal association encoding features of the time series distribution of the summary semantic features of the emotional logs, and finally determines whether to generate a psychological intervention prompt signal based on the emotion recognition result. In this way, by continuously monitoring and analyzing the emotional fluctuations and their patterns of the elderly, it helps to more accurately capture the emotional state, thereby improving the timeliness, effectiveness, and intelligence of psychological intervention. Description of the Drawings

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained according to these drawings without creative efforts. In the drawings:

[0012] Figure 1 It is a system block diagram of an artificial intelligence-based postoperative rehabilitation nursing system for the elderly according to an embodiment of the present application.

[0013] Figure 2 It is a block diagram of a data acquisition and preprocessing module in an artificial intelligence-based postoperative rehabilitation nursing system for the elderly according to an embodiment of the present application.

[0014] Figure 3It is a block diagram of an emotional state recognition module in an artificial intelligence-based postoperative rehabilitation nursing system for the elderly according to an embodiment of the present application.

[0015] Figure 4 It is a block diagram of a log time series summary embedding unit in an artificial intelligence-based postoperative rehabilitation nursing system for the elderly according to an embodiment of the present application.

[0016] Figure 5 It is a block diagram of a log time series encoding unit in an artificial intelligence-based postoperative rehabilitation nursing system for the elderly according to an embodiment of the present application. Detailed implementation manners

[0017] Next, exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein.

[0018] It should be noted that in the present application, all actions of obtaining signals, information, or data are carried out on the premise of complying with the corresponding data protection regulations and policies of the country where the location is located and obtaining the authorization given by the owner of the corresponding device.

[0019] In the postoperative rehabilitation nursing of the elderly, psychological intervention is of great significance. During the operation and recovery period, elderly patients often have negative emotions such as worry about the operation result and anxiety about future self-care ability due to physical discomfort and psychological pressure. These emotions not only affect sleep and appetite, but also may weaken the immune function, delay wound healing, and increase the risk of postoperative complications. Therefore, implementing effective psychological intervention is crucial for improving the mental health of patients, promoting active rehabilitation, and enhancing the quality of life.

[0020] However, traditional psychological intervention faces many challenges in the postoperative rehabilitation nursing of the elderly. On the one hand, its evaluation methods mainly rely on the observation of medical staff and the self-report of patients, which are easily affected by individual differences and are difficult to comprehensively and accurately capture the real emotional changes of patients. On the other hand, due to limited human resources, it is difficult to continuously monitor the emotional state of patients, and often misses the best intervention opportunity. In addition, traditional sentiment analysis mostly stays on the surface and is difficult to deeply analyze the emotional fluctuations and evolution laws behind the emotional logs of patients. The existence of these problems weakens the effect of traditional psychological intervention in the postoperative rehabilitation nursing of the elderly.

[0021] To address the above problems, the technical concept of this application is to first collect the emotional log dataset of the target elderly object, then perform encryption processing and data regularization on it to obtain the time series distribution of the encrypted emotional log, and then use artificial intelligence-based data analysis methods to summarize the content and embed semantics of the time series distribution of the encrypted emotional log to obtain the time series distribution of the summary semantic features of the emotional log, and obtain the emotion recognition result based on the context causal association coding features of the time series distribution of the summary semantic features of the emotional log. Finally, it is determined whether to generate a psychological intervention prompt signal based on the emotion recognition result. In this way, through continuous emotion monitoring and analysis, and accurately mining the deep emotional fluctuations and their evolution laws behind the emotional logs of the elderly in the analysis, the real emotional state of the elderly can be captured more timely, comprehensively, and accurately, which is beneficial to improving the timeliness, effectiveness, and intelligence of psychological intervention.

[0022] Figure 1 FIG. is a system block diagram of an artificial intelligence-based postoperative rehabilitation nursing system for the elderly according to an embodiment of the present application. As Figure 1 shown, in the artificial intelligence-based postoperative rehabilitation nursing system 100 for the elderly, it includes: a data acquisition and preprocessing module 110, configured to collect the emotional logs of the target elderly object to obtain a dataset of emotional logs, and perform encryption processing and data regularization on the dataset of the emotional logs to obtain the time series distribution of the encrypted emotional logs; an emotional state recognition module 120, configured to perform text-semantic-based emotion analysis on the time series distribution of the encrypted emotional logs to obtain an emotional state recognition result; an intervention signal generation module 130, configured to determine whether to generate a psychological intervention prompt signal according to the emotional state recognition result.

[0023] In an embodiment of the present application, the data acquisition and preprocessing module 110 is configured to collect the emotional logs of the target elderly object to obtain a dataset of emotional logs, and perform encryption processing and data regularization on the dataset of the emotional logs to obtain the time series distribution of the encrypted emotional logs. Specifically, Figure 2 FIG. is a block diagram of the data acquisition and preprocessing module in the artificial intelligence-based postoperative rehabilitation nursing system for the elderly according to an embodiment of the present application. As Figure 2 shown, the data acquisition and preprocessing module 110 includes: an emotional log acquisition unit 111, configured to collect the emotional logs of the target elderly object to obtain the dataset of emotional logs; an emotional log encryption processing unit 112, configured to perform encryption processing on each emotional log in the dataset of the emotional logs to obtain a dataset of encrypted emotional logs; an emotional log time series distribution adjustment unit 113, configured to perform data distribution pattern adjustment on the dataset of the encrypted emotional logs based on the time dimension according to the time stamp to obtain the time series distribution of the encrypted emotional logs.

