Personalized burn rehabilitation path planning method and system

Abstract

The invention relates to the technical field of physiological information processing, and discloses a personalized burn rehabilitation path planning method and system, and the method comprises the steps: obtaining historical rehabilitation data and real-time evaluation data, constructing a state time sequence, extracting a score fluctuation vector and an emotion change vector, and carrying out the window adjustment processing of a fluctuation feature vector set, a rhythm factor and a training factor are calculated in each window range, so that path analysis has the capability of synchronously evaluating rhythm change and a training state, the phenomenon of excessive path adjustment caused by short-time abnormal fluctuation is avoided, a rhythm feature sequence is input into a whale optimization algorithm to execute path updating processing, and the path updating efficiency is improved. The method ensures that the path generation process has the dynamic adjustment capability, enhances the adaptability of the algorithm to the recovery rhythm change, performs grading processing on each stage according to the stage sequence set, can effectively output the rehabilitation path matrix, and realizes the unified expression of the multi-stage rhythm and the training grade.

Description

Technical Field

[0001] The present invention relates to the field of physiological information processing technology, and more specifically, to a personalized burn rehabilitation path planning method and system. Background Art

[0002] In existing rehabilitation path planning technologies, a phased control method based on evaluation indicators is usually adopted to divide the patient's rehabilitation process into multiple stages. Different training goals and intervention intensities are set for each stage in combination with functional assessment scales, physiological data or clinical experience rules. Some methods introduce machine learning or expert systems to analyze the patient's basic characteristics in order to recommend personalized rehabilitation paths. However, most of these methods are based on static indicators. Once the path planning plan is set, the adjustment flexibility during the implementation process is low. In addition, the impact of state fluctuations and rhythm changes in the patient's rehabilitation process on path continuity and rhythm matching is generally ignored, which can easily lead to reduced rehabilitation efficiency or intervention rhythm disorder.

[0003] However, existing technologies have the risk of over-response when dealing with short-term fluctuations in evaluation indicators during the rehabilitation process. Especially in multi-cycle or long-course paths, it is easy to frequently adjust the path strategy due to small abnormal fluctuations in indicators, resulting in stage mismatches or repeated corrections in the rehabilitation path, reducing the stability and rationality of the path. At the same time, existing optimization algorithms are mostly based on fixed strategies or static search strategies, and lack responsive adjustment mechanisms when dealing with dynamic changes in patients' recovery rhythm. It is difficult to adapt to the recovery speed and rhythm changes of different patients. Therefore, how to reduce the risk of path over-adjustment caused by fluctuations in rehabilitation indicators and achieve the rationality of the stage output of the rehabilitation path under the conditions of adaptability of intelligent optimization algorithms to changes in recovery rhythm has become a key technical problem that needs to be solved urgently.

[0004] In view of this, the present invention proposes a personalized burn rehabilitation path planning method and system to solve the above problems. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art, the present invention provides a personalized burn rehabilitation path planning method and system.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] First, a personalized burn rehabilitation pathway planning method is provided, including:

[0008] Acquire historical rehabilitation data and real-time assessment data, perform alignment processing on the historical rehabilitation data and the real-time assessment data based on the time axis, and obtain a state time series;

[0009] Perform stability calculation processing on each time period according to the state time series, extract the score fluctuation vector and the sentiment change vector, and construct a set of fluctuation feature vectors based on the score fluctuation vector and the sentiment change vector;

[0010] Performing window adjustment processing on the set of fluctuation feature vectors, setting the window range according to the score fluctuation vector and the emotion change vector, calculating the rhythm factor and training factor within each window range, and obtaining the rhythm feature sequence based on the rhythm factor and training factor;

[0011] Input the rhythm feature sequence into the whale optimization algorithm, perform path update processing, and output a set of stage sequences;

[0012] The stages are graded according to the stage sequence set to obtain the rehabilitation pathway matrix.

[0013] In some embodiments, the historical rehabilitation data includes a training item sequence and a stage score, and the real-time assessment data includes muscle activity values ​​and emotion change values. A method for aligning the historical rehabilitation data and the real-time assessment data based on a time axis to obtain a state time series includes:

[0014] According to the training item sequence and stage scores in the historical rehabilitation data, the timestamp corresponding to each training record is extracted to construct a training time index;

[0015] Based on the muscle activity values ​​and emotion change values ​​in the real-time assessment data, the collection time corresponding to each assessment record is extracted to construct an assessment time index;

[0016] Perform index alignment processing on the training time index and the evaluation time index at fixed time intervals, and combine the corresponding training items, stage scores, muscle activity values, and emotion change values ​​at the aligned time points to generate a joint time series data structure;

[0017] Arrange the joint time series data structure in ascending time order to construct a complete state time series.

