A dynamic recommendation method and system for rehabilitation programs in clinical nursing

Abstract

The present application provides a method and system for dynamically recommending rehabilitation plans in clinical nursing. The method includes: collecting a nursing data sequence and a rehabilitation plan sequence of a target user within a historical time window, wherein the rehabilitation plan includes rehabilitation exercise data; using the nursing data sequence and the rehabilitation plan sequence to predict the blood sugar of the target user and obtain predicted blood sugar; collecting the actual blood sugar of the target user, combining the predicted blood sugar to calculate the blood sugar deviation, and compensating and correcting the blood sugar deviation based on the blood sugar prediction error and the blood sugar fluctuation parameter of the target user to obtain a corrected blood sugar deviation; based on the corrected blood sugar deviation, classifying and obtaining nursing abnormality dimensions, configuring nursing abnormality prediction resources, performing nursing abnormality pattern analysis, obtaining nursing abnormality patterns, and dynamically adjusting and recommending rehabilitation plans. This solves the technical problem of poor adaptability of rehabilitation plans in clinical nursing in the prior art.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and in particular to a method and system for dynamically recommending rehabilitation programs in clinical nursing. Background Art

[0002] In the clinical care of diabetes, recommending appropriate rehabilitation programs to patients can effectively alleviate dysglycemia. However, traditional rehabilitation program recommendation methods rely on blood glucose monitoring values ​​and empirical rules for extensive decision-making and recommendation. They do not deeply analyze the abnormal nursing patterns that lead to dysglycemia and adjust the rehabilitation program accordingly. For example, when a patient experiences a blood glucose peak (12.8mmol / L), traditional methods only mechanically increase the duration of exercise, but fail to identify the true cause: the patient's deteriorating sleep quality. Therefore, traditional extensive recommendation methods lead to poor adaptability of rehabilitation programs. Summary of the Invention

[0003] The present invention aims to solve the technical problem of poor adaptability of rehabilitation programs in clinical nursing in the prior art and provides a method and system for dynamically recommending rehabilitation programs in clinical nursing.

[0004] The technical solution of the present invention to solve the above technical problems is as follows:

[0005] In a first aspect, the present invention provides a method for dynamically recommending rehabilitation plans in clinical nursing, comprising:

[0006] Collecting nursing data sequences and rehabilitation program sequences of the target user within a historical time window, wherein the rehabilitation program includes rehabilitation exercise data;

[0007] Using the nursing data sequence and rehabilitation program sequence, predicting the blood glucose of the target user to obtain predicted blood glucose;

[0008] Collecting the actual blood glucose of the target user, combining it with the predicted blood glucose, calculating a blood glucose deviation, and compensating and correcting the blood glucose deviation based on the blood glucose prediction error and the blood glucose fluctuation parameter of the target user to obtain a corrected blood glucose deviation;

[0009] According to the corrected blood sugar deviation, the nursing abnormality dimension is obtained by classification, nursing abnormality prediction resources are configured, nursing abnormality pattern analysis is performed, nursing abnormality pattern is obtained, and dynamic adjustment recommendation of rehabilitation plan is performed.

[0010] In a second aspect, the present invention provides a dynamic recommendation system for rehabilitation programs in clinical nursing, comprising:

[0011] A data collection module is used to collect nursing data sequences and rehabilitation program sequences of target users within a historical time window, wherein the rehabilitation program includes rehabilitation exercise data;

[0012] A blood glucose prediction module, configured to use the nursing data sequence and the rehabilitation program sequence to predict the blood glucose of the target user and obtain predicted blood glucose;

[0013] a compensation and correction module, configured to collect the actual blood glucose of the target user, combine it with the predicted blood glucose, calculate a blood glucose deviation, and perform compensation and correction on the blood glucose deviation based on the blood glucose prediction error and the blood glucose fluctuation parameter of the target user to obtain a corrected blood glucose deviation;

[0014] The adjustment recommendation module is used to classify and obtain nursing abnormality dimensions based on the corrected blood sugar deviation, configure nursing abnormality prediction resources, perform nursing abnormality pattern analysis, obtain nursing abnormality patterns, and make dynamic adjustment recommendations for rehabilitation plans.

[0015] The beneficial effects of the present invention are:

[0016] This application first collects the nursing data and rehabilitation plans of the target user in multiple time periods within the historical time window, and establishes a nursing data sequence and a rehabilitation plan sequence with time series characteristics based on the time period, so as to provide necessary data support for the subsequent dynamic recommendation of rehabilitation plans. Then, the nursing data sequence and the rehabilitation plan sequence are used to predict the blood sugar of the target user and obtain the predicted blood sugar, that is, the ideal blood sugar value of the target user under the nursing data sequence and the rehabilitation plan sequence. Then, the actual blood sugar of the target user is collected, combined with the predicted blood sugar, and the blood sugar deviation is calculated. According to the blood sugar prediction error and the blood sugar fluctuation parameter of the target user, the blood sugar deviation is compensated and corrected to obtain the corrected blood sugar deviation, eliminating the interference between the prediction error and the individual physiological fluctuation, and improving the accuracy of the abnormal blood sugar analysis. Finally, according to the corrected blood sugar deviation, the nursing abnormality dimension is classified, the nursing abnormality prediction resources are configured, the nursing abnormality pattern analysis is performed, the nursing abnormality pattern is obtained, and the dynamic adjustment recommendation of the rehabilitation plan is performed.

[0017] Through the above technical solution, this application collects the nursing data sequence and rehabilitation program sequence of the target user within the historical time window; then uses the nursing data sequence and rehabilitation program sequence to predict the blood sugar of the target user and obtain the predicted blood sugar; then collects the actual blood sugar of the target user, combines the predicted blood sugar, calculates the blood sugar deviation, and compensates and corrects the blood sugar deviation based on the blood sugar prediction error and the blood sugar fluctuation parameter of the target user to obtain the corrected blood sugar deviation; finally, based on the corrected blood sugar deviation, classifies and obtains the nursing abnormality dimension, configures the nursing abnormality prediction resources, performs nursing abnormality pattern analysis, obtains the nursing abnormality pattern, and performs dynamic adjustment and recommendation of the rehabilitation program. In this way, the nursing abnormality pattern that causes abnormal blood sugar is obtained, and the rehabilitation program is adjusted in time accordingly, thereby improving the adaptability of the rehabilitation program. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1A schematic diagram of a flow chart of a method for dynamically recommending rehabilitation programs in clinical nursing provided by the present invention;

[0019] Figure 2 This is a structural diagram of a dynamic recommendation system for rehabilitation plans in clinical nursing provided by the present invention.

[0020] In the accompanying drawings, the components represented by the reference numerals are as follows:

[0021] Data collection module 11, blood glucose prediction module 12, compensation correction module 13, adjustment recommendation module 14. DETAILED DESCRIPTION

[0022] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts shall fall within the scope of protection of the present invention.