[0024] In the embodiment of the present application, the emotional log collection unit 111 is used to collect the emotional logs of the target elderly object to obtain a data set of the emotional logs. It should be understood that the data set of the collected emotional logs mainly includes the record information of the emotional experiences of the target elderly object at different time points, and may also include the description information of the events that trigger emotions. In addition, it will also include the correlation record information between physical conditions and emotions, etc. Specifically, the record of emotional experience in the emotional log is the most direct basis for judging the emotional state. If the proportion of negative emotions in the record is high and the intensity is large, it indicates that the elderly person may be in a negative emotional state. For example, if there are records of high-intensity anxiety or sadness for several consecutive days, key attention is required. The description of the event that triggers emotions can help the model understand the background of the emotion generation. When the events that trigger negative emotions appear frequently or the intensity increases, such as the wound always hurting or often having unpleasant exchanges with others, this may indicate that the emotion will continue to develop negatively. The correlation record between physical conditions and emotions can dynamically help the model evaluate the emotional state. For example, if the physical condition continues to be poor and the emotion has been low, or the improvement of the physical condition does not bring about an improvement in the emotion, this may indicate the existence of potential emotional problems. Generally speaking, through in-depth analysis of the data set of emotional logs, it is possible to more accurately identify the emotional fluctuations of the elderly object that need to be concerned, and automatically generate a psychological intervention prompt signal when necessary to ensure that medical staff can intervene in time to provide necessary support and help for elderly patients.

[0025] In the embodiment of the present application, the emotional log encryption processing unit 112 is used to perform encryption processing on each emotional log in the data set of the emotional logs to obtain a data set of encrypted emotional logs. Correspondingly, considering that the emotional logs contain a large amount of personal sensitive information, such as the emotional state and inner feelings of the target elderly object, if this information is leaked, it may infringe on the personal privacy of the elderly object. Based on this, the present application performs encryption processing on each emotional log in the data set of the emotional logs to effectively protect the privacy and security of the elderly. It is worth mentioning that since the encryption process ensures the integrity of the data, in the subsequent detailed data processing stage, it is all based on the original data that has not been tampered with, which can make the finally obtained emotional state recognition result more reliable, and thus ensure the accuracy of the generated psychological intervention prompt signal.

[0026] In an embodiment of the present application, the emotional log time-series distribution adjustment unit 113 is configured to adjust the data distribution pattern of the encrypted emotional log dataset based on the time dimension according to timestamps to obtain the time-series distribution of the encrypted emotional log. It should be understood that the emotional state of the elderly is not fixed but fluctuates continuously over time. By adjusting the data distribution pattern of the encrypted emotional log based on the time dimension according to timestamps, the emotional logs can be arranged in chronological order, so that the changes in the emotions of the elderly at different stages of postoperative rehabilitation, different time periods of a day, and other time scales can be systematically observed. For example, some elderly people may have greater emotional fluctuations in the early postoperative period, and as their bodies gradually recover, their emotions tend to stabilize; or during the day, they are more likely to experience anxiety at night. That is, organizing the data in chronological order can help the model clearly capture the emotional change patterns, and then better judge the development direction of the emotional state of the elderly object.

[0027] In an embodiment of the present application, the emotional state recognition module 120 is configured to perform text-semantic-based emotional analysis on the time-series distribution of the encrypted emotional log to obtain an emotional state recognition result. Specifically, Figure 3 It is a block diagram of the emotional state recognition module in the artificial-intelligence-based postoperative rehabilitation nursing system for the elderly according to an embodiment of the present application. As Figure 3 shown, the emotional state recognition module 120 includes: a log time-series summary embedding unit 121, configured to perform content summary and semantic embedding on the time-series distribution of the encrypted emotional log to obtain the time-series distribution of the emotional log summary semantic embedding coding features; a log time-series coding unit 122, configured to perform emotional log context causal association coding on the time-series distribution of the emotional log summary semantic embedding coding features to obtain the emotional state time-series evolution semantic coding features; and a recognition result generation unit 123, configured to obtain the emotional state recognition result based on the emotional state time-series evolution semantic coding features.

[0028] In an embodiment of the present application, the log time-series summary embedding unit 121 is configured to perform content summary and semantic embedding on the time-series distribution of the encrypted emotional log to obtain the time-series distribution of the emotional log summary semantic embedding coding features. Specifically, Figure 4 It is a block diagram of the log time-series summary embedding unit in the artificial-intelligence-based postoperative rehabilitation nursing system for the elderly according to an embodiment of the present application. As Figure 4As shown, the log time series summary embedding unit 121 includes: an emotional log summary generation subunit 1211, which is configured to input each encrypted emotional log in the time series distribution of the encrypted emotional logs into a content summary encoder based on a large language model to obtain a time series distribution of emotional log summaries; and an emotional log summary embedding encoding subunit 1212, which is configured to perform semantic embedding encoding on each emotional log summary in the time series distribution of the emotional log summaries to obtain a time series distribution of emotional log summary semantic embedding encoding vectors as the time series distribution of the emotional log summary semantic embedding encoding features.

[0029] In an embodiment of the present application, the emotional log summary generation subunit 1211 is configured to input each encrypted emotional log in the time series distribution of the encrypted emotional logs into a content summary encoder based on a large language model to obtain a time series distribution of emotional log summaries. It should be understood that the time series distribution of the encrypted emotional logs contains a large amount of detailed information, which may make subsequent analysis complex and redundant. Considering that the large language model is an artificial intelligence model based on deep learning and trained with a large amount of text data, it can understand and generate natural language. By learning language structure and semantic information, it can not only generate coherent and meaningful text, but also perform context-aware tasks, such as extracting key information or summarizing the main content from a long document. Based on this, in the present application, each encrypted emotional log in the time series distribution of the encrypted emotional logs is input into a content summary encoder based on a large language model to simplify these detailed emotional logs, filter out events and emotional expressions closely related to mood changes, and remove irrelevant information, making the data easier to process and analyze. The obtained time series distribution of emotional log summaries is a concise data representation form that contains key emotional information. Compared with the original encrypted emotional logs, the summary is more refined and facilitates the model to quickly understand and analyze the emotional states of the elderly at different time points.

[0030] The following is a detailed elaboration of a specific implementation process of "inputting each encrypted emotional log in the time series distribution of the encrypted emotional logs into a content summary encoder based on a large language model to obtain a time series distribution of emotional log summaries":

[0031] First, preliminary preparation work needs to be carried out. This includes selecting a suitable large language model. There are various large language models available in the market, such as GPT series of OpenAI, BERT of Google, etc. When selecting, multiple factors need to be comprehensively considered, such as the language understanding and generation ability of the model, and its performance in sentiment analysis tasks. At the same time, it is also necessary to obtain each encrypted sentiment log data from the temporal distribution of the encrypted sentiment logs, and decrypt it using the corresponding encryption key to restore it to the original text format. After that, clean the decrypted text, remove irrelevant special characters, garbled codes, etc., and unify the text format to ensure the standardization and consistency of the text, laying a good foundation for subsequent processing.