[0018] In some embodiments, the method of setting a window range based on a rating fluctuation vector and a sentiment change vector includes:

[0019] The score fluctuation vector and the emotion change vector are respectively indexed into sequence pairs according to the time index, and the fluctuation feature vector set is constructed based on the maximum difference between the score fluctuation value and the emotion fluctuation value in several adjacent pairs of sequences;

[0020] Perform sliding window statistical processing on the set of fluctuation feature vectors, and calculate the sum of the change gradients of the score fluctuation value and the sentiment change value in each window as the fluctuation density indicator;

[0021] Compare the volatility concentration index with a preset volatility threshold. When the volatility concentration index is greater than the preset volatility threshold, reduce the sliding window length. When the volatility concentration index is less than or equal to the preset volatility threshold, expand the sliding window length.

[0022] The adjusted sliding window length is used as the basis for time period division, and the time window segmentation processing is performed on the set of fluctuation feature vectors to set the window range.

[0023] In some embodiments, a method for obtaining a rhythm feature sequence based on a rhythm factor and a training factor includes:

[0024] Obtain a rhythm factor sequence and a training factor sequence, wherein the rhythm factor sequence is composed of multiple rhythm factors, and the training factor sequence is composed of multiple training factors, and the rhythm factor sequence and the training factor sequence correspond to each other in time sequence to form a dual-channel sequence pair set;

[0025] According to the rhythm factor sequence and the training factor sequence, the rhythm factor and the training factor corresponding to each time period are extracted, and the set of values ​​are combined into a two-dimensional feature pair to form a sliding window feature pair sequence;

[0026] Perform partition mapping processing on the sliding window feature pair sequence, determine the two-dimensional plane region index where the set of features is located based on the normalized values ​​of the rhythm factor and the training factor, and construct the original rhythm index sequence;

[0027] Perform smoothing and denoising on the original rhythm index sequence, adopt the consistent weight distribution of rhythm indexes in adjacent time periods, perform attenuation control on local high-frequency changes, and generate a smooth rhythm index sequence;

[0028] The rhythm levels are arranged in chronological order according to the smoothed rhythm index sequence to construct a rhythm feature sequence.

[0029] In a second aspect, a personalized burn rehabilitation path planning system is provided, which is used to implement the above-mentioned personalized burn rehabilitation path planning method, including:

[0030] Sequence construction module: used to obtain historical rehabilitation data and real-time assessment data, perform alignment processing on the historical rehabilitation data and real-time assessment data based on the time axis to obtain the state time series;

[0031] Vector construction module: used to perform stability calculation processing for each time period based on the state time series, extract the score fluctuation vector and sentiment change vector, and construct a fluctuation feature vector set based on the score fluctuation vector and sentiment change vector;

[0032] The first processing module is used to perform window adjustment processing on the fluctuation feature vector set, set the window range according to the score fluctuation vector and the emotion change vector, calculate the rhythm factor and training factor within each window range, and obtain the rhythm feature sequence based on the rhythm factor and training factor;

[0033] The second processing module is used to input the rhythm feature sequence into the whale optimization algorithm, perform path update processing, and output a stage sequence set;

[0034] Path planning module: used to perform hierarchical classification processing on each stage according to the stage sequence set to obtain the rehabilitation path matrix.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] The present invention first obtains historical rehabilitation data and real-time evaluation data, constructs a state time series, and extracts score fluctuation vectors and emotion change vectors, thereby solving the problem in the existing technology that the path planning process cannot accurately capture the patient's state fluctuations. Furthermore, by performing window adjustment processing on the fluctuation feature vector set, the rhythm factor and training factor are calculated within each window range, so that the path analysis has the ability to synchronously evaluate the rhythm change and training state, thereby avoiding the phenomenon of excessive path adjustment caused by short-term abnormal fluctuations. The rhythm feature sequence is input into the whale optimization algorithm to perform path update processing, ensuring that the path generation process has dynamic adjustment capabilities, enhancing the algorithm's adaptability to recovery rhythm changes, and performing level division processing on each stage according to the stage sequence set. The rehabilitation path matrix can be effectively output, realizing the unified expression of multi-stage rhythm and training level. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 Schematic diagram of the process of the personalized burn rehabilitation pathway planning method of the present invention;

[0038] Figure 2 Schematic diagram of the structure of the personalized burn rehabilitation path planning system in the present invention. DETAILED DESCRIPTION

[0039] In order to make the objectives, technical solutions and advantages of the present invention more clear, the present invention is further described in detail below in combination with specific embodiments and with reference to the accompanying drawings. In the following detailed description, many specific details are set forth to provide a thorough understanding of the described exemplary embodiments. However, it is obvious to those skilled in the art that the described embodiments can be practiced without some or all of these specific details. In other exemplary embodiments, well-known structures are not described in detail to avoid unnecessarily obscuring the concepts of the present disclosure. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. At the same time, the various aspects described in the embodiments can be arbitrarily combined without conflict.