[0023] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature specified as "first" or "second" may explicitly or implicitly include one or more of the specified features. In the description of the present invention, "plurality" means two or more, unless otherwise specifically defined.

[0024] In the description of the present invention, the term "for example" is used to mean "used as an example, illustration or illustration". Any embodiment of the present invention described as "for example" is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any person skilled in the art to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes are not elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed herein.

[0025] Example 1, as Figure 1 As shown, an embodiment of the present invention provides a method for dynamically recommending rehabilitation plans in clinical nursing, including:

[0026] S10: Collecting nursing data sequences and rehabilitation program sequences of the target user within a historical time window, wherein the rehabilitation program includes rehabilitation exercise data;

[0027] Blood sugar fluctuations are closely linked to exercise patterns. Measured data show that, at the same exercise duration and intensity, brisk walking is more effective than cycling in lowering blood sugar levels (reducing blood sugar by an average of 0.8 mmol / L), and exercising two hours after a meal is more effective than exercising on an empty stomach. Therefore, in the care of diabetic patients, recommending appropriate rehabilitation programs based on their individual characteristics can help alleviate abnormal blood sugar levels.

[0028] To address the above issues, this application collects the nursing data and rehabilitation plans of the target users in multiple time periods within the historical time window, and establishes nursing data sequences and rehabilitation plan sequences with time series characteristics based on the time periods, so as to analyze the impact of different rehabilitation plans on blood sugar.

[0029] Specifically, step S10 in the method includes:

[0030] Collecting nursing data and rehabilitation plans of the target user over multiple time periods within a historical time window, wherein the rehabilitation plan includes rehabilitation exercise data;

[0031] Integrate nursing data and rehabilitation plans in multiple time periods to obtain nursing data sequences and rehabilitation plan sequences.

[0032] In an embodiment of the present application, the nursing data and rehabilitation plan of the target user in multiple time periods (smaller time units, such as 1 day, 3 hours, etc.) within a historical time window (such as the past 7 days, 10 days, etc.) are first collected, wherein the nursing data includes treatment data and eating data, such as food intake, food category, etc., and the rehabilitation plan includes rehabilitation exercise data, such as exercise type, duration, intensity, etc. For example, the nursing data of a patient is collected every day (multiple time periods) within the past 14 days (historical time window), including data such as morning fasting blood sugar (such as 5.6mmol / L), postprandial blood sugar (such as 8.2mmol / L), insulin dosage (8U in the morning / 6U in the evening), and a daily rehabilitation plan, such as a 30-minute brisk walk at 4 pm every day, with a heart rate range of 100-110 beats / minute.

[0033] Secondly, the nursing data and rehabilitation plans from multiple time periods are integrated to obtain nursing data sequences and rehabilitation plan sequences. Specifically, the collected nursing data and rehabilitation plans are integrated by time period (e.g., day) to form nursing data sequences and rehabilitation plan sequences with time series characteristics. For example, the nursing data (including treatment data and dietary data) and rehabilitation plans (including rehabilitation exercise data) from the diabetic patient for each day (multiple time periods) over the past 14 days (historical time window) are integrated to obtain nursing data sequences and rehabilitation plan sequences.

[0034] In summary, this application collects nursing data and rehabilitation plans for target users over multiple time periods within a historical time window, and establishes nursing data sequences and rehabilitation plan sequences with time series characteristics based on the time periods. In this way, the time-series data sequence fully preserves the dynamic relationship between blood sugar fluctuations and rehabilitation plans, providing the necessary data support for the dynamic recommendation of subsequent rehabilitation plans.

[0035] S20: using the nursing data sequence and the rehabilitation program sequence to predict the blood glucose of the target user and obtain predicted blood glucose;

[0036] During the care process for diabetic patients, abnormalities in care (such as a patient's mood, poor sleep quality, or failure to adhere to strict dietary restrictions) can lead to blood sugar deviations. For example, a patient's rehabilitation plan may be reasonable, but excessive psychological stress can lead to persistent insomnia, which in turn causes blood sugar deviations. Therefore, when analyzing blood sugar deviations for target users and dynamically adjusting rehabilitation plans, it is necessary to obtain the target user's ideal blood sugar levels under the nursing data and rehabilitation plan sequences, and use this information to analyze the blood sugar deviations caused by actual care deviations.

[0037] To address the above problems, this application obtains multiple approximate users based on the characteristic information of the target user, and then extracts the blood sugar of the approximate users after the nursing data sequence and rehabilitation program sequence, marks them as sample predicted blood sugar sets, and then uses this as a training set to train and optimize the blood sugar predictor constructed by machine learning to obtain a blood sugar predictor. Finally, the nursing data sequence and rehabilitation program sequence are input to predict and output the predicted blood sugar of the target user.

[0038] Specifically, step S20 in the method includes:

[0039] Calling a blood glucose predictor that has been trained in advance, wherein the blood glucose predictor is trained using sample nursing data of an approximate user;

[0040] The nursing data sequence and the rehabilitation program sequence are input into the blood glucose predictor, and the predicted blood glucose of the target user is obtained by the prediction output.

[0041] In this embodiment, a previously trained blood glucose predictor is first called. The blood glucose predictor is built based on machine learning and trained using sample nursing data from a similar user. It can predict the target user's ideal blood glucose level under a nursing data sequence and a rehabilitation program sequence.

[0042] The nursing data sequence and rehabilitation regimen sequence are then input into a blood glucose predictor, which produces a predicted blood glucose value for the target user. This predicted blood glucose value represents the target user's ideal blood glucose value under these nursing data and rehabilitation regimen sequences. For example, if the target user's nursing data sequence (lunch: 100g carbohydrates, 30g protein, 200g vegetables) and rehabilitation regimen sequence (30 minutes of brisk walking two hours after lunch, heart rate range: 100-110 bpm) are input into the blood glucose predictor, the predicted blood glucose value for the target user after a 30-minute brisk walk two hours after lunch is 7.1 mmol / L.

[0043] Furthermore, the training step of the blood glucose predictor includes:

[0044] Obtaining user characteristic information of the target user, and screening to obtain a plurality of similar users based on the user characteristic information;

[0045] According to the nursing data of the plurality of approximate users in the historical time, a sample nursing data sequence set and a sample rehabilitation program sequence set are collected, and the blood glucose of the approximate user after collecting different sample nursing data sequences and sample rehabilitation program sequences is approximated and marked as a sample predicted blood glucose set;

[0046] Use machine learning to build a blood sugar predictor;

[0047] The blood glucose predictor is supervised trained and tested using the sample nursing data sequence set, the sample rehabilitation program sequence set and the sample predicted blood glucose set. When the test accuracy is greater than or equal to the accuracy threshold, the training is completed and the blood glucose prediction accuracy is obtained.