[0032] Next is the model parameter setting section. If calling the API, the parameters need to be set according to the API documentation of the model. Taking the call to GPT-3.5 as an example, the temperature parameter is used to control the randomness of the generated text, and its value range is usually between 0 and 1. The closer the value is to 0, the more certain the generated result; the maximum number of generated tokens (max_tokens) is used to limit the length of the generated text. If it is a locally deployed model, similar parameters need to be set according to the configuration file of the model to optimize the generation effect.

[0033] After completing the preparation and parameter setting, it enters the stage of interacting with the large language model. Use programming languages (such as Python) and corresponding libraries (such as the openai library of OpenAI, the transformers library of HuggingFace) to write code to achieve interaction with the large language model. Build a request body in the code and pass the cleaned sentiment log text as input to the model.

[0034] When the emotional log text enters the large language model, a series of complex operations will be carried out inside the model to achieve content summary encoding. First is text encoding. The large language model will split the input emotional log text into individual tokens, and then map each token to a vector through the embedding layer. These vectors contain the semantic information of the tokens, and the formed sequence becomes the basic data for the subsequent processing of the model. Next, the model uses its own attention mechanism to process the token vector sequence. The attention mechanism enables the model to focus on the parts of the text related to emotions and key events, such as words describing emotions (like "anxiety", "happy"), events triggering emotions (such as "the pain of the surgical wound has increased"), and related information such as time and location. Through the multi-layer Transformer architecture, the model conducts in-depth semantic understanding of the text, mining the implicit information and semantic associations therein. After that, based on the extracted key information and semantic understanding of the text, the model starts to generate the emotional log summary. Based on the language generation patterns it has learned, it predicts the next most likely token and gradually constructs the summary text. During the generation process, it will fully consider the context information to ensure that the generated summary is coherent and logical. In addition, the generated summary may be inaccurate or not concise enough, and the model will use its own optimization mechanism to adjust it. For example, it adopts the method of reinforcement learning and optimizes the generated summary according to a certain reward strategy (such as the matching degree between the generated summary and the key information of the original text, the fluency of the language, etc.) to improve the quality of the summary.

[0035] Finally, receive the emotional log summary result returned by the large language model and organize it. Arrange the generated summary in the time order of the encrypted emotional log to form the time series distribution of the emotional log summary. To ensure the accuracy of the summary, it is also necessary to verify the generated summary, checking whether the summary accurately reflects the key emotions and events of the original emotional log and whether there are logical errors or information omissions. Automatic evaluation metrics can be used, such as the ROUGE metric, to calculate the similarity between the generated summary and the reference summary (if any) to evaluate the quality of the summary. Through this series of implementation operations, the time series distribution of the emotional log summary can be obtained from the time series distribution of the encrypted emotional log, thereby providing strong support for subsequent emotion analysis and intervention.

[0036] In an embodiment of the present application, the emotional log summary embedding coding subunit 1212 is used to perform semantic embedding coding on each emotional log summary in the temporal distribution of the emotional log summary to obtain the temporal distribution of the emotional log summary semantic embedding coding vector as the temporal distribution of the emotional log summary semantic embedding coding feature. Specifically, in an embodiment of the present application, the emotional log summary embedding coding subunit is used to: use a semantic encoder based on a bidirectional recurrent neural network model to perform semantic embedding coding on each emotional log summary in the temporal distribution of the emotional log summary to obtain the temporal distribution of the emotional log summary semantic embedding coding vector. Accordingly, considering that the emotional log summary is text information, it is difficult to capture deep emotional and semantic associations based on text analysis directly. In order to enable the computer to more effectively process and understand the complex semantic information contained therein, so as to more accurately grasp the emotional state of the elderly, it is necessary to perform semantic embedding coding on each emotional log summary in the temporal distribution of the emotional log summary in the present application to obtain the temporal distribution of the emotional log summary semantic embedding coding vector as the temporal distribution of the emotional log summary semantic embedding coding feature. The semantic embedding coding operation can mine the semantic information and potential features in the emotional log summary. For example, it can identify the semantic similarity and relevance between different words and phrases. For emotion-related words, such as "anxiety", "worry", "fear", etc., the close connection between them can be found in the vector space through encoding, so as to more accurately understand the semantic connotation behind the emotional expression of the elderly. In particular, in a specific embodiment of the present application, a semantic encoder based on a bidirectional recurrent neural network model is used to perform semantic embedding encoding on each emotion log summary in the temporal distribution of the emotion log summary to obtain the temporal distribution of the semantic embedding encoding vector of the emotion log summary. Those of ordinary skill in the art should know that the bidirectional recurrent neural network consists of two unidirectional recurrent neural networks, one is forward, processing data from front to back according to the time order of the input sequence; the other is reverse, processing data in order from back to front. In the temporal distribution of the emotion log summary, it may contain records of emotion changes over a long time span. The use of a bidirectional recurrent neural network model can better mine the law of emotion evolution through bidirectional information processing to avoid information loss or forgetting. For example, the factors that trigger emotions (which may be at the front of the emotion log summary) and the subsequent impacts caused by emotions (which may be at the back of the emotion log summary) can be effectively encoded into the semantic embedding coding vector of the emotion log summary through the bidirectional recurrent neural network model, thereby improving the accuracy of the semantic understanding of the emotion log summary.

[0037] The following is a detailed elaboration of a specific implementation process of "performing semantic embedding encoding on each sentiment log summary in the temporal distribution of the sentiment log summary using a semantic encoder based on a bidirectional recurrent neural network model to obtain the temporal distribution of the sentiment log summary semantic embedding encoding vectors":

[0038] First is data preprocessing. In this step, each sentiment log summary needs to be comprehensively cleaned, removing special characters and punctuation marks, and at the same time unifying the case of the text to reduce the interference caused by lexical diversity. After cleaning, tokenize the text. For Chinese text, tools such as Jieba can be used for tokenization, and for English text, libraries such as NLTK and SpaCy can be used to split the text into individual words or tokens. Based on the tokenization results, count all tokens to build a vocabulary, assign a unique integer index to each token, and also set special indices to represent tokens not present in the vocabulary. Then, convert the tokens in each sentiment log summary into an integer index sequence according to the vocabulary. If the sequence lengths are inconsistent, pad (usually padding with a specific index value 0 at the beginning and end of the sequence) or truncate them to make them of equal length for subsequent processing.