[0040] Example 1

[0041] See also Figure 1 As shown, this embodiment discloses a personalized burn rehabilitation path planning method, including:

[0042] S10: Acquire historical rehabilitation data and real-time assessment data, perform alignment processing on the historical rehabilitation data and the real-time assessment data based on the time axis, and obtain a state time series;

[0043] In this embodiment, historical rehabilitation data refers to the training task information and corresponding evaluation results recorded during the patient's previous rehabilitation process, specifically including a sequence of training items arranged in chronological order, stage-by-stage scoring values, training duration, etc., which are used to reflect the patient's training content and recovery performance at each stage.

[0044] Real-time assessment data refers to quantitative information related to the patient's status obtained in real time by sensor equipment or assessment systems during the current rehabilitation process, including muscle activity values ​​(such as electromyographic signal amplitude and activation frequency) and emotional change values ​​(such as emotional indexes derived from voice, facial expressions or self-assessment questionnaires), which are used to characterize the patient's current physical reactions and emotional state.

[0045] For example, historical rehabilitation data include training project sequences and stage scores. The training project sequence refers to the collection of names or numbers of training tasks that the patient actually participated in during the rehabilitation process, which are arranged in chronological order to form an ordered sequence. Common training projects include upper limb extension, weighted walking, and joint range of motion improvement training. This sequence is usually automatically generated by the task recording module of the rehabilitation guidance system or training equipment. The stage score refers to the quantitative evaluation value given by the rehabilitation assessor or system to the patient's performance in a specific rehabilitation stage. Common forms include FIM scores, VAS scores, or muscle strength level scores, etc., and can be obtained through regular assessment and scoring, automatic system-generated scores, or calculations after sorting out patient questionnaires.

[0046] Real-time assessment data includes muscle activity values ​​and emotion change values. Muscle activity values ​​refer to muscle electrical signal parameters collected by surface electromyography sensors or wearable devices during training or rest. Common indicators include the mean amplitude of the electromyography signal, the integrated electromyography value, the spectral energy density, etc., which are used to characterize the degree and stability of muscle activation. Emotion change values ​​refer to the emotion index value derived during training or assessment intervals through voice intonation analysis, facial expression image recognition, psychological state questionnaire scoring, etc. Common representation methods include a 1-5 level emotion scale or a continuous emotion curve based on time series changes, which are used to measure the patient's psychological fluctuation state.

[0047] In this embodiment, the historical rehabilitation data includes training item sequences and stage scores, and the real-time assessment data includes muscle activity values ​​and emotion change values. The method for aligning the historical rehabilitation data and the real-time assessment data based on the time axis to obtain the state time series includes:

[0048] According to the training item sequence and stage scores in the historical rehabilitation data, the timestamp corresponding to each training record is extracted to construct a training time index;

[0049] Based on the muscle activity values ​​and emotion change values ​​in the real-time assessment data, the collection time corresponding to each assessment record is extracted to construct an assessment time index;

[0050] Perform index alignment processing on the training time index and the evaluation time index at fixed time intervals, and combine the corresponding training items, stage scores, muscle activity values, and emotion change values ​​at the aligned time points to generate a joint time series data structure;

[0051] Arrange the joint time series data structure in ascending time order to construct a complete state time series.

[0052] In this embodiment, alignment processing is performed on historical rehabilitation data and real-time evaluation data based on the time axis, which means extracting the training record time from the training item sequence and stage score, extracting the evaluation collection time from the muscle activity value and the emotion change value, constructing the training time index and the evaluation time index respectively, and synchronously aligning the two types of indexes at preset time intervals. At each alignment point, the corresponding training information and evaluation information are combined into data items in a unified format to form a joint time series data structure, and then the structure is sorted in chronological order to construct a state time series. The main purpose of this step is to ensure that data from different sources remain consistent in the time dimension, and to provide a unified input basis for subsequent stability calculations and path planning by time period.

[0053] S20: Performing stability calculation processing on each time period according to the state time series, extracting the score fluctuation vector and the emotion change vector, and constructing a fluctuation feature vector set based on the score fluctuation vector and the emotion change vector;

[0054] The method of performing stability calculation processing for each time period based on the state time series and extracting the score fluctuation vector includes:

[0055] Divide the state time series into continuous non-overlapping time periods according to a fixed time length T, and let the jth segment contain the score sequence ;

[0056] Perform improved variance superposition calculation on each score sequence, using the following score fluctuation calculation formula:

[0057] ;

[0058] in, It is The score fluctuation value corresponding to the segment score sequence, Indicates the total number of scoring points contained in this scoring sequence. Represents the counting index used to traverse the scoring points in the scoring sequence; Indicates the Section 1 The rating value of the rating points, Represents the average value of the rating sequence of this segment; represents the average of the squared deviations of the rating values ​​from the mean, is the rhythmic disturbance term, where is the adjustment coefficient, Represents the sum of the absolute differences between adjacent rating values.

[0059] In this embodiment, the state time series is divided into continuous non-overlapping time periods according to a fixed time length T, and after the score sequence in each segment is extracted, an improved variance superposition calculation process is performed. This means that in each score sequence, the average of the square deviations of the score values ​​of the segment relative to the mean is first calculated and used as the basic fluctuation term. On the basis of the original, a rhythm disturbance term is introduced into the score sequence, that is, the average of the absolute differences between adjacent score points, as a sensitive enhancement term for short-term drastic changes within the sequence. This improves the ability to respond to "intermittent jumps" or "short-term deterioration" in rehabilitation training while maintaining the ability to capture the overall fluctuation trend. The superposition of these two items together constitutes the score fluctuation value, which effectively enhances the robust recognition ability of nonlinear rhythm fluctuations in rehabilitation task evaluation.

[0060] It should be noted that the rhythm disturbance term is introduced as the core enhancement module in the score fluctuation calculation in this embodiment. Its essential principle is: by counting the absolute change values ​​between adjacent score points, a quantitative signal of score trend transfer or rhythm interruption in the stage score sequence is constructed. In the rehabilitation assessment scenario, patient scores are often affected by external assistance, fatigue level or environmental fluctuations. The probability of short-term drastic fluctuations is much higher than the data sequence under the natural physiological state. If only the variance term is used, it will be difficult to characterize these rhythmic abnormalities. Therefore, this embodiment proposes a disturbance detection structure centered on the micro-grade changes between score points. Compared with the existing methods such as sliding average and weighted sliding, this structure does not rely on window delay but directly faces local differences, and has higher time response characteristics. At the same time, the disturbance term is globally controlled through the μ adjustment coefficient, and the disturbance sensitivity can be dynamically adjusted according to different rehabilitation tasks, further improving the adaptability of the system in various rehabilitation scenarios.

[0061] The method of performing stability calculation processing on each time period according to the state time series and extracting the emotion change vector includes:

[0062] Divide the state time series into continuous non-overlapping time periods according to a fixed time length T, and let the jth segment contain the emotion sequence ; Perform improved fluctuation calculation on each emotion sequence, using the following emotion change value calculation formula:

[0063] ;

[0064] in, It is The emotional fluctuation value corresponding to the emotional sequence, Indicates the total number of emotion points contained in this emotion sequence. Indicates the counting index used to traverse the emotion points in the emotion sequence. Indicates the Section 1 The emotional value of the emotional point, represents the average value of the emotion sequence. represents the average of the squared deviations of the sentiment values ​​from the mean, The average of the differences between all sentiment values ​​below the average and the average value in the segment, is the adjustment coefficient.

[0065] Furthermore, in the calculation structure of the emotion change value, the first half continues the variance term structure consistent with the score fluctuation, which is used to measure the overall volatility of the emotion value, while the second half constructs an asymmetric detection structure for negative emotion changes. This design is based on the high volatility and psychological bias characteristics of emotion indicators in rehabilitation assessment scenarios. In practice, patients' negative emotions (such as anxiety, fatigue, and fear) usually have a greater impact on path adjustment. Therefore, by weighted amplification of the part below the average, the model's sensitivity to negative deviations can be improved. Compared with the common mean filtering or bilateral smoothing methods in existing technologies, this structure can introduce directional intervention logic for psychological data without destroying the overall variance structure, thereby better meeting the actual clinical emotion evaluation needs and enhancing the system's ability to detect stage-by-stage emotional degradation.

[0066] It should be added that, based on the score fluctuation vector and the emotion change vector, constructing a set of fluctuation feature vectors means combining the score fluctuation values ​​and the corresponding emotion fluctuation values ​​in each time period in chronological order, and constructing a two-dimensional feature vector for each time period. ,in Indicates the The score fluctuation value of the segment, Indicates the The emotional fluctuation value of the segment; arrange the two-dimensional feature vectors corresponding to all time periods in the order of time periods to form a set of fluctuation feature vectors ,in Indicates the total number of time periods.

[0067] S30: Window adjustment processing is performed on the fluctuation feature vector set, a window range is set according to the score fluctuation vector and the emotion change vector, a rhythm factor and a training factor are calculated within each window range, and a rhythm feature sequence is obtained based on the rhythm factor and the training factor;

[0068] In this embodiment, performing window adjustment processing on the fluctuation feature vector set refers to using a fixed length or dynamic adjustment mechanism to perform sliding window division on the vector sequence according to a preset sliding window step size. The window corresponding to the sliding window is ,in and are the feature vectors corresponding to the start and end time periods of the window respectively.