[0048] In an embodiment of the present application, user characteristic information of the target user is first obtained, and then a plurality of approximate users are obtained by screening based on the user characteristic information. The characteristic information includes multidimensional feature vectors such as the target user's age, gender, BMI, initial blood glucose value, blood pressure, and complications. By calculating the cosine similarity of the multidimensional feature vectors, a group of approximate users with a similarity greater than 0.85 is screened out. For example, the user characteristic information of the target user is first obtained, including age, gender, BMI, initial blood glucose value, blood pressure, complications, etc., and then a plurality of approximate users with a similarity greater than 0.85 are screened based on the characteristic information.

[0049] Next, based on the historical nursing data of multiple similar users, a set of sample nursing data sequences and a set of sample rehabilitation program sequences are collected. The blood glucose levels of the similar users after collecting different sample nursing data sequences and sample rehabilitation program sequences are labeled as a sample predicted blood glucose set. The sample predicted blood glucose set can predict the ideal blood glucose level of the target user under the nursing data sequence and rehabilitation program sequence. For example, based on historical nursing data, the blood glucose levels of multiple similar users after collecting nursing data sequences and rehabilitation program sequences are labeled as a sample predicted blood glucose set.

[0050] Secondly, machine learning is used to construct a blood glucose predictor. Specifically, the machine learning architecture of the blood glucose predictor adopts a multimodal spatiotemporal fusion design, with its core consisting of a spatiotemporal feature extraction layer, a personalized adaptation module, and a physiological constraint unit. The spatiotemporal feature extraction layer captures the local fluctuation patterns of the blood glucose sequence through dilated causal convolution, combines it with a bidirectional GRU network to model long-range dependencies, and uses a self-attention mechanism to dynamically focus on key time nodes (such as the 90-minute metabolic window after exercise). The personalized adaptation module encodes user feature information (age, BMI, blood type, etc.) into a 64-dimensional embedding vector, which is cross-modally fused with dynamic temporal features through a gating mechanism. The physiological constraint unit uses the ReLU activation function to hard-limit the hourly blood glucose change rate to ≤3mmol / L, and uses a risk-sensitive loss function to apply a 5x error weight to high and low blood glucose intervals.

[0051] Finally, supervised training and testing of the blood glucose predictor are performed using a set of sample nursing data sequences, a set of sample rehabilitation plan sequences, and a set of sample predicted blood glucose. Training is completed and the blood glucose prediction accuracy is obtained when the test accuracy is greater than or equal to an accuracy threshold (e.g., set to 90%). The training process can be implemented using the following technical approaches: 1. Preparation of the training dataset: The sample nursing data sequences, the sample rehabilitation plan sequences, and the sample predicted blood glucose sequences are divided into a training set, a validation set, and a test set in a ratio of 7.5:1.5:1.5. 2. Model training: A curriculum learning strategy is used for training and optimization. The model is first trained on the training set, then fine-tuned by augmenting difficult samples using a generative adversarial network (GAN). Bayesian hyperparameter search is introduced during the model optimization phase. Ultimately, achieving an accuracy of over 90% on the test set is considered convergence. For example, supervised training and testing of the model using a set of sample nursing data sequences, a set of sample rehabilitation plan sequences, and a set of sample predicted blood glucose sequences yields a blood glucose predictor with a blood glucose prediction accuracy of 92%.

[0052] In summary, compared to the existing technology, this application obtains multiple approximate users based on the characteristic information of the target user, and then extracts the blood sugar of the approximate user after the nursing data sequence and rehabilitation program sequence, annotates it as a sample predicted blood sugar set, and then uses this as a training set to train and optimize the blood sugar predictor constructed by machine learning to obtain a blood sugar predictor. Finally, the nursing data sequence and rehabilitation program sequence are input to predict and output the predicted blood sugar of the target user. In this way, the ideal blood sugar value of the target user under the nursing data sequence and rehabilitation program sequence is obtained, providing the necessary data support for subsequent abnormal analysis.

[0053] S30: collecting the actual blood glucose of the target user, combining it with the predicted blood glucose, calculating a blood glucose deviation, and compensating and correcting the blood glucose deviation based on the blood glucose prediction error and the blood glucose fluctuation parameter of the target user to obtain a corrected blood glucose deviation;

[0054] When calculating the blood glucose deviation based on the predicted blood glucose output by the aforementioned blood glucose predictor and the actual blood glucose obtained during the test, and then performing blood glucose deviation analysis, both the prediction error of the blood glucose predictor and normal blood glucose fluctuations may cause blood glucose deviation. For example, the blood glucose predictor may fail to predict, thereby causing blood glucose deviation, the target user's normal metabolism may cause blood glucose fluctuations, and normal blood glucose fluctuations may cause blood glucose deviation. Therefore, in order to improve the accuracy of the analysis, it is necessary to compensate for the blood glucose deviation.

[0055] In response to the above problems, the present application performs double compensation correction on the calculated blood glucose deviation to obtain the corrected blood glucose deviation: first, compensation correction calculation is performed based on the blood glucose prediction error to obtain the predicted corrected blood glucose deviation, and then compensation correction calculation is performed on the predicted corrected blood glucose deviation based on the blood glucose fluctuation parameters of the target user to obtain the corrected blood glucose deviation.

[0056] Specifically, step S30 in the method includes:

[0057] Collecting the actual blood glucose of the target user, calculating the difference between the actual blood glucose and the predicted blood glucose, and obtaining a blood glucose deviation;

[0058] Obtain the blood sugar prediction accuracy of the blood sugar prediction and calculate the blood sugar prediction error rate;

[0059] performing compensation correction calculation on the blood glucose deviation according to the blood glucose prediction error rate to obtain a predicted and corrected blood glucose deviation;

[0060] Obtaining a blood glucose fluctuation value of the target user under normal care status as a blood glucose fluctuation parameter;

[0061] The predicted and corrected blood glucose deviation is compensated and corrected based on the blood glucose fluctuation parameter to obtain the corrected blood glucose deviation.

[0062] In the embodiment of the present application, the actual blood glucose of the target user is first collected, and the difference between the actual blood glucose and the predicted blood glucose is calculated to obtain the blood glucose deviation, wherein the blood glucose deviation = actual blood glucose - predicted blood glucose. For example, if the actual blood glucose collected from a patient is 7.3 mmol / L and the predicted blood glucose value output by the blood glucose predictor is 6.9 mmol / L, then the blood glucose deviation = 7.3 mmol / L - 6.9 mmol / L = 0.4 mmol / L. Generally, the subsequent steps are performed only when the actual blood glucose is greater than the predicted blood glucose, that is, when the calculated blood glucose deviation is a positive value.