[0039] After data preprocessing, start building a bidirectional recurrent neural network (BiRNN) semantic encoder model. First, design the model architecture, which generally includes an input layer, a bidirectional recurrent layer, and an output layer. The input layer receives the preprocessed text sequence and converts it into a word vector representation, which can use pre-trained word embeddings (such as Word2Vec, GloVe) or randomly initialize the word embedding matrix. The bidirectional recurrent layer processes the input sequence forward and backward to capture the forward and backward context information, and concatenates the forward and backward outputs to obtain a richer semantic representation. The output layer maps the output of the bidirectional recurrent layer to the required dimension to obtain the semantic embedding encoding vector of the sentiment log summary. Finally, initialize the weights and biases of the model, and commonly used methods such as Xavier initialization and He initialization are used to help the model converge faster.

[0040] After the model is constructed, it enters the training stage. First, divide the dataset. The preprocessed data is divided into a training set, a validation set, and a test set according to common ratios (70%-80% for the training set, 10%-15% for the validation set, and 10%-15% for the test set). The training set is used for model training, the validation set is used to adjust hyperparameters (such as learning rate, batch size), and the test set is used to evaluate the final performance. Then define the loss function and the optimizer. Select appropriate loss functions (such as mean squared error loss, cross-entropy loss) and optimizers (such as stochastic gradient descent, Adam) according to the task. The optimizer is used to minimize the loss function value during training. Subsequently, perform multiple rounds of iterative training on the training set. In each round of iteration, batch the input data into the model, calculate the output result and calculate the loss value according to the loss function, and use the optimizer to update the model parameters based on the loss value. During the training process, regularly evaluate the model performance on the validation set, monitor the loss value and other evaluation metrics (such as accuracy, recall rate), and prevent overfitting. If the performance of the validation set no longer improves, stop training early. After training is completed, use the test set to evaluate the performance of the trained model, calculate relevant metrics such as accuracy, recall rate, and F1 value, and judge whether the model performance meets the requirements based on the evaluation results. If the model performance does not meet the expectations, you can try to adjust the hyperparameters (such as learning rate, hidden layer dimension), increase the amount of training data, and continuously optimize to improve the model performance and stability, ensuring that the semantic encoder based on the bidirectional recurrent neural network model can efficiently and accurately perform semantic embedding encoding on the emotional log summary.

[0041] Finally, load the model with trained and saved parameters, and input each emotional log summary in the time series distribution of the emotional log summary into the loaded model after processing according to the preprocessing process. The model processes the input text sequence, extracts context information through the bidirectional recurrent layer, and finally obtains the corresponding semantic embedding encoding vector at the output layer. Arrange these vectors in chronological order to form the time series distribution of the semantic embedding encoding vectors of the emotional log summary.

[0042] In the embodiment of the present application, the log time series encoding unit 122 is used to perform emotional log context causal association encoding on the time series distribution of the semantic embedding encoding features of the emotional log summary to obtain the emotional state time series evolution semantic encoding features. Specifically, Figure 5 It is a block diagram of the log time series encoding unit in the artificial intelligence-based postoperative rehabilitation nursing system for the elderly according to the embodiment of the present application. As Figure 5As shown, the log time series encoding unit 122 includes: an emotional log summary implicit semantic feature mining subunit 1221, configured to perform emotional log summary implicit semantic feature mining on the time series distribution of the semantic embedding encoding vector of the emotional log summary to obtain the time series distribution of the semantic deep implicit embedding encoding vector of the emotional log summary; an emotional log summary semantic causal analysis subunit 1222, configured to perform emotional log summary semantic causal association factor calculation and emotional log summary semantic causal triggering on the time series distribution of the semantic deep implicit embedding encoding vector of the emotional log summary to obtain an emotional log summary semantic causal association topological feature matrix; an emotional state time series evolution semantic encoding feature generation subunit 1223, configured to, based on the emotional log summary semantic causal association topological feature matrix, perform log semantic context dynamic walk encoding fusion on the time series distribution of the semantic deep implicit embedding encoding vector of the emotional log summary and the time series distribution of the semantic embedding encoding vector of the emotional log summary to obtain an emotional state time series evolution semantic encoding vector as the emotional state time series evolution semantic encoding feature.

[0043] It should be understood that there are certain limitations in the time series data processing method of the bidirectional recurrent neural network model. Specifically, taking the association between wound healing and emotions as an example, there may be records such as "the wound infection worsens, the pain intensifies, resulting in extremely low emotions". Although the traditional processing method can detect the sequence of changes in wound conditions and emotions over time, it is difficult to clearly model the causal relationship of how the worsening of wound infection directly leads to low emotions. Moreover, the context of the time series distribution of the semantic embedding encoding features of the emotional log summary presents multi-level associations, including not only explicit semantic information (such as directly expressed emotions), but also implicit complex background information and potential psychological states. The traditional single-layer analysis method is difficult to effectively distinguish these different levels of information, resulting in incomplete and one-sided integration of the context semantics of the emotional log summary. Based on this, in this application, emotional log context causal association encoding is performed on the time series distribution of the semantic embedding encoding features of the emotional log summary to obtain emotional state time series evolution semantic encoding features. In this way, through the emotional log context causal association encoding operation, not only can the causal relationship be accurately sorted out, but also the in-depth mining of multi-level semantics of the original features is realized, which is beneficial for the model to more sensitively capture the potential trend of emotional changes, and thus improve the understanding ability of the emotional state of the elderly object.

[0044] Specifically, first, emotional log summary implicit semantic feature mining is performed on the time series distribution of the semantic embedding encoding vector of the emotional log summary to obtain the time series distribution of the semantic deep implicit embedding encoding vector of the emotional log summary. The above process can be expressed by the formula:

[0045] O = {x1, x2,..., x i ,..., xn}

[0046] v i = Sigmoid[Conv 1×1 (x i )]

[0047] D = {v1, v2,..., v i ,..., v n}

[0048] where O is the time series distribution of the semantic embedding encoding vectors of the emotional log summary, x1, x2, x i and x n are respectively the 1st, 2nd, ith, and nth semantic embedding encoding vectors of the emotional log summary in the time series distribution of the semantic embedding encoding vectors of the emotional log summary, Conv 1×1 is point convolution encoding, Sigmoid is the activation function of convolution encoding, v1, v2, v i and v n are respectively the 1st, 2nd, ith, and nth semantic deep implicit embedding encoding vectors of the emotional log summary in the time series distribution of the semantic deep implicit embedding encoding vectors of the emotional log summary, and D is the time series distribution of the semantic deep implicit embedding encoding vectors of the emotional log summary.

[0049] It should be understood that the original semantic embedding encoding vectors of the emotional log summary may contain a large amount of redundant information and noise, which will interfere with subsequent causal association analysis. Through the mining of implicit semantic features of the emotional log summary, deep and meaningful semantic features can be extracted from these vectors to reduce unnecessary information interference. That is, the generated semantic deep implicit embedding encoding vectors of the emotional log summary can more accurately reflect the core features of the emotional state of the elderly object, and can provide richer clues for the model to comprehensively understand the emotional state of the elderly.

[0050] Specifically, in the embodiment of the present application, the semantic causal analysis sub-unit of the emotional log summary includes: a first-level sub-unit for calculating the semantic causal association factor of the emotional log summary, which is used to calculate the semantic causal association factor between any two semantic deep implicit embedding encoding vectors of the emotional log summary in the time series distribution of the semantic deep implicit embedding encoding vectors of the emotional log summary to obtain a semantic causal association topology matrix composed of multiple semantic causal association factors of the emotional log summary; a first-level sub-unit for triggering the semantic causal relationship of the emotional log summary, which is used to perform semantic causal triggering of the emotional log summary based on a gating function on the semantic causal association topology matrix of the emotional log summary to obtain the semantic causal association topology feature matrix.

[0051] More specifically, in the embodiments of the present application, the first-level sub-unit for calculating the semantic causal association factor of the emotional log summary is configured to: calculate the association matrix between any two semantic depth implicit embedding coding vectors in the time series distribution of the semantic depth implicit embedding coding vectors of the emotional log summary to obtain the time series distribution of the semantic depth association embedding coding matrix of the emotional log summary; calculate the semantic causal association factor of each semantic depth association embedding coding matrix in the time series distribution of the semantic depth association embedding coding matrix of the emotional log summary to obtain the semantic causal association topology matrix of the emotional log summary composed of multiple semantic causal association factors of the emotional log summary, wherein the semantic causal association factor of the emotional log summary is calculated from the maximum value, mean value, variance and causal association bias term of the semantic depth association embedding coding matrix of the emotional log summary; in response to the variance of the semantic depth association embedding coding matrix of the emotional log summary being greater than or equal to a preset threshold, taking the weighted average of the distances between any two semantic depth implicit embedding coding vectors in the time series distribution of the semantic depth implicit embedding coding vectors of the emotional log summary as the causal association bias term, and in response to the variance of the semantic depth association embedding coding matrix of the emotional log summary being less than the preset threshold, taking the weighted average of the semantic depth association embedding coding matrix of the emotional log summary as the causal association bias term. The above process can be represented by the following formula:

[0052]

[0053]

[0054] Wherein, v i , v j are respectively the i-th and j-th semantic depth implicit embedding coding vectors in the time series distribution of the semantic depth implicit embedding coding vectors of the emotional log summary, is matrix multiplication, v j T is the transposed vector of v j , M i-j is the semantic depth association embedding coding matrix between v i and v j , σ 2 (M i-j ) is the variance of M i-j , max(M i-j ) is to take the maximum value in M i-j , μ(M i-j ) is the mean value of M i-j , λ is the causal association bias term, t i-j is the semantic causal association factor of the emotional log summary corresponding to M i-j , d(vi , v j ) is v i and v j is the distance between, L is the number of vectors in D, ε is a preset threshold, α and β are weighted hyperparameters, t 1-1 , t 1-n , t n-1 and t n-n are respectively the semantic causal association factors of the emotional log summary at each position in the semantic causal association topology matrix of the emotional log summary, and T is the semantic causal association topology matrix of the emotional log summary.

[0055] Correspondingly, in order to understand the causal mechanism behind the emotional changes at different time points, it is necessary to quantify the causal association between the semantic depth implicit embedding coding vectors of the emotional log summary between any two time points, that is, it is necessary to calculate the semantic causal association factor of the emotional log summary between any two semantic depth implicit embedding coding vectors of the emotional log summary. The semantic causal association factor of the emotional log summary calculated through the causal association energy metric function can accurately deduce the causal association between two features in the original time series distribution, helping the model to deeply explore the internal relationship between different features. The constructed semantic causal association topology matrix of the emotional log summary can present these causal relationships in a quantitative manner, facilitating the subsequent in-depth analysis of the internal mechanism of the emotional changes of the elderly.

[0056] Specifically, by regarding the causal association of low-level emotional changes in a complex system as a molecular-level relationship inferred based on statistical correlation, it is possible to further perform intervention prediction of the causal association energy of emotional changes on the basis of the global fine-grained statistical association representation, so as to study the fine-grained structure and its dynamic regulation of the causal association of emotional changes with a high-dimensional and heterogeneous representation based on the omics of causal relationships of emotional changes. Among them, when the aggregated distribution representation of the causal graph is greater than the preset threshold, there is a bias in the integration of source data based on the matrix representation of the node effect of the semantic causal association factor of the emotional log summary, while when the aggregated distribution representation of the causal graph is less than the preset threshold, the condensed structure modeling can be directly carried out through the integration and compression of the feature pattern. In this way, not only can the causal association energy of the emotional changes in the system be encoded and described, but also the results of the implicit causal intervention prediction of the emotional changes can be condensed, so as to obtain a more efficient revelation of the key causal association of the emotional changes.