[0069] Methods for setting the window range based on the rating fluctuation vector and the sentiment change vector include:

[0070] The score fluctuation vector and the emotion change vector are respectively indexed into sequence pairs according to the time index, and the fluctuation feature vector set is constructed based on the maximum difference between the score fluctuation value and the emotion fluctuation value in several adjacent pairs of sequences;

[0071] Perform sliding window statistical processing on the set of fluctuation feature vectors, and calculate the sum of the change gradients of the score fluctuation value and the sentiment change value in each window as the fluctuation density indicator;

[0072] Compare the volatility concentration index with a preset volatility threshold. When the volatility concentration index is greater than the preset volatility threshold, reduce the sliding window length. When the volatility concentration index is less than or equal to the preset volatility threshold, expand the sliding window length.

[0073] The adjusted sliding window length is used as the basis for time period division, and the time window segmentation processing is performed on the set of fluctuation feature vectors to set the window range.

[0074] In this embodiment, setting the window range based on the score fluctuation vector and the emotion change vector means constructing a time series pair of score fluctuation values ​​and emotion change values, calculating the maximum difference between these two types of fluctuation quantities in multiple adjacent time periods, and forming a set of fluctuation feature vectors based on this. The set reflects the coupled change trend of rehabilitation stability and emotional stability in each time period. On this basis, sliding window statistical processing is performed on the set of fluctuation feature vectors. By calculating the sum of the change gradients of the score fluctuation value and the emotion change value in each window, the degree of state change in the window is quantified and a fluctuation density index is constructed. This index essentially reflects the intensity of fluctuations in a local time period. Compared with the traditional fixed window method, this method introduces a mechanism for dynamically adjusting the window length with the intensity of fluctuations over time, and adaptively shortens or expands the sliding window length based on the comparison relationship between the fluctuation density index and the preset fluctuation amplitude threshold, thereby realizing a rhythm distribution modeling method with refined identification in high-fluctuation areas and compressed processing in low-fluctuation areas, thereby improving the state slicing accuracy and rhythm boundary detection capability before path planning.

[0075] The method for calculating the rhythm factor within each window range includes:

[0076] Set the number of time periods in the current window to s, and extract the score fluctuation value of each time period in the window and mood swings , construct the score fluctuation sequence and the emotion fluctuation sequence; perform rhythm synergy calculation based on these two types of sequences, and use the following rhythm factor calculation formula:

[0077] ;

[0078] in, Indicates the The rhythm factor corresponding to the window, Indicates the The score fluctuation value of each time period, Indicates the The emotional fluctuation value of a time period, A non-zero constant to set.

[0079] Furthermore, in the process of calculating the rhythm factor within each window range, the score fluctuation value and the emotion change value contained in the current window are respectively constructed into a one-to-one corresponding score fluctuation sequence and emotion fluctuation sequence, and a rhythm synergy calculation is performed on them. The calculation is based on the similarity of the numerical trends of the two sequences, and the normalized difference formula is used as the core calculation structure. The difference part measures the immediate synchronization between the score and the emotion, and the denominator balances the influence of the two dimensions through the square + root operation, so that it has a unified response capability to fluctuations of different scales. Different from the prior art that uses the Pearson correlation coefficient or the standard cosine similarity to model the rhythm synergy relationship, this embodiment constructs a more concise, real-time embeddable and segmented interpretable synergy measurement method, which is particularly suitable for nonlinear behavior trend modeling in the rehabilitation process. The numerical changes of the rhythm factor can clearly reflect the stage stability and rhythm deviation, providing strong support for subsequent path quality evaluation.

[0080] Methods for calculating the training factor within each window range include:

[0081] Set the number of time periods in the current window to w, and extract the score fluctuation value corresponding to each time period in the window and the difference between the scores before and after , construct the score fluctuation sequence and score change sequence; perform training response index calculation based on the score fluctuation value and score change amount, and use the following training factor calculation formula:

[0082] ;

[0083] in, Indicates the The training factor corresponding to the sliding window is Indicates the and The difference in stage scores between time periods, Indicates the The score fluctuation value of the time period, is a non-zero constant term that is set.