[0063] Next, the blood glucose prediction accuracy of the blood glucose predictor is obtained, and the blood glucose prediction error rate is calculated, where the blood glucose prediction error rate = 1 - blood glucose prediction accuracy rate. For example, if the blood glucose prediction accuracy of the blood glucose predictor is 96%, the blood glucose prediction error rate = 1 - 96% = 4%.

[0064] Next, based on the blood glucose prediction error rate, a compensation correction calculation is performed on the blood glucose deviation to obtain the predicted corrected blood glucose deviation, where the predicted corrected blood glucose deviation = blood glucose deviation * (1-blood glucose prediction error rate). For example, if a patient's blood glucose deviation is 0.4mmol / L and the blood glucose prediction error rate is 4%, then the predicted corrected blood glucose deviation = 0.4mmol / L * (1-4%) = 0.384mmol / L. Compensating and correcting the blood glucose deviation based on the blood glucose prediction error rate can avoid misjudgments caused by the inherent error of the blood glucose predictor (blood glucose prediction error rate), effectively improving analysis accuracy.

[0065] Furthermore, the target user's blood glucose fluctuation value under normal care is obtained as a blood glucose fluctuation parameter. The blood glucose fluctuation value = (|target user's maximum blood glucose value under normal care - average blood glucose value|) / average blood glucose value. For example, if the target user's average blood glucose value under normal care is 6.9 mmol / L and the maximum blood glucose value is 7.9 mmol / L, then the blood glucose fluctuation value = (|7.9 mmol / L - 6.9 mmol / L|) / 6.9 mmol / L = 14.5%.

[0066] Finally, the predicted-corrected blood glucose deviation is compensated and corrected based on the blood glucose fluctuation parameter to obtain the corrected blood glucose deviation, where the corrected blood glucose deviation = predicted-corrected blood glucose deviation * (1 - blood glucose fluctuation value). For example, if the predicted-corrected blood glucose deviation is 0.384 mmol / L and the blood glucose fluctuation value is 14.5%, then the corrected blood glucose deviation = 0.384 mmol / L * (1 - 14.5%) = 0.328 mmol / L.

[0067] In summary, compared to the prior art, this application first performs a compensation correction calculation based on the blood glucose prediction error to obtain the predicted and corrected blood glucose deviation. Then, it performs a compensation correction calculation on the predicted and corrected blood glucose deviation based on the target user's blood glucose fluctuation parameters to obtain the corrected blood glucose deviation. In this way, compensation correction is performed on the blood glucose prediction error and blood glucose fluctuation parameters, eliminating the interference between the prediction error and individual physiological fluctuations and improving the accuracy of abnormality analysis.

[0068] S40: Based on the corrected blood sugar deviation, classify and obtain nursing abnormality dimensions, configure nursing abnormality prediction resources, perform nursing abnormality pattern analysis, obtain nursing abnormality patterns, and make dynamic adjustment recommendations for rehabilitation programs.

[0069] When analyzing abnormal nursing patterns based on corrected blood sugar deviations, the larger the blood sugar deviation, the more complex the possible abnormal nursing patterns. More abnormal nursing prediction resources need to be configured for abnormal analysis to obtain accurate abnormal nursing patterns and make dynamic adjustments to rehabilitation plans accordingly.

[0070] In response to the above problems, this application inputs the corrected blood sugar deviation into the nursing abnormality dimension classification table, classifies the nursing abnormality dimension, and then configures the nursing abnormality prediction resources based on the nursing abnormality dimension. The corrected blood sugar deviation is then input into the nursing abnormality analyzer, and the nursing abnormality pattern is output. Finally, based on the nursing abnormality pattern, dynamic adjustment recommendations for the rehabilitation plan are made.

[0071] Specifically, step S40 in the method includes:

[0072] Based on the nursing data in the historical time, the sample blood glucose deviation set is collected, and all the nursing abnormality patterns that appear under different sample blood glucose deviations are collected, and the number of nursing abnormality patterns is extracted to obtain the sample nursing abnormality dimension set;

[0073] Constructing a mapping relationship between the sample blood glucose deviation set and the sample nursing abnormality dimension set to obtain a nursing abnormality dimension classification table;

[0074] Inputting the corrected blood sugar deviation into the nursing abnormality dimension classification table, and classifying to obtain the nursing abnormality dimension;

[0075] According to the nursing abnormality dimension, a nursing abnormality prediction resource is configured, and a nursing abnormality pattern analysis is performed on the corrected blood glucose deviation to obtain a nursing abnormality pattern;

[0076] Based on the abnormal nursing pattern, dynamic adjustment recommendations for the rehabilitation plan are made.

[0077] In this embodiment, a sample blood glucose deviation set is first collected based on historical nursing data. All abnormal nursing patterns (such as bad mood, excessive stress, poor sleep quality, overeating, etc.) that occur under different sample blood glucose deviations are collected. The number of abnormal nursing patterns is then extracted to obtain a sample abnormal nursing dimension set. For example, when the blood glucose deviation is 0.4 mmol / L, there are three abnormal nursing patterns (bad mood, excessive stress, and poor sleep quality). Therefore, the abnormal nursing dimension obtained when the blood glucose deviation is 0.4 mmol / L is 3.

[0078] Next, a mapping relationship is constructed between the set of sample blood glucose deviations and the set of sample nursing anomaly dimensions to obtain a nursing anomaly dimension classification table. In this nursing anomaly dimension classification table, there is a one-to-one mapping relationship between blood glucose deviations and nursing anomaly dimensions. By inputting a blood glucose deviation, a unique matching nursing anomaly dimension can be retrieved and output.

[0079] Next, the corrected blood glucose deviation is input into the nursing abnormality dimension classification table, and the nursing abnormality dimension is obtained by classification. For example, the corrected blood glucose deviation of 0.328 mmol / L is input into the nursing abnormality dimension classification table, and the nursing abnormality dimension 3 is obtained as an output.

[0080] Furthermore, based on the nursing anomaly dimension (e.g., 2 or 3), nursing anomaly prediction resources are allocated to analyze the corrected blood glucose deviation for nursing anomaly patterns, obtaining nursing anomaly patterns (e.g., bad mood, excessive stress, poor sleep quality, overeating, etc.). This analysis is performed by a pre-trained nursing anomaly analyzer comprising multiple nursing anomaly analysis branches. The corrected blood glucose deviation is input into each of these branches, and the output is filtered to obtain the nursing anomaly pattern with the highest occurrence rate, which is then used as the final nursing anomaly pattern. Furthermore, the larger the nursing anomaly dimension, the more complex the potential nursing anomaly, and the more nursing anomaly prediction resources should be allocated to analyze the nursing anomaly pattern. For example, when the nursing anomaly dimension is 1, the nursing anomaly is relatively uncomplex, requiring only two nursing anomaly analysis branches for nursing anomaly pattern analysis. When the nursing anomaly dimension is 4, the nursing anomaly is relatively complex, requiring six nursing anomaly analysis branches for nursing anomaly pattern analysis.