[0057] Next, perform semantic causal triggering of the emotional log summary based on the gating function on the semantic causal association topology matrix of the emotional log summary to obtain the semantic causal association topology feature matrix of the emotional log summary. The above process can be expressed by the formula:

[0058]

[0059] where, ti-j is M i-j The corresponding semantic causal association factor of the emotional log summary, T is the semantic causal association topology matrix of the emotional log summary, softmax is a non-linear activation function, τ is a normalization threshold, and f trigger (T) is the gated activation process for T, and M is the semantic causal association topology feature matrix of the emotional log summary.

[0060] It should be understood that the dynamic gating mechanism and non-linear activation function in the semantic causal trigger operation of the emotional log summary can perform more delicate feature extraction and modeling on the causal topology relationship of emotional changes. Specifically, the dynamic gating mechanism can accurately identify the key causal paths in the complex and changing semantic dynamics of the emotional log summary, thereby highlighting the important emotional change associations, weakening the noise interference, and making the causal relationship of emotional changes clearer and more definite; the non-linear activation function, by introducing non-linear characteristics, greatly enhances the model's ability to capture the high-order regularities in the complex causal structure of emotional changes, and improves the model's understanding and analysis ability of the causal relationship of emotional changes. That is, the generated semantic causal association topology feature matrix of the emotional log summary successfully filters out a large amount of interference information and highlights the key causal paths. Among the many factors affecting the emotions of the elderly, it can accurately identify which are the core causal factors that truly dominate emotional changes and which are secondary or insignificant factors. This can provide more accurate and key information for subsequent processing, and thus improve the accuracy and effectiveness of the model's analysis of the causal relationship of emotional changes.

[0061] Specifically, in the embodiment of the present application, the emotion state time-series evolution semantic encoding feature generation sub-unit is used to: perform log semantic dynamic random walk encoding based on the graph convolutional neural network model on the time-series distributions of the semantic causal association topology feature matrix of the emotional log summary and the semantic embedding encoding vector of the emotional log summary to obtain the surface context dynamic random walk semantic encoding vector of the emotional log summary. This process can be expressed by the formula:

[0062]

[0063] where x i is the i-th semantic embedding encoding vector in the time-series distribution of the semantic embedding encoding vector of the emotional log summary, M is the semantic causal association topology feature matrix of the emotional log summary, GCN is the graph convolutional processing, and H surface is the surface context dynamic random walk semantic encoding vector of the emotional log summary;

[0064] Perform the log semantic dynamic random walk encoding based on the graph convolutional neural network model on the temporal distributions of the semantic causal association topological feature matrix of the emotional log summary and the semantic depth implicit embedding encoding vector of the emotional log summary to obtain the hidden layer context dynamic random walk semantic encoding vector of the emotional log summary. This process can be expressed by the formula:

[0065]

[0066] where, v i is the i-th semantic depth implicit embedding encoding vector in the temporal distribution of the semantic depth implicit embedding encoding vector of the emotional log summary, M is the semantic causal association topological feature matrix of the emotional log summary, GCN is the graph convolutional processing, and H hidden is the hidden layer context dynamic random walk semantic encoding vector of the emotional log summary;

[0067] Fuse the hidden layer context dynamic random walk semantic encoding vector of the emotional log summary and the surface layer context dynamic random walk semantic encoding vector of the emotional log summary to obtain the semantic encoding vector of the temporal evolution of the emotional state. This process can be expressed by the formula:

[0068] H final = γ · H surface + (1 - γ) · H hidden

[0069] where, H surface is the surface layer context dynamic random walk semantic encoding vector of the emotional log summary, H hidden is the hidden layer context dynamic random walk semantic encoding vector of the emotional log summary, γ is the fusion weighting parameter, and H final is the semantic encoding vector of the temporal evolution of the emotional state.

[0070] Correspondingly, in order to integrate the explicit semantic information in the original emotional log summary and provide a clear and definite basis for the model to intuitively understand the emotional state of the elderly, it is necessary to perform the log semantic dynamic random walk encoding process based on the graph convolutional neural network model on the temporal distributions of the semantic causal association topological feature matrix of the emotional log summary and the semantic embedding encoding vector of the emotional log summary. Specifically, by simulating the propagation process of the semantic features of the emotional log summary in the topological structure through the dynamic random walk mechanism and recursively aggregating the explicit semantics between nodes from local to global, the explicit semantic information in the emotional log summary can be comprehensively and systematically integrated, making up for the deficiencies of the traditional single-layer analysis method.

[0071] It should be understood that the latent semantics of the emotional log summary involves the potential embedding of features, and the information it contains is more complex and hidden. When performing semantic encoding, attention should be paid to the multi-hop propagation of high-order information and the distributed decoupling of deep features to avoid the over-smoothing phenomenon caused by the propagation of deep topological features. In this application, by performing the log semantic dynamic random walk encoding process based on the graph convolutional neural network model on the temporal distribution of the semantic causal association topological feature matrix of the emotional log summary and the semantic depth implicit embedding encoding vector of the emotional log summary, the complex semantic patterns hidden in the emotional log summary can be deeply mined, the profound temporal dependence relationships can be captured, and the generalization ability of the latent semantic feature expression of the emotional log summary can be enhanced, so that the emotional state of the elderly can be understood more comprehensively and deeply. That is, the generated latent context dynamic random walk semantic encoding vector of the emotional log summary can uncover the potential influence of the elderly's long-term psychological state on the current emotion, or the subtle emotional change connections between different rehabilitation stages, providing a deep perspective and basis for the model to deeply understand the emotional changes of the elderly.

[0072] Correspondingly, the surface context dynamic random walk semantic encoding vector of the emotional log summary captures the explicit semantic information of the emotional log summary, while the latent context dynamic random walk semantic encoding vector of the emotional log summary captures the implicit semantic information of the emotional log summary, and the two are highly complementary. By fusing the vectors containing semantic information at different levels, their respective advantages can be fully utilized to form a more comprehensive, accurate, and more discriminative and representative integrity emotion state temporal evolution semantic encoding vector. That is, the fused emotion state temporal evolution semantic encoding vector comprehensively integrates the explicit and implicit semantic information in the emotional log summary, as well as the causal relationship and multi-level context association information therein. It can keenly capture the potential trend of the elderly's emotional changes, thereby enhancing the model's ability to understand the emotional state of the elderly.