[0084] It should be noted that in the process of calculating the training factor within each window range, a training response evaluation structure is constructed for the score fluctuation value and the score value difference. The score change difference is used to capture the improvement or decline trend of the training status during the execution of the rehabilitation task, while the score fluctuation value reflects the stability of the training execution in this stage. After the two are superimposed and entered into the denominator, a nonlinear buffering effect for short-term drastic changes is established through square accumulation. The normalized logarithmic value range is constrained in the overall structure to give it a unified interpretability. Compared with the common fixed score difference threshold judgment or point-by-point dynamic evaluation method in the existing technology, this structure can integrate the dual characteristics of training trend and stage stability into an integrated indicator, and at the same time has a natural inhibitory ability for abnormally high differences, ensuring that the training factor has good robustness and generalization ability in path matching or stage reordering tasks.

[0085] Methods for obtaining rhythm feature sequences based on rhythm factors and training factors include:

[0086] Obtain a rhythm factor sequence and a training factor sequence, wherein the rhythm factor sequence is composed of multiple rhythm factors, and the training factor sequence is composed of multiple training factors, and the rhythm factor sequence and the training factor sequence correspond to each other in time sequence to form a dual-channel sequence pair set;

[0087] According to the rhythm factor sequence and the training factor sequence, the rhythm factor and the training factor corresponding to each time period are extracted, and the set of values ​​are combined into a two-dimensional feature pair to form a sliding window feature pair sequence;

[0088] Perform partition mapping processing on the sliding window feature pair sequence, determine the two-dimensional plane region index where the set of features is located based on the normalized values ​​of the rhythm factor and the training factor, and construct the original rhythm index sequence;

[0089] Perform smoothing and denoising on the original rhythm index sequence, adopt the consistent weight distribution of rhythm indexes in adjacent time periods, perform attenuation control on local high-frequency changes, and generate a smooth rhythm index sequence;

[0090] The rhythm levels are arranged in chronological order according to the smoothed rhythm index sequence to construct a rhythm feature sequence.

[0091] S40: Input the rhythm feature sequence into the whale optimization algorithm, perform path update processing, and output a stage sequence set;

[0092] The method of inputting the rhythm feature sequence into the whale optimization algorithm, performing path update processing, and outputting a stage sequence set includes:

[0093] Arrange the rhythm levels corresponding to each time period in the rhythm feature sequence in chronological order to form a one-dimensional vector structure with a length of 𝐿, and use it as the initial encoding input of the path individual;

[0094] Construct an initial population set, where each individual in the population encodes a path whose length is consistent with the rhythm feature sequence. Fixed start and end boundaries are used and the order of the intermediate stages is randomly initialized to generate an initial path candidate set.

[0095] The rhythm feature sequence is matched and evaluated with each candidate path individual. The fitness evaluation matrix is ​​constructed based on the accumulated value of the rhythm difference as the fitness index, and the optimal individual and the optimal fitness are recorded.

[0096] Call the whale optimization algorithm, introduce the adaptive convergence factor and rhythm disturbance adjustment operator based on the current optimal path individual, average rhythm difference and historical optimal stage combination, and perform path update processing;

[0097] All updated individual paths are re-evaluated, and the individual paths that converge to the minimum rhythm difference are extracted as the final stage sequence set, which is used as the output structure of the path planning result.

[0098] S50: performing a level classification process on each stage according to the stage sequence set to obtain a rehabilitation pathway matrix.

[0099] Methods for performing hierarchical classification processing on each stage according to the stage sequence set to obtain a rehabilitation pathway matrix include:

[0100] According to the stage sequence set, the consecutive stage numbers in each path are extracted and a stage index sequence is constructed. Each index sequence represents a set of candidate paths.

[0101] Perform feature collection processing on the stage nodes in each stage index sequence, extract the corresponding values ​​of the rhythm factor and training factor in the stage, and construct a set of stage indicator pairs;

[0102] Performing partition level mapping processing on the set of stage indicators, mapping them into rhythm level values ​​and training level values ​​respectively according to the preset level intervals corresponding to the rhythm factor and the training factor;

[0103] Combine the stage nodes in each path with their corresponding level values ​​to generate a path level sequence, and then arrange them in chronological order to construct a path level coding table;

[0104] All pathway level coding tables are integrated into a multi-row and multi-column matrix structure and output as a rehabilitation pathway matrix.

[0105] In this embodiment, the grading process is performed on each stage according to the stage sequence set. This means that the stage sequence after the completed path optimization is used as structured input, and the rhythm factor and training factor values ​​corresponding to each stage node are extracted one by one. These values ​​are combined to form a set of indicator pairs. Subsequently, each pair of indicator values ​​is mapped according to the pre-defined partition grade standard to obtain a rhythm grade value and a training grade value, respectively. The core principle of this process is to discretize the eigenvectors in the form of continuous numerical values ​​into grade labels, which facilitates the subsequent unified management and visual output in the matrix structure. At the same time, this dual-factor partition mapping method can more comprehensively reflect the rhythm stability and training execution quality of the rehabilitation process compared to the traditional single-indicator scoring method. Finally, the grades of each stage are combined with the corresponding time sequence one by one to form a path grade coding table. After integrating all path grades, the rehabilitation path matrix is ​​output. This matrix not only provides the grade distribution of each stage in the path, but also provides a unified reference structure for doctors or systems to interpret results and adjust personalized plans.

[0106] Example 2

[0107] See also Figure 2 As shown, based on the same inventive concept, this embodiment discloses a personalized burn rehabilitation path planning system. For details not provided in this embodiment, please refer to the description of the relevant parts in Example 1. The system includes:

[0108] Sequence construction module: used to obtain historical rehabilitation data and real-time assessment data, perform alignment processing on the historical rehabilitation data and real-time assessment data based on the time axis to obtain the state time series;

[0109] Vector construction module: used to perform stability calculation processing for each time period based on the state time series, extract the score fluctuation vector and sentiment change vector, and construct a fluctuation feature vector set based on the score fluctuation vector and sentiment change vector;

[0110] The first processing module is used to perform window adjustment processing on the fluctuation feature vector set, set the window range according to the score fluctuation vector and the emotion change vector, calculate the rhythm factor and training factor within each window range, and obtain the rhythm feature sequence based on the rhythm factor and training factor;

[0111] The second processing module is used to input the rhythm feature sequence into the whale optimization algorithm, perform path update processing, and output a stage sequence set;

[0112] Path planning module: used to perform hierarchical classification processing on each stage according to the stage sequence set to obtain the rehabilitation path matrix.

[0113] The detailed description set forth above in conjunction with the accompanying drawings describes examples and does not represent all examples that can be implemented or fall within the scope of the claims. The terms "example" and "exemplary" when used in this specification mean "used as an example, instance or illustration" and do not mean "better than or better than other examples."

[0114] Reference throughout this specification to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the present invention. Therefore, use of these phrases may refer to more than just one embodiment, and further, the described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0115] It should also be noted that these embodiments may be described as a process depicted as a flowchart, structure diagram, or block diagram, and that although the flowchart may describe the operations as a sequential process, many of the operations can be performed in parallel or concurrently, and the order of the operations may be rearranged.

Claims

1. A personalized burn rehabilitation pathway planning method, characterized by: include: Acquire historical rehabilitation data and real-time assessment data, perform alignment processing on the historical rehabilitation data and the real-time assessment data based on the time axis, and obtain a state time series; Perform stability calculation processing on each time period according to the state time series, extract the score fluctuation vector and the sentiment change vector, and construct a set of fluctuation feature vectors based on the score fluctuation vector and the sentiment change vector; Performing window adjustment processing on the set of fluctuation feature vectors, setting the window range according to the score fluctuation vector and the emotion change vector, calculating the rhythm factor and training factor within each window range, and obtaining the rhythm feature sequence based on the rhythm factor and training factor; Input the rhythm feature sequence into the whale optimization algorithm, perform path update processing, and output a set of stage sequences; The stages are graded according to the stage sequence set to obtain the rehabilitation pathway matrix.

2. The personalized burn rehabilitation path planning method according to claim 1, characterized in that: Historical rehabilitation data includes training item sequences and stage scores, and real-time assessment data includes muscle activity values ​​and emotion change values. Aligning historical rehabilitation data and real-time assessment data based on the time axis to obtain a state time series involves the following methods: According to the training item sequence and stage scores in the historical rehabilitation data, the timestamp corresponding to each training record is extracted to construct a training time index; Based on the muscle activity values ​​and emotion change values ​​in the real-time assessment data, the collection time corresponding to each assessment record is extracted to construct an assessment time index; Perform index alignment processing on the training time index and the evaluation time index at fixed time intervals, and combine the corresponding training items, stage scores, muscle activity values, and emotion change values ​​at the aligned time points to generate a joint time series data structure; Arrange the joint time series data structure in ascending time order to construct a complete state time series.

3. The personalized burn rehabilitation path planning method according to claim 2, characterized in that: The method of performing stability calculation processing for each time period based on the state time series and extracting the score fluctuation vector includes: Divide the state time series into continuous non-overlapping time periods according to a fixed time length T, and let the jth segment contain the score sequence ; Performs a modified variance stacking calculation on each scoring sequence.

4. The personalized burn rehabilitation path planning method according to claim 3, characterized in that: The method of performing stability calculation processing on each time period according to the state time series and extracting the emotion change vector includes: The state time series is divided into continuous non-overlapping time periods according to a fixed time length T, and the jth segment contains the emotion sequence ; Perform improved fluctuation calculation on each emotion sequence.