[0081] Finally, based on the abnormal nursing patterns, a dynamic adjustment recommendation for the rehabilitation plan is made. For example, based on the abnormal nursing pattern (excessive stress), the original rehabilitation plan (30 minutes of brisk walking at 4 pm every day) is adjusted to (30 minutes of slow walking at 4 pm every day and listening to 30 minutes of soothing piano music).

[0082] The step of “configuring nursing abnormality prediction resources according to the nursing abnormality dimension, performing nursing abnormality pattern analysis on the corrected blood glucose deviation, and obtaining a nursing abnormality pattern” includes:

[0083] Obtain the total number of nursing abnormality patterns as the maximum nursing abnormality dimension;

[0084] Calculating the ratio of the nursing abnormality dimension to the maximum nursing abnormality dimension to obtain a nursing abnormality prediction resource coefficient;

[0085] Calling a pre-trained nursing anomaly analyzer including multiple nursing anomaly analysis branches, and calculating the number of nursing branches according to the nursing anomaly prediction resource coefficient and the number of the multiple nursing anomaly analysis branches;

[0086] Randomly call the nursing abnormality analysis branch of the nursing branch number, input the corrected blood sugar deviation, output the nursing abnormality pattern of the nursing branch number, and screen the nursing abnormality pattern with the largest occurrence ratio.

[0087] In the embodiment of the present application, the number of all abnormal nursing patterns is first obtained as the maximum abnormal nursing dimension. For example, if the number of all abnormal nursing patterns is 10 (including bad mood, excessive stress, poor sleep quality, overeating, etc.), 10 is used as the maximum abnormal nursing dimension.

[0088] Next, calculate the ratio of the nursing anomaly dimension to the maximum nursing anomaly dimension to obtain the nursing anomaly prediction resource coefficient, where the nursing anomaly prediction resource coefficient = nursing anomaly dimension / maximum nursing anomaly dimension. For example, if the corrected blood glucose deviation is input into the nursing anomaly dimension classification table, the nursing anomaly dimension obtained by classification is 3, and the maximum nursing anomaly dimension is 10, then the nursing anomaly prediction resource coefficient = 3 / 10 = 0.3.

[0089] Next, call the nursing anomaly analyzer that has been trained in advance and includes multiple nursing anomaly analysis branches, and calculate the number of nursing branches based on the nursing anomaly prediction resource coefficient and the number of multiple nursing anomaly analysis branches. The nursing anomaly prediction resource coefficient is multiplied by the total number of multiple nursing anomaly analysis branches and rounded to the nearest integer to obtain the number of nursing branches. For example, the nursing anomaly prediction resource coefficient is 0.3, and the total number of nursing anomaly analysis branches built into the nursing anomaly analyzer is 15. Then, according to 0.3*15=4.5, rounded to 5, the number of nursing branches obtained is 5.

[0090] Finally, the nursing anomaly analysis branch is randomly called, and the corrected blood glucose deviation is input into each branch. The resulting nursing anomaly patterns are then output, and the nursing anomaly pattern with the highest occurrence rate is selected as the final nursing anomaly pattern. For example, if the calculated number of nursing branches is 5 and the total number of nursing anomaly analysis branches built into the nursing anomaly analyzer is 15, 5 of the 15 nursing anomaly analysis branches are randomly called. The corrected blood glucose deviation (0.328 mmol / L) is then input into each of the five nursing anomaly analysis branches, resulting in five nursing anomaly patterns. The final nursing anomaly pattern is selected as the one with the highest occurrence rate. For example, if the first nursing anomaly analysis branch outputs "excessive stress," the second outputs "overeating," the third outputs "excessive stress," the fourth outputs "alcohol consumption," and the fifth outputs "excessive stress," and the "excessive stress" pattern has the highest occurrence rate (0.6), then "excessive stress" is selected as the final nursing anomaly pattern.

[0091] Furthermore, the training steps of the "nursing abnormality analyzer" include:

[0092] In the historical nursing data, a set of sample blood glucose deviations is collected, and nursing abnormality patterns under different sample blood glucose deviations are collected, and the sample nursing abnormality pattern set is obtained by annotation;

[0093] Performing random division with replacement on the sample blood glucose deviation set and the sample nursing abnormality pattern set to obtain multiple sets of nursing abnormality analysis training data;

[0094] Machine learning is used to construct multiple nursing anomaly analysis branches. The multiple nursing anomaly analysis training data are used respectively to perform supervised training on the multiple nursing anomaly analysis branches until the accuracy converges, thereby obtaining a nursing anomaly analyzer.

[0095] In this embodiment of the present application, a set of sample blood glucose deviations is first collected from the historical nursing data, and abnormal nursing patterns under different sample blood glucose deviations are collected and labeled to obtain a set of sample abnormal nursing patterns. For example, abnormal nursing patterns (overeating, poor sleep quality) under a blood glucose deviation (0.4mmol / L) in the historical nursing data are collected, and overeating and poor sleep quality are labeled as abnormal labels for a blood glucose deviation of 0.4mmol / L.

[0096] Secondly, the sample blood glucose deviation set and the sample nursing anomaly pattern set were randomly partitioned with replacement to obtain multiple sets of training data for nursing anomaly analysis. The sample blood glucose deviation set and the sample nursing anomaly pattern set were randomly resampled with replacement to generate multiple differentiated training subsets (e.g., 10). Each subset retained approximately 63.2% of the original data (e.g., 632 data items were extracted from 1000 data items). The remaining 36.8% of unselected samples automatically formed an out-of-bag (OOB) validation set. By forcing random combinations of sample duplications and omissions, data diversity was effectively enhanced, data sparsity was alleviated, and overfitting of a single model to noisy data was avoided.

[0097] Finally, machine learning is used to construct multiple nursing anomaly analysis branches. Each branch is supervised trained using the multiple nursing anomaly analysis training data until the accuracy converges, resulting in a nursing anomaly analyzer. The nursing anomaly analyzer employs a multi-branch ensemble architecture, with each branch being an independent lightweight model (such as 1D-CNN, GBDT, or logistic regression). Furthermore, the training process can be implemented through the following technical approaches: 1. Prepare the training dataset: Split the multiple nursing anomaly analysis training data sets into training, validation, and test sets in a ratio of 7.5:1.5:1.5. 2. Model training: Branch-by-branch training is performed using a weighted cross-entropy loss function (with rare patterns weighted 5x). Adam dynamic learning rate decay and early stopping are used to prevent overfitting. Ultimately, achieving an accuracy of over 90% on the test set is considered model convergence.