[0073] In an embodiment of the present application, the recognition result generation unit 123 is configured to obtain the emotional state recognition result based on the semantic encoding features of the temporal evolution of the emotional state. Specifically, in an embodiment of the present application, the recognition result generation unit is configured to: input the semantic encoding vector of the temporal evolution of the emotional state into an emotional state recognizer based on a classifier to obtain the emotional state recognition result, where the emotional state recognition result is used to indicate whether the emotional state of the target elderly object is evolving positively or negatively. It should be understood that by inputting the semantic encoding vector of the temporal evolution of the emotional state into an emotional state recognizer based on a classifier, the classification function of the classifier can be utilized to judge the processed semantic encoding vector, so as to accurately determine the evolution direction of the emotional state. Specifically, a classifier is a machine learning model, and its main function is to classify the input data features into different categories. It can learn the relationship between different emotional evolution patterns and their corresponding features through a large amount of labeled training data. After training is completed, the classifier can classify the newly input semantic encoding vector of the temporal evolution of the emotional state into the corresponding emotional state recognition result according to the learned feature mapping pattern. The generated recognition result can provide a key decision-making basis for psychological intervention. In a specific embodiment of the present application, inputting the semantic encoding vector of the temporal evolution of the emotional state into an emotional state recognizer based on a classifier to obtain the emotional state recognition result, where the emotional state recognition result is used to indicate whether the emotional state of the target elderly object is evolving positively or negatively, includes: performing fully connected encoding on the semantic encoding vector of the temporal evolution of the emotional state using the fully connected layer of the classifier to obtain a fully connected encoding feature vector of the semantic encoding of the temporal evolution of the emotional state; inputting the fully connected encoding feature vector of the semantic encoding of the temporal evolution of the emotional state into the Softmax classification function of the classifier to obtain the probability values of the semantic encoding vector of the temporal evolution of the emotional state belonging to each classification label, where the classification labels include those used to indicate that the emotional state of the target elderly object is evolving positively and those used to indicate that the emotional state of the target elderly object is evolving negatively; and determining the classification label corresponding to the maximum value among the probability values as the emotional state recognition result.

[0074] In the embodiments of the present application, the intervention signal generation module 130 is configured to determine whether to generate a psychological intervention prompt signal according to the emotion state recognition result. Specifically, in the embodiments of the present application, the intervention signal generation module is configured to: generate the psychological intervention prompt signal in response to the emotion state of the target elderly object evolving negatively. It should be understood that the negative evolution of the emotions of the elderly after surgery may have many adverse effects on their recovery. By promptly generating the psychological intervention prompt signal, it helps medical staff to learn in a timely manner that the elderly may be experiencing emotional distress, intervene in a timely manner, and provide personalized psychological support and intervention measures for the target elderly object to reduce the impact of negative emotions on the elderly. Specifically, the psychological intervention prompt signal may include basic information of the elderly object (such as name, ward number), specific descriptions of the changes in the emotion state (such as "showing high anxiety for three consecutive days"), and other information, and is accurately transmitted to relevant medical staff through the hospital's internal communication system, the work mobile phone application of medical staff, etc. To ensure that the psychological intervention prompt signal is indeed received and actions are taken, the system includes a feedback mechanism that requires medical staff to confirm that they have read the prompt after receiving it and to fill in the next plan or the measures already taken. In this way, it can be ensured that when it is found that the elderly object has a negative emotion tendency, necessary psychological support can be provided quickly and effectively.

[0075] In summary, the artificial intelligence-based postoperative rehabilitation nursing system 100 for the elderly according to the embodiments of the present application is elucidated. First, it encrypts and regularizes the collected emotional log data set of the target elderly object to obtain the time series distribution of the encrypted emotional log. Then, it uses an artificial intelligence-based data analysis method to summarize the content and embed the semantics of the time series distribution of the encrypted emotional log to obtain the time series distribution of the summary semantic features of the emotional log. And it obtains the emotion recognition result based on the context causal association coding feature of the time series distribution of the summary semantic features of the emotional log. Finally, it determines whether to generate a psychological intervention prompt signal based on the emotion recognition result. In this way, by continuously monitoring and analyzing the emotional fluctuations and their patterns of the elderly, it helps to capture the emotion state more accurately, and further improves the timeliness, effectiveness, and intelligence of psychological intervention.

Claims

1. An artificial intelligence-based postoperative rehabilitation nursing system for the elderly, characterized in that, Including: A data acquisition and preprocessing module, which is used to collect the emotional logs of the target elderly object to obtain a dataset of emotional logs, and perform encryption processing and data regularization on the dataset of the emotional logs to obtain the time series distribution of the encrypted emotional logs; An emotional state recognition module, which is used to perform text-semantic-based emotional analysis on the time series distribution of the encrypted emotional logs to obtain an emotional state recognition result. Among them, the emotional state recognition module includes: a log time series summary and embedding unit, which is used to perform content summary and semantic embedding on the time series distribution of the encrypted emotional logs to obtain the time series distribution of the semantic embedding encoding features of the emotional log summary; a log time series encoding unit, which is used to perform emotional log context causal association encoding on the time series distribution of the semantic embedding encoding features of the emotional log summary to obtain the time series evolution semantic encoding features of the emotional state; a recognition result generation unit, which is used to obtain the emotional state recognition result based on the time series evolution semantic encoding features of the emotional state; An intervention signal generation module, which is used to judge whether to generate a psychological intervention prompt signal according to the emotional state recognition result.

2. The elderly postoperative rehabilitation nursing system based on artificial intelligence according to claim 1, wherein, The data acquisition and preprocessing module includes: An emotional log acquisition unit, which is used to collect the emotional logs of the target elderly object to obtain the dataset of the emotional logs; An emotional log encryption processing unit, which is used to perform encryption processing on each emotional log in the dataset of the emotional logs to obtain a dataset of encrypted emotional logs; An emotional log time series distribution adjustment unit, which is used to perform data distribution pattern adjustment based on the time dimension on the dataset of the encrypted emotional logs according to the time stamp to obtain the time series distribution of the encrypted emotional logs.