5. The personalized burn rehabilitation path planning method according to claim 4, characterized in that: Methods for setting the window range based on the rating fluctuation vector and the sentiment change vector include: The score fluctuation vector and the emotion change vector are respectively indexed into sequence pairs according to the time index, and the fluctuation feature vector set is constructed based on the maximum difference between the score fluctuation value and the emotion fluctuation value in several adjacent pairs of sequences; Perform sliding window statistical processing on the set of fluctuation feature vectors, and calculate the sum of the change gradients of the score fluctuation value and the sentiment change value in each window as the fluctuation density indicator; Compare the volatility concentration index with a preset volatility threshold. When the volatility concentration index is greater than the preset volatility threshold, reduce the sliding window length. When the volatility concentration index is less than or equal to the preset volatility threshold, expand the sliding window length. The adjusted sliding window length is used as the basis for time period division, and the time window segmentation processing is performed on the set of fluctuation feature vectors to set the window range.

6. The personalized burn rehabilitation path planning method according to claim 5, characterized in that: Methods for setting the window range based on the rating fluctuation vector and the sentiment change vector include: The score fluctuation vector and the emotion change vector are respectively indexed into sequence pairs according to the time index, and the fluctuation feature vector set is constructed based on the maximum difference between the score fluctuation value and the emotion fluctuation value in several adjacent pairs of sequences; Perform sliding window statistical processing on the set of fluctuation feature vectors, and calculate the sum of the change gradients of the score fluctuation value and the sentiment change value in each window as the fluctuation density indicator; Compare the volatility concentration index with a preset volatility threshold. When the volatility concentration index is greater than the preset volatility threshold, reduce the sliding window length. When the volatility concentration index is less than or equal to the preset volatility threshold, expand the sliding window length. The adjusted sliding window length is used as the basis for time period division, and the time window segmentation processing is performed on the set of fluctuation feature vectors to set the window range.

7. The personalized burn rehabilitation path planning method according to claim 6, characterized in that: The method for calculating the rhythm factor within each window range includes: Set the number of time periods in the current window to s, and extract the score fluctuation value of each time period in the window and mood swings , construct a score fluctuation sequence and an emotion fluctuation sequence; perform rhythm synergy calculation based on these two types of sequences.

8. The personalized burn rehabilitation path planning method according to claim 7, characterized in that: Methods for calculating the training factor within each window range include: Set the number of time periods in the current window to w, and extract the score fluctuation value corresponding to each time period in the window and the difference between the scores before and after , construct a score fluctuation sequence and a score change sequence; perform training response index calculation based on the score fluctuation value and the score change amount.

9. The personalized burn rehabilitation path planning method according to claim 8, characterized in that: Methods for obtaining rhythm feature sequences based on rhythm factors and training factors include: Obtain a rhythm factor sequence and a training factor sequence, wherein the rhythm factor sequence is composed of multiple rhythm factors, and the training factor sequence is composed of multiple training factors, and the rhythm factor sequence and the training factor sequence correspond to each other in time sequence to form a dual-channel sequence pair set; According to the rhythm factor sequence and the training factor sequence, the rhythm factor and the training factor corresponding to each time period are extracted, and the set of values ​​are combined into a two-dimensional feature pair to form a sliding window feature pair sequence; Perform partition mapping processing on the sliding window feature pair sequence, determine the two-dimensional plane region index where the set of features is located based on the normalized values ​​of the rhythm factor and the training factor, and construct the original rhythm index sequence; Perform smoothing and denoising on the original rhythm index sequence, adopt the consistent weight distribution of rhythm indexes in adjacent time periods, perform attenuation control on local high-frequency changes, and generate a smooth rhythm index sequence; The rhythm levels are arranged in chronological order according to the smoothed rhythm index sequence to construct a rhythm feature sequence.

10. A personalized burn rehabilitation path planning system, which is used to implement the personalized burn rehabilitation path planning method according to any one of claims 1 to 9, characterized in that: include: Sequence construction module: used to obtain historical rehabilitation data and real-time assessment data, perform alignment processing on the historical rehabilitation data and real-time assessment data based on the time axis to obtain the state time series; Vector construction module: used to perform stability calculation processing for each time period based on the state time series, extract the score fluctuation vector and sentiment change vector, and construct a fluctuation feature vector set based on the score fluctuation vector and sentiment change vector; The first processing module is used to perform window adjustment processing on the fluctuation feature vector set, set the window range according to the score fluctuation vector and the emotion change vector, calculate the rhythm factor and training factor within each window range, and obtain the rhythm feature sequence based on the rhythm factor and training factor; The second processing module is used to input the rhythm feature sequence into the whale optimization algorithm, perform path update processing, and output a stage sequence set; Path planning module: used to perform hierarchical classification processing on each stage according to the stage sequence set to obtain the rehabilitation path matrix.

Images