[0098] In summary, compared to the existing technology, this application inputs the corrected blood sugar deviation into the nursing abnormality dimension classification table, classifies the nursing abnormality dimension, and then configures the nursing abnormality prediction resource based on the nursing abnormality dimension. The corrected blood sugar deviation is then input into the nursing abnormality analyzer, and the nursing abnormality pattern is output. Finally, based on the nursing abnormality pattern, dynamic adjustment and recommendation of the rehabilitation plan are made. In this way, the nursing abnormality pattern that leads to abnormal blood sugar is obtained, and the rehabilitation plan can be adjusted in a timely manner accordingly.

[0099] In summary, the embodiments of the present application have at least the following technical effects:

[0100] Compared to existing technologies, this application first collects nursing data and rehabilitation plans for target users over multiple time periods within a historical time window, and then establishes nursing data sequences and rehabilitation plan sequences with time series characteristics based on the time periods. In this way, the time-series data sequence fully preserves the dynamic relationship between blood sugar fluctuations and rehabilitation plans, providing the necessary data support for the dynamic recommendation of subsequent rehabilitation plans.

[0101] Next, based on the target user's characteristic information, multiple approximate users are screened. The blood glucose levels of these approximate users after the nursing data sequence and rehabilitation program sequence are extracted and labeled as a sample predicted blood glucose set. This is then used as a training set to train and optimize a blood glucose predictor constructed using machine learning. Finally, the blood glucose predictor is input with the nursing data sequence and rehabilitation program sequence, resulting in a predicted output for the target user's predicted blood glucose. This yields the target user's ideal blood glucose value under the nursing data sequence and rehabilitation program sequence, providing the necessary data support for subsequent analysis of abnormal blood glucose levels.

[0102] Thirdly, the present application performs a double compensation correction on the calculated blood glucose deviation to obtain the corrected blood glucose deviation: first, a compensation correction calculation is performed based on the blood glucose prediction error to obtain the predicted corrected blood glucose deviation, and then a compensation correction calculation is performed on the predicted corrected blood glucose deviation based on the blood glucose fluctuation parameters of the target user to obtain the corrected blood glucose deviation. In this way, the blood glucose prediction error and blood glucose fluctuation parameters are compensated and corrected, eliminating the interference between the prediction error and individual physiological fluctuations, and improving the accuracy of blood glucose abnormality analysis.

[0103] Finally, this application inputs the corrected blood sugar deviation into the nursing abnormality dimension classification table, classifies the nursing abnormality dimension, and then configures the nursing abnormality prediction resource based on the nursing abnormality dimension. The corrected blood sugar deviation is then input into the nursing abnormality analyzer, and the nursing abnormality pattern is output. Finally, based on the nursing abnormality pattern, dynamic adjustment and recommendation of the rehabilitation plan are made. In this way, the nursing abnormality pattern that leads to abnormal blood sugar is obtained, and the rehabilitation plan can be adjusted in a timely manner accordingly.

[0104] Through the above technical solution, this application collects the nursing data sequence and rehabilitation program sequence of the target user within the historical time window; then uses the nursing data sequence and rehabilitation program sequence to predict the blood sugar of the target user and obtain the predicted blood sugar; then collects the actual blood sugar of the target user, combines the predicted blood sugar, calculates the blood sugar deviation, and compensates and corrects the blood sugar deviation based on the blood sugar prediction error and the blood sugar fluctuation parameter of the target user to obtain the corrected blood sugar deviation; finally, based on the corrected blood sugar deviation, classifies and obtains the nursing abnormality dimension, configures the nursing abnormality prediction resources, performs nursing abnormality pattern analysis, obtains the nursing abnormality pattern, and performs dynamic adjustment and recommendation of the rehabilitation program. In this way, the nursing abnormality pattern that causes abnormal blood sugar is obtained, and the rehabilitation program is adjusted in time accordingly, thereby improving the adaptability of the rehabilitation program.

[0105] Example 2, as Figure 2 As shown, based on the same inventive concept as the method for dynamically recommending rehabilitation programs in clinical nursing provided in Example 1, an embodiment of the present invention further provides a dynamic recommendation system for rehabilitation programs in clinical nursing, comprising:

[0106] The data collection module 11 is used to collect the nursing data sequence and rehabilitation program sequence of the target user within the historical time window, wherein the rehabilitation program includes rehabilitation exercise data;

[0107] A blood glucose prediction module 12 is configured to use the nursing data sequence and the rehabilitation program sequence to perform blood glucose prediction on the target user and obtain predicted blood glucose;

[0108] The compensation and correction module 13 is configured to collect the actual blood glucose of the target user, combine it with the predicted blood glucose, calculate the blood glucose deviation, and compensate and correct the blood glucose deviation based on the blood glucose prediction error and the blood glucose fluctuation parameter of the target user to obtain a corrected blood glucose deviation;

[0109] The adjustment recommendation module 14 is used to classify and obtain nursing abnormality dimensions based on the corrected blood sugar deviation, configure nursing abnormality prediction resources, perform nursing abnormality pattern analysis, obtain nursing abnormality patterns, and make dynamic adjustment recommendations for rehabilitation programs.

[0110] The data acquisition module 11 is specifically used for:

[0111] Collecting nursing data and rehabilitation plans of the target user over multiple time periods within a historical time window, wherein the rehabilitation plan includes rehabilitation exercise data;

[0112] Integrate nursing data and rehabilitation plans in multiple time periods to obtain nursing data sequences and rehabilitation plan sequences.

[0113] The blood glucose prediction module 12 is specifically configured to:

[0114] Calling a blood glucose predictor that has been trained in advance, wherein the blood glucose predictor is trained using sample nursing data of an approximate user;

[0115] The nursing data sequence and the rehabilitation program sequence are input into the blood glucose predictor, and the predicted blood glucose of the target user is obtained by the prediction output.

[0116] Furthermore, the training steps of the "blood sugar predictor" include:

[0117] Obtaining user characteristic information of the target user, and screening to obtain a plurality of similar users based on the user characteristic information;

[0118] According to the nursing data of the plurality of approximate users in the historical time, a sample nursing data sequence set and a sample rehabilitation program sequence set are collected, and the blood glucose of the approximate user after collecting different sample nursing data sequences and sample rehabilitation program sequences is approximated and marked as a sample predicted blood glucose set;

[0119] Use machine learning to build a blood sugar predictor;

[0120] The blood glucose predictor is supervised trained and tested using the sample nursing data sequence set, the sample rehabilitation program sequence set and the sample predicted blood glucose set. When the test accuracy is greater than or equal to the accuracy threshold, the training is completed and the blood glucose prediction accuracy is obtained.

[0121] The compensation and correction module 13 is specifically configured to:

[0122] Collecting the actual blood glucose of the target user, calculating the difference between the actual blood glucose and the predicted blood glucose, and obtaining a blood glucose deviation;

[0123] Obtain the blood sugar prediction accuracy of the blood sugar prediction and calculate the blood sugar prediction error rate;

[0124] performing compensation correction calculation on the blood glucose deviation according to the blood glucose prediction error rate to obtain a predicted and corrected blood glucose deviation;

[0125] Obtaining a blood glucose fluctuation value of the target user under normal care status as a blood glucose fluctuation parameter;

[0126] The predicted and corrected blood glucose deviation is compensated and corrected based on the blood glucose fluctuation parameter to obtain the corrected blood glucose deviation.

[0127] The adjustment recommendation module 14 is specifically configured to:

[0128] Based on the nursing data in the historical time, the sample blood glucose deviation set is collected, and all the nursing abnormality patterns that appear under different sample blood glucose deviations are collected, and the number of nursing abnormality patterns is extracted to obtain the sample nursing abnormality dimension set;

[0129] Constructing a mapping relationship between the sample blood glucose deviation set and the sample nursing abnormality dimension set to obtain a nursing abnormality dimension classification table;

[0130] Inputting the corrected blood sugar deviation into the nursing abnormality dimension classification table, and classifying to obtain the nursing abnormality dimension;

[0131] According to the nursing abnormality dimension, a nursing abnormality prediction resource is configured, and a nursing abnormality pattern analysis is performed on the corrected blood glucose deviation to obtain a nursing abnormality pattern;

[0132] Based on the abnormal nursing pattern, dynamic adjustment recommendations for the rehabilitation plan are made.

[0133] Specifically, the "configuring nursing abnormality prediction resources according to the nursing abnormality dimension, performing nursing abnormality pattern analysis on the corrected blood glucose deviation, and obtaining a nursing abnormality pattern" includes:

[0134] Obtain the total number of nursing abnormality patterns as the maximum nursing abnormality dimension;

[0135] Calculating the ratio of the nursing abnormality dimension to the maximum nursing abnormality dimension to obtain a nursing abnormality prediction resource coefficient;

[0136] Calling a pre-trained nursing anomaly analyzer including multiple nursing anomaly analysis branches, and calculating the number of nursing branches according to the nursing anomaly prediction resource coefficient and the number of the multiple nursing anomaly analysis branches;

[0137] Randomly call the nursing abnormality analysis branch of the nursing branch number, input the corrected blood sugar deviation, output the nursing abnormality pattern of the nursing branch number, and screen the nursing abnormality pattern with the largest occurrence ratio.

[0138] Furthermore, the training steps of the "nursing abnormality analyzer" include:

[0139] In the historical nursing data, a set of sample blood glucose deviations is collected, and nursing abnormality patterns under different sample blood glucose deviations are collected, and the sample nursing abnormality pattern set is obtained by annotation;

[0140] Performing random division with replacement on the sample blood glucose deviation set and the sample nursing abnormality pattern set to obtain multiple sets of nursing abnormality analysis training data;

[0141] Machine learning is used to construct multiple nursing anomaly analysis branches. The multiple nursing anomaly analysis training data are used respectively to perform supervised training on the multiple nursing anomaly analysis branches until the accuracy converges, thereby obtaining a nursing anomaly analyzer.

[0142] In summary, the embodiments of the present application have at least the following technical effects:

[0143] The data acquisition module collects nursing data and rehabilitation plans for a target user over multiple time periods within a historical time window and constructs a nursing data sequence and rehabilitation plan sequence with time series characteristics based on the time periods, providing the necessary data support for subsequent dynamic rehabilitation plan recommendations. The blood glucose prediction module uses the nursing data sequence and rehabilitation plan sequence to predict the target user's blood glucose level, obtaining a predicted blood glucose level—the target user's ideal blood glucose level under the nursing data sequence and rehabilitation plan sequence. The compensation and correction module collects the target user's actual blood glucose level and, combining it with the predicted blood glucose level, calculates the blood glucose deviation. Based on the blood glucose prediction error and the target user's blood glucose fluctuation parameters, it compensates for the deviation to obtain the corrected blood glucose deviation. This eliminates the interference between the prediction error and individual physiological fluctuations and improves the accuracy of blood glucose abnormality analysis. The adjustment and recommendation module classifies the corrected blood glucose deviation to obtain nursing abnormality dimensions, allocates nursing abnormality prediction resources, analyzes nursing abnormality patterns, obtains nursing abnormality patterns, and dynamically adjusts rehabilitation plans. This identifies the nursing abnormality patterns that lead to blood glucose abnormalities and allows for timely adjustments to rehabilitation plans, thereby improving the adaptability of rehabilitation plans.

[0144] It should be noted that, in the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0145] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0146] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0147] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0148] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0149] Although preferred embodiments of the present invention have been described, additional changes and modifications to these embodiments may occur to those skilled in the art once the basic inventive concepts become known.

[0150] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the present invention and its equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for dynamically recommending rehabilitation programs in clinical nursing, characterized in that: The method comprises: Collecting nursing data sequences and rehabilitation program sequences of the target user within a historical time window, wherein the rehabilitation program includes rehabilitation exercise data; Using the nursing data sequence and rehabilitation program sequence, predicting the blood glucose of the target user to obtain predicted blood glucose; Collecting the actual blood glucose of the target user, combining it with the predicted blood glucose, calculating a blood glucose deviation, and compensating and correcting the blood glucose deviation based on the blood glucose prediction error and the blood glucose fluctuation parameter of the target user to obtain a corrected blood glucose deviation; According to the corrected blood sugar deviation, the nursing abnormality dimension is obtained by classification, nursing abnormality prediction resources are configured, nursing abnormality pattern analysis is performed, nursing abnormality pattern is obtained, and dynamic adjustment recommendation of rehabilitation plan is made, including: Based on the nursing data in the historical time, the sample blood glucose deviation set is collected, and all the nursing abnormality patterns that appear under different sample blood glucose deviations are collected, and the number of nursing abnormality patterns is extracted to obtain the sample nursing abnormality dimension set; Constructing a mapping relationship between the sample blood glucose deviation set and the sample nursing abnormality dimension set to obtain a nursing abnormality dimension classification table; Inputting the corrected blood sugar deviation into the nursing abnormality dimension classification table, and classifying to obtain the nursing abnormality dimension; According to the nursing abnormality dimension, a nursing abnormality prediction resource is configured, and a nursing abnormality pattern analysis is performed on the corrected blood glucose deviation to obtain a nursing abnormality pattern, including: Obtain the total number of nursing abnormality patterns as the maximum nursing abnormality dimension; Calculating the ratio of the nursing abnormality dimension to the maximum nursing abnormality dimension to obtain a nursing abnormality prediction resource coefficient; Calling a pre-trained nursing anomaly analyzer including multiple nursing anomaly analysis branches, and calculating the number of nursing branches based on the nursing anomaly prediction resource coefficient and the number of the multiple nursing anomaly analysis branches, wherein the number of nursing branches is obtained by multiplying the nursing anomaly prediction resource coefficient by the total number of the multiple nursing anomaly analysis branches and rounding the result; Randomly calling the nursing abnormality analysis branch of the number of nursing branches, inputting the corrected blood glucose deviation, outputting the nursing abnormality pattern of the number of nursing branches, and screening the nursing abnormality pattern with the largest occurrence ratio; Based on the abnormal nursing pattern, dynamic adjustment recommendations for the rehabilitation plan are made.

2. The method for dynamic recommendation of rehabilitation programs in clinical nursing according to claim 1, characterized in that: Collect the target user's nursing data sequence and rehabilitation plan sequence within the historical time window, including: Collecting nursing data and rehabilitation plans of the target user over multiple time periods within a historical time window, wherein the rehabilitation plan includes rehabilitation exercise data; Integrate nursing data and rehabilitation plans in multiple time periods to obtain nursing data sequences and rehabilitation plan sequences.

3. The method for dynamic recommendation of rehabilitation programs in clinical nursing according to claim 1, characterized in that: Using the nursing data sequence and the rehabilitation program sequence to predict the blood glucose of the target user to obtain the predicted blood glucose includes: Calling a blood glucose predictor that has been trained in advance, wherein the blood glucose predictor is trained using sample nursing data of an approximate user; The nursing data sequence and the rehabilitation program sequence are input into the blood glucose predictor, and the predicted blood glucose of the target user is obtained by the prediction output.

4. The method for dynamic recommendation of rehabilitation programs in clinical nursing according to claim 3, characterized in that: The training step of the blood glucose predictor comprises: Obtaining user characteristic information of the target user, and screening to obtain a plurality of similar users based on the user characteristic information; According to the nursing data of the plurality of approximate users in the historical time, a sample nursing data sequence set and a sample rehabilitation program sequence set are collected, and the blood glucose of the approximate user after collecting different sample nursing data sequences and sample rehabilitation program sequences is approximated and marked as a sample predicted blood glucose set; Use machine learning to build a blood sugar predictor; The blood glucose predictor is supervised trained and tested using the sample nursing data sequence set, the sample rehabilitation program sequence set and the sample predicted blood glucose set. When the test accuracy is greater than or equal to the accuracy threshold, the training is completed and the blood glucose prediction accuracy is obtained.

5. The method for dynamic recommendation of rehabilitation programs in clinical nursing according to claim 1, characterized in that: The actual blood glucose of the target user is collected, combined with the predicted blood glucose, to calculate a blood glucose deviation, and the blood glucose deviation is compensated and corrected according to the blood glucose prediction error and the blood glucose fluctuation parameter of the target user to obtain a corrected blood glucose deviation, including: Collecting the actual blood glucose of the target user, calculating the difference between the actual blood glucose and the predicted blood glucose, and obtaining a blood glucose deviation; Obtain the blood sugar prediction accuracy of the blood sugar prediction and calculate the blood sugar prediction error rate; performing compensation correction calculation on the blood glucose deviation according to the blood glucose prediction error rate to obtain a predicted and corrected blood glucose deviation; Obtaining a blood glucose fluctuation value of the target user under normal care status as a blood glucose fluctuation parameter; The predicted and corrected blood glucose deviation is compensated and corrected based on the blood glucose fluctuation parameter to obtain the corrected blood glucose deviation.

6. The method for dynamic recommendation of rehabilitation programs in clinical nursing according to claim 1, characterized in that: The training steps of the nursing abnormality analyzer include: In the historical nursing data, a set of sample blood glucose deviations is collected, and nursing abnormality patterns under different sample blood glucose deviations are collected, and the sample nursing abnormality pattern set is obtained by annotation; Performing random division with replacement on the sample blood glucose deviation set and the sample nursing abnormality pattern set to obtain multiple sets of nursing abnormality analysis training data; Machine learning is used to construct multiple nursing anomaly analysis branches. The multiple nursing anomaly analysis training data are used respectively to perform supervised training on the multiple nursing anomaly analysis branches until the accuracy converges, thereby obtaining a nursing anomaly analyzer.

7. A dynamic recommendation system for rehabilitation programs in clinical nursing, characterized by: Used to perform the method according to any one of claims 1 to 6, comprising: A data collection module is used to collect nursing data sequences and rehabilitation program sequences of target users within a historical time window, wherein the rehabilitation program includes rehabilitation exercise data; A blood glucose prediction module, configured to use the nursing data sequence and the rehabilitation program sequence to predict the blood glucose of the target user and obtain predicted blood glucose; a compensation and correction module, configured to collect the actual blood glucose of the target user, combine it with the predicted blood glucose, calculate a blood glucose deviation, and perform compensation and correction on the blood glucose deviation based on the blood glucose prediction error and the blood glucose fluctuation parameter of the target user to obtain a corrected blood glucose deviation; The adjustment recommendation module is used to classify and obtain nursing abnormality dimensions based on the corrected blood sugar deviation, configure nursing abnormality prediction resources, perform nursing abnormality pattern analysis, obtain nursing abnormality patterns, and make dynamic adjustment recommendations for rehabilitation plans, including: Based on the nursing data in the historical time, the sample blood glucose deviation set is collected, and all the nursing abnormality patterns that appear under different sample blood glucose deviations are collected, and the number of nursing abnormality patterns is extracted to obtain the sample nursing abnormality dimension set; Constructing a mapping relationship between the sample blood glucose deviation set and the sample nursing abnormality dimension set to obtain a nursing abnormality dimension classification table; Inputting the corrected blood sugar deviation into the nursing abnormality dimension classification table, and classifying to obtain the nursing abnormality dimension; According to the nursing abnormality dimension, a nursing abnormality prediction resource is configured, and a nursing abnormality pattern analysis is performed on the corrected blood glucose deviation to obtain a nursing abnormality pattern, including: Obtain the total number of nursing abnormality patterns as the maximum nursing abnormality dimension; Calculating the ratio of the nursing abnormality dimension to the maximum nursing abnormality dimension to obtain a nursing abnormality prediction resource coefficient; Calling a pre-trained nursing anomaly analyzer including multiple nursing anomaly analysis branches, and calculating the number of nursing branches based on the nursing anomaly prediction resource coefficient and the number of the multiple nursing anomaly analysis branches, wherein the number of nursing branches is obtained by multiplying the nursing anomaly prediction resource coefficient by the total number of the multiple nursing anomaly analysis branches and rounding the result; Randomly calling the nursing abnormality analysis branch of the number of nursing branches, inputting the corrected blood glucose deviation, outputting the nursing abnormality pattern of the number of nursing branches, and screening the nursing abnormality pattern with the largest occurrence ratio; Based on the abnormal nursing pattern, dynamic adjustment recommendations for the rehabilitation plan are made.

Images