3. The elderly postoperative rehabilitation nursing system based on artificial intelligence according to claim 2, characterized in that, The log time series summary and embedding unit includes: An emotional log summary generation subunit, which is used to input each encrypted emotional log in the time series distribution of the encrypted emotional logs into a content summary encoder based on a large language model to obtain the time series distribution of the emotional log summary; An emotional log summary embedding encoding subunit, which is used to perform semantic embedding encoding on each emotional log summary in the time series distribution of the emotional log summary to obtain the time series distribution of the semantic embedding encoding vectors of the emotional log summary as the time series distribution of the semantic embedding encoding features of the emotional log summary.

4. The elderly postoperative rehabilitation nursing system based on artificial intelligence according to claim 3, characterized in that, The emotional log summary embedding encoding subunit is used to: use a semantic encoder based on a bidirectional recurrent neural network model to perform semantic embedding encoding on each emotional log summary in the time series distribution of the emotional log summary to obtain the time series distribution of the semantic embedding encoding vectors of the emotional log summary.

5. The artificial intelligence-based postoperative rehabilitation nursing system for the elderly according to claim 1, wherein The log time series encoding unit includes: An emotional log summary implicit semantic feature mining subunit, which is used to perform emotional log summary implicit semantic feature mining on the time series distribution of the semantic embedding encoding vectors of the emotional log summary to obtain the time series distribution of the semantic deep implicit embedding encoding vectors of the emotional log summary; An emotional log summary semantic causal analysis subunit, which is used to calculate the emotional log summary semantic causal association factors and perform emotional log summary semantic causal triggering on the temporal distribution of the emotional log summary semantic depth implicit embedding coding vectors to obtain an emotional log summary semantic causal association topological feature matrix; An emotional state temporal evolution semantic coding feature generation subunit, which is used to perform log semantic context dynamic walk coding fusion on the temporal distribution of the emotional log summary semantic depth implicit embedding coding vectors and the temporal distribution of the emotional log summary semantic embedding coding vectors based on the emotional log summary semantic causal association topological feature matrix to obtain an emotional state temporal evolution semantic coding vector as the emotional state temporal evolution semantic coding feature.

6. The elderly postoperative rehabilitation nursing system based on artificial intelligence according to claim 5, characterized in that, The emotional log summary semantic causal analysis subunit includes: An emotional log summary semantic causal association factor calculation primary subunit, which is used to calculate the emotional log summary semantic causal association factors between any two emotional log summary semantic depth implicit embedding coding vectors in the temporal distribution of the emotional log summary semantic depth implicit embedding coding vectors to obtain an emotional log summary semantic causal association topological matrix composed of multiple emotional log summary semantic causal association factors; An emotional log summary semantic causal triggering primary subunit, which is used to perform emotional log summary semantic causal triggering based on a gating function on the emotional log summary semantic causal association topological matrix to obtain the emotional log summary semantic causal association topological feature matrix.

7. The elderly postoperative rehabilitation nursing system based on artificial intelligence according to claim 6, characterized in that, The emotional log summary semantic causal association factor calculation primary subunit is used to: Calculate the association matrix between any two emotional log summary semantic depth implicit embedding coding vectors in the temporal distribution of the emotional log summary semantic depth implicit embedding coding vectors to obtain the temporal distribution of the emotional log summary semantic depth association embedding coding matrix; Calculate the emotional log summary semantic causal association factors of each emotional log summary semantic depth association embedding coding matrix in the temporal distribution of the emotional log summary semantic depth association embedding coding matrix to obtain the emotional log summary semantic causal association topological matrix composed of multiple emotional log summary semantic causal association factors, where the emotional log summary semantic causal association factor is calculated from the maximum value, mean value, variance, and causal association bias term of the emotional log summary semantic depth association embedding coding matrix; In response to the variance of the emotional log summary semantic depth association embedding coding matrix being greater than or equal to a preset threshold, use the weighted average of the distances between any two emotional log summary semantic depth implicit embedding coding vectors in the temporal distribution of the emotional log summary semantic depth implicit embedding coding vectors as the causal association bias term, In response to the variance of the emotional log summary semantic depth association embedding coding matrix being less than the preset threshold, use the weighted average of the emotional log summary semantic depth association embedding coding matrix as the causal association bias term.

8. The artificial intelligence-based postoperative rehabilitation nursing system for the elderly according to claim 7, characterized in that The emotional state temporal evolution semantic coding feature generation subunit is used to: Perform log semantic dynamic walk encoding based on a graph convolutional neural network model on the temporal distribution of the semantic causal association topological feature matrix of the emotional log summary and the semantic embedded encoding vector of the emotional log summary to obtain the surface context dynamic walk semantic encoding vector of the emotional log summary; Perform the log semantic dynamic walk encoding based on the graph convolutional neural network model on the temporal distribution of the semantic causal association topological feature matrix of the emotional log summary and the semantic deep implicit embedded encoding vector of the emotional log summary to obtain the hidden layer context dynamic walk semantic encoding vector of the emotional log summary; Fuse the hidden layer context dynamic walk semantic encoding vector of the emotional log summary and the surface context dynamic walk semantic encoding vector of the emotional log summary to obtain the semantic encoding vector of the temporal evolution of the emotional state.

9. The artificial intelligence-based postoperative rehabilitation nursing system for the elderly according to claim 8, characterized in that, The recognition result generation unit is configured to: input the semantic encoding vector of the temporal evolution of the emotional state into an emotional state recognizer based on a classifier to obtain the emotional state recognition result, and the emotional state recognition result is used to indicate whether the emotional state of the target elderly object evolves positively or negatively.

10. The artificial intelligence-based postoperative rehabilitation nursing system for the elderly according to claim 9, characterized in that, The intervention signal generation module is configured to: generate the psychological intervention prompt signal in response to the emotional state of the target elderly object evolving negatively.

Images