Method and system for assisting upper limb function rehabilitation after breast cancer operation

Abstract

The invention discloses a breast cancer postoperative upper limb function rehabilitation assisting method and system, and the method achieves the precise rehabilitation through multi-dimensional data collection and intelligent scheme generation: collecting the joint movement range, muscle strength index, lymphedema data, VAS pain score and Borg fatigue index of the upper limb of a patient, and carrying out the fusion to generate an upper limb function feature set; dividing the rehabilitation process into four stages, and generating a rehabilitation target corresponding to each stage by means of a preset mapping mechanism based on the feature set; generating a candidate rehabilitation scheme by applying a rehabilitation scheme generation model based on machine learning according to the staged rehabilitation target; through a prediction model based on multi-dimensional evaluation, rehabilitation prediction scores are generated in combination with the upper limb function feature set and the candidate rehabilitation schemes, and the candidate scheme with the highest score is selected as the optimal rehabilitation scheme. Personalized and dynamic management of upper limb rehabilitation after the breast cancer operation is achieved, and adaptability and effectiveness of a rehabilitation scheme are improved.

Description

Technical Field

[0001] This invention belongs to the field of rehabilitation assistive technology, and in particular relates to a method and system for assistive rehabilitation of upper limb function after breast cancer surgery. Background Technology

[0002] With the development of medical rehabilitation technology, postoperative upper limb functional rehabilitation techniques for breast cancer have emerged, focusing on promoting the recovery of patients' upper limb mobility after surgery. Currently, postoperative upper limb functional rehabilitation for breast cancer mainly relies on traditional rehabilitation program development methods. Traditional techniques typically involve doctors dividing rehabilitation stages (e.g., early, middle, and late stages) based on clinical experience, formulating standardized rehabilitation goals (e.g., joint range of motion indicators, muscle strength recovery goals) and fixed training programs (e.g., passive and active exercise combinations), with less emphasis on incorporating individual patient physiological data (e.g., age, body mass index) and real-time monitored rehabilitation indicators (e.g., changes in lymphedema, muscle fatigue status) for dynamic adjustments. However, current traditional rehabilitation program development methods have significant shortcomings: insufficient personalization, making it difficult to adapt to differences in physical condition and postoperative complications among patients; lack of dynamic analysis of real-time patient rehabilitation data (e.g., joint range of motion, changes in lymphedema volume), making it impossible to optimize training intensity and duration in a timely manner according to the rehabilitation progress; and insufficient quantitative assessment of key factors such as the risk of lymphedema worsening and muscle fatigue accumulation, which may lead to poor rehabilitation outcomes or secondary injuries, failing to meet the needs of precise and dynamic postoperative rehabilitation. Summary of the Invention

[0003] Therefore, it is necessary to provide a rehabilitation method and system for upper limb function after breast cancer surgery that can solve the above problems.

[0004] Firstly, this application provides an auxiliary method for upper limb functional rehabilitation after breast cancer surgery, including:

[0005] The patient's upper limb function data were collected and fused to generate an upper limb function feature set. The upper limb function data included joint range of motion, muscle strength indicators, lymphedema data, VAS pain score and Borg fatigue index.

[0006] The rehabilitation phase is divided into four stages: 1-7 days post-surgery, 8-14 days post-surgery, 15 days-3 months post-surgery, and 3-6 months post-surgery. Based on the upper limb functional feature set, a pre-defined mapping mechanism is used to generate corresponding phased rehabilitation goals.

[0007] Based on phased rehabilitation goals, a machine learning-based rehabilitation plan generation model is used to generate candidate rehabilitation plans.

[0008] Using a prediction model based on multidimensional assessment, a rehabilitation prediction score is generated by combining upper limb functional feature set and candidate rehabilitation programs, and the candidate rehabilitation program with the highest score is selected as the optimal rehabilitation program.

[0009] In one embodiment, based on phased rehabilitation goals, a machine learning-based rehabilitation plan generation model is used to generate candidate rehabilitation plans, including:

[0010] Obtain individual patient data, including age, body mass index, and postoperative complication level;

[0011] By using a motion type generation model, combined with phased rehabilitation goals and individual patient data, a set of candidate motion types that match the current rehabilitation stage is generated through feature matching.

[0012] Based on muscle strength indicators, lymphedema data, VAS pain scores, and Borg fatigue index from the upper limb functional feature set, baseline intensity parameters for each candidate movement type are determined through a training intensity prediction model. The baseline intensity parameters include unit movement resistance value and maximum safe repetition range.

[0013] Based on the phased rehabilitation goals, the baseline intensity parameters are mapped to training duration configurations using a preset intensity-duration conversion matrix. The duration configurations include the duration of a single training session and the training frequency.

[0014] A set of candidate rehabilitation plans is generated by dynamically combining candidate action type sets, associated baseline intensity parameters, and their mapped duration configurations.

[0015] In one embodiment, a rehabilitation prediction score is generated using a prediction model based on multi-dimensional assessment, combining an upper limb functional feature set and candidate rehabilitation programs, including:

[0016] Based on the movement type, intensity parameters, and training duration configuration of candidate rehabilitation programs, a predictive model is used in conjunction with the joint range of motion and muscle strength indicators of the upper limb functional feature set to generate a predicted value of functional recovery progress.

[0017] Based on the lymphedema data of the upper limb functional feature set, the historical trends of VAS pain scores and Borg fatigue index, and the training intensity parameters of candidate rehabilitation programs, the risk level of edema worsening and the cumulative risk value of muscle pain and fatigue were calculated.

[0018] By combining patient age, body mass index, and training frequency configuration of candidate rehabilitation programs in individual patient data, the feasibility and suitability of the training programs are evaluated.

[0019] The predicted value of functional recovery progress, the risk level of edema worsening, the cumulative risk value of muscle pain and fatigue, and the feasibility fit are normalized and weighted to generate a rehabilitation prediction score.

[0020] In one embodiment, the candidate rehabilitation plan with the highest score is selected as the optimal rehabilitation plan using the following formula:

[0021]

[0022] in:

[0023] R * This indicates the optimal rehabilitation plan. R represents the set of candidate rehabilitation protocols. p Indicates the risk of pain worsening, R b Let P(·) represent the cumulative fatigue level, and let P(·) represent the predicted functional recovery progress, calculated using an LSTM network.

[0024]

[0025] Where, θ ROM (t) represents the joint range of motion vector at time t, μ strength (t) represents the muscle strength index at time t, a k Scheme A k Includes action type encoding, s k Scheme A k Strength parameters;

[0026] R e (·) indicates the risk level of worsening edema, and the calculation satisfies:

[0027]

[0028] Where ∈(t) represents the current rate of change of lymphedema volume, ∈ base Indicates the safety threshold for the current recovery stage, ||s k ||2 represents scheme A k The baseline strength parameters are w1 and w2, which represent clinical risk weight coefficients, and σ represents the Sigmoid function.

[0029] A d (·) represents the executability fit, calculated using the following formula:

[0030]

[0031] Among them, f f Scheme A k The training frequency, BMI represents body mass index, age represents patient age, and k1, k2 and k3 represent individual adaptive adjustment coefficients;

[0032] α, β, γ, δ, and η represent dynamic weighting coefficients, and α, β, and γ satisfy:

[0033]

[0034] Where, when ∈(t)>∈threshold When (lymphedema threshold at each recovery stage), β = 1.0.

[0035] In one embodiment, the method further includes:

[0036] Calculate the dynamic deviation between the dynamic monitoring parameters of the upper limb functional feature set and the preset target parameters of the current stage rehabilitation program;

[0037] Based on the parameter deviation, the corresponding parameters of the rehabilitation plan generation model are adjusted through backpropagation algorithm using an online learning mechanism.

[0038] In one embodiment, the dynamic monitoring parameters include real-time angle values ​​of joint range of motion, instantaneous detection values ​​of muscle strength indicators, real-time volume change rate of lymphedema data, VAS pain score, and Borg fatigue index; the corresponding parameters of the rehabilitation program generation model are adjusted using an online learning mechanism and a backpropagation algorithm, including the priority weight matrix of the movement type generation model and the mapping coefficients of the intensity-duration conversion matrix.

[0039] In one embodiment, the method further includes:

[0040] The upper limb functional feature set is visualized by using a three-dimensional skeletal motion model to generate a multi-dimensional rehabilitation progress report that includes joint range of motion change curves, muscle strength recovery heat maps, edema volume comparison maps, pain trend curves, and fatigue heat maps.

[0041] The visualization refresh frequency is set according to the update cycle of the upper limb functional feature set, and the multi-dimensional rehabilitation progress report is overlaid and compared with the phased rehabilitation goals based on the refresh frequency to generate a visual comparison result of rehabilitation progress.

[0042] Secondly, this application also provides a post-mastectomy upper limb functional rehabilitation assistive system, comprising:

[0043] The data acquisition and fusion module is used to collect data on the range of motion of the patient's upper limb joints, muscle strength indicators, and lymphedema through a wearable sensor array, and then fuse these data to generate a set of upper limb functional features.

[0044] The phase goal generation module is used to divide the rehabilitation phase into four stages: 1-7 days post-surgery, 8-14 days post-surgery, 15 days-3 months post-surgery, and 3-6 months post-surgery. Based on the upper limb functional feature set, it generates corresponding phased rehabilitation goals using a preset mapping mechanism.

[0045] The rehabilitation plan generation module is used to generate candidate rehabilitation plans based on phased rehabilitation goals and using a machine learning-based rehabilitation plan generation model.

[0046] The evaluation and selection module is used to generate a rehabilitation prediction score by using a prediction model based on multi-dimensional assessment, combined with upper limb functional feature set and candidate rehabilitation programs, and select the candidate rehabilitation program with the highest score as the optimal rehabilitation program.

[0047] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0048] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0049] The aforementioned method and system for upper limb functional rehabilitation after breast cancer surgery, along with computer equipment and storage media, collects data on the patient's upper limb joint range of motion, muscle strength indicators, lymphedema data, VAS pain score, and Borg fatigue index, and integrates these data to generate a functional feature set. Combined with a phased rehabilitation goal generation mechanism (1-7 days post-surgery, 8-14 days, 15 days-3 months, and 3-6 months post-surgery), a machine learning model dynamically generates candidate rehabilitation plans. Through multi-dimensional evaluation and prediction models, the plans are optimized by comprehensively considering factors such as functional recovery progress, edema risk, muscle fatigue, and individual adaptability. Based on individual data, precise division of rehabilitation stages and dynamic adjustment of goals are achieved. Machine learning enhances the personalized adaptability of rehabilitation plans, and a multi-dimensional quantitative evaluation mechanism effectively controls the risk of lymphedema and the accumulation of muscle fatigue. This addresses the problems of insufficient personalization, lack of dynamic optimization capabilities, and incomplete risk assessment in traditional plans, enabling precise and dynamic management of upper limb rehabilitation after breast cancer surgery and significantly improving the adaptability and effectiveness of rehabilitation plans. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 This is a flowchart of an auxiliary method for upper limb functional rehabilitation after breast cancer surgery according to the present invention;

[0052] Figure 2 This is a structural diagram of an upper limb functional rehabilitation assistive system after breast cancer surgery according to the present invention. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0054] The implementation environment of this invention includes a sensor array, a computing terminal, and a server. During the postoperative upper limb function rehabilitation of breast cancer patients, when it is necessary to formulate a dynamic rehabilitation plan based on the individual rehabilitation status, the wearable sensor array collects data and transmits it to the computing terminal or server. The machine learning model generates candidate rehabilitation plans, and the plans are optimized by a multi-dimensional evaluation model and then transmitted to the display terminal to realize the personalized generation of rehabilitation plans.

[0055] In one embodiment, such as Figure 1 As shown, a method for assisting upper limb function rehabilitation after breast cancer surgery is provided. This embodiment illustrates the application of this method to a computing terminal. It is understood that this method can also be applied to a server, or to a system including both a computing terminal and a server, and implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0056] S01. Collect the patient's upper limb function data and fuse them to generate an upper limb function feature set. The upper limb function data includes joint range of motion, muscle strength indicators, lymphedema data, VAS pain score and Borg fatigue index.

[0057] Specifically, wearable sensor arrays (such as wearable devices integrating inertial measurement units, electromyography sensors, and volume measurement modules) can be used to collect data on the patient's upper limb joint range of motion (flexion, extension, and rotation range of shoulder / elbow / wrist joints), muscle contraction force electrical signals (muscle strength indicators), lymphedema volume change rate (e.g., obtained through infrared light scattering or pressure sensing), and VAS pain scores. The Borg fatigue index can also be obtained through a scale. Multi-source heterogeneous data (time series, physiological signals, spatial motion data, VAS pain scores, and Borg fatigue index) are spatiotemporally aligned, denoised, filtered, and feature-extracted, and fused to generate an upper limb functional feature set containing joint range of motion vectors, muscle strength feature matrices, edema status parameters, and pain and fatigue parameters. This provides a quantitative data foundation for the subsequent generation of phased rehabilitation goals and optimization of programs.

[0058] S02 divides the rehabilitation stage into four phases: 1-7 days post-surgery, 8-14 days post-surgery, 15 days-3 months post-surgery, and 3-6 months post-surgery. Based on the upper limb functional feature set, a pre-defined mapping mechanism is used to generate corresponding phased rehabilitation goals.

[0059] Based on the postoperative tissue repair physiological cycle and clinical rehabilitation guidelines, the rehabilitation process can be divided into four stages with clear time boundaries: 1-7 days postoperatively (acute phase), 8-14 days postoperatively (subacute phase), 15 days to 3 months postoperatively (recovery phase), and 3-6 months postoperatively (consolidation phase). Using an upper limb functional feature set (including joint range of motion vectors, muscle strength index matrix, edema parameters, and pain and fatigue parameters), a pre-defined clinical parameter mapping mechanism (such as an evidence-based stage target threshold table or a machine learning mapping model) maps real-time collected functional data to quantitative rehabilitation goals for each stage. For example, the goal for 1-7 days postoperatively is to control the rate of change in lymphedema volume to ≤ a safe threshold and maintain passive joint range of motion ≥ 60% of the baseline value; the goal for 3-6 months is to achieve active joint range of motion of ≥ 90% on the healthy side and muscle strength recovery to 80% of normal levels. By establishing a dynamic mapping relationship between functional features and stage goals, personalized and precise setting of rehabilitation goals can be achieved.

[0060] S03. Based on the phased rehabilitation goals, a machine learning-based rehabilitation plan generation model is used to generate candidate rehabilitation plans.

[0061] Based on phased rehabilitation goals, candidate rehabilitation plans can be generated using machine learning models through the following steps: First, individual patient data is acquired. Then, using a movement type generation model, the phased rehabilitation goals are matched with the individual data to generate a set of candidate movement types suitable for the current phase (e.g., passive movement exercises for days 1-7 post-surgery). Second, based on the upper limb functional feature set, the baseline intensity parameters for each candidate movement are calculated using a training intensity prediction model. Third, according to the phased rehabilitation goals, a pre-defined transformation matrix is ​​used to map the baseline intensity parameters to duration configurations such as single training session duration and training frequency. Fourth, candidate movement types, baseline intensity parameters, and duration configurations are dynamically combined to generate multiple sets of candidate rehabilitation plans. This approach achieves personalized generation of movement types and intensity parameters through machine learning models and quantitative configuration of plans by combining them with a clinical transformation matrix, ensuring the phased adaptability and individual relevance of candidate plans.

[0062] S04. Using a prediction model based on multi-dimensional assessment, combined with upper limb functional feature set and candidate rehabilitation programs, a rehabilitation prediction score is generated, and the candidate rehabilitation program with the highest score is selected as the optimal rehabilitation program.

[0063] Specifically, a predictive model based on multi-dimensional assessment is used to configure the movement type, intensity parameters, and duration of candidate rehabilitation programs. This, combined with joint range of motion vectors and muscle strength indices from the upper limb functional feature set, generates a predicted value for functional recovery progress. Based on the historical rate of change of lymphedema data and the baseline intensity parameters of the program, the risk level of edema worsening can be calculated, simultaneously assessing the cumulative risk of muscle pain and fatigue. Combining patient age, body mass index, and program training frequency, the feasibility fit is calculated, and a weighted summation is used to generate a comprehensive score. The candidate program with the highest score is selected, achieving a balance between functional recovery efficiency, risk control, and individual adaptability.

[0064] The aforementioned method for assisting upper limb functional rehabilitation after breast cancer surgery collects data on the patient's upper limb joint range of motion, muscle strength indicators, lymphedema data, VAS pain score, and Borg fatigue index, fusing these data to generate an upper limb functional feature set. This allows for precise quantification of the patient's individual physiological state, addressing the problem of insufficient personalized data collection in traditional methods. The rehabilitation process is scientifically divided into four stages: 1-7 days post-surgery, 8-14 days, 15 days-3 months, and 3-6 months post-surgery. Based on the upper limb functional feature set, a dynamic phased rehabilitation goal is generated through a pre-defined mapping mechanism, overcoming the limitations of traditional fixed-stage divisions and static goal setting, and enabling dynamic management of the rehabilitation process. A machine learning-based rehabilitation program generation model is used to generate candidate rehabilitation programs by combining phased goals and the functional feature set. A multi-dimensional assessment-based predictive model is used for quantitative scoring and selection of the optimal program, addressing the problems of incomplete risk assessment and inability to quantify and control the risk of secondary injury in traditional techniques. Through the organic combination of these technical steps, personalized and dynamic management of upper limb rehabilitation after breast cancer surgery is achieved, improving the adaptability and effectiveness of rehabilitation programs and solving the technical problems of insufficient personalization, lack of dynamic optimization capabilities, and incomplete risk assessment in traditional rehabilitation programs.

[0065] In one embodiment, based on phased rehabilitation goals, a machine learning-based rehabilitation plan generation model is used to generate candidate rehabilitation plans, including:

[0066] S11, Obtain individual patient data, including age, body mass index, and postoperative complication level;

[0067] S12, using the action type generation model, combined with phased rehabilitation goals and individual patient data, generates a set of candidate action types that match the current rehabilitation stage through feature matching;

[0068] S13, based on muscle strength indicators, lymphedema data, VAS pain scores and Borg fatigue index in the upper limb functional feature set, determines the baseline intensity parameters for each candidate movement type through a training intensity prediction model. The baseline intensity parameters include the unit movement resistance value and the maximum safe repetition range.

[0069] S14. Based on the phased rehabilitation goals, the baseline intensity parameters are mapped to training duration configurations using a preset intensity-duration conversion matrix. The duration configurations include the duration of a single training session and the training frequency.

[0070] S15 generates a set of candidate rehabilitation plans by dynamically combining the candidate action type set, the associated baseline intensity parameters and their mapped duration configuration.

[0071] Specifically, individual data such as patient age, body mass index, and postoperative complication levels are acquired and used as a basis. A movement type generation model (trained based on decision trees and clustering algorithms) is then used to match phased rehabilitation goals with individual data, generating a set of candidate movement types suitable for the current rehabilitation stage, such as generating passive movement types for the postoperative acute phase. Based on muscle strength indicators, lymphedema data, VAS pain scores, and Borg fatigue index from the upper limb functional feature set, a training intensity prediction model determines the baseline intensity parameters for each candidate movement, including but not limited to unit movement resistance value (determined by muscle strength indicators) and the maximum safe repetition range (determined by Borg fatigue index), ensuring that the intensity setting conforms to the patient's current physiological state (lymphedema data limits training intensity, and VAS pain scores dynamically adjust the intensity threshold). According to the phased rehabilitation goals, a pre-set intensity-duration conversion matrix is ​​used to map the baseline intensity parameters to a duration configuration including the duration of a single training session and training frequency, achieving a quantitative conversion of intensity parameters to the training time dimension. By dynamically combining candidate action type sets, corresponding baseline intensity parameters, and their mapped duration configurations, multiple sets of candidate rehabilitation plans are generated. With the help of the adaptive capabilities of machine learning models and clinical parameter mapping mechanisms, personalized rehabilitation plans are generated, effectively adapting to individual patient differences and postoperative stage needs, and solving the problems of insufficient personalization and poor stage adaptability in traditional plans.

[0072] In one embodiment, a rehabilitation prediction score is generated using a prediction model based on multi-dimensional assessment, combining an upper limb functional feature set and candidate rehabilitation programs, including:

[0073] S21, based on the movement type, intensity parameters and training duration configuration of the candidate rehabilitation program, uses a prediction model combined with the muscle strength index and joint range of motion of the upper limb functional feature set to generate a predicted value of functional recovery progress.

[0074] S22. Based on the lymphedema data of the upper limb functional feature set, the historical trend of VAS pain score and Borg fatigue index, and the training intensity parameters of candidate rehabilitation programs, calculate the risk level of edema worsening and the cumulative risk value of muscle pain and fatigue.

[0075] S23. Combine patient age, body mass index and training frequency configuration of candidate rehabilitation programs in individual patient data to evaluate the feasibility and suitability of the training program.

[0076] S24 normalizes and weights the predicted value of functional recovery progress, the risk level of edema worsening, the cumulative risk value of muscle pain and fatigue, and the feasibility fit to generate a rehabilitation prediction score.

[0077] For example, based on the movement types, intensity parameters, and training duration configurations of candidate rehabilitation programs, combined with muscle strength indicators and joint range of motion data from the upper limb functional feature set, a predictive model generates predicted values ​​for functional recovery progress, quantifying the expected effect of the program on the patient's functional recovery. Based on the historical trends of lymphedema data, VAS pain scores, and Borg fatigue indices from the upper limb functional feature set, and the training intensity parameters of the candidate programs, the risk level of edema worsening and the cumulative risk value of muscle fatigue are calculated. By analyzing the correlation between the rate of change in edema volume, training-induced pain and fatigue, and intensity parameters, the degree of pain and fatigue that training may induce, and the risk of lymphedema worsening, are quantified. Combining individual patient physiological conditions such as age and body mass index with the training frequency configuration of the candidate programs, the feasibility and suitability of the training program are assessed, determining its adaptability in actual implementation. Normalized and weighted processing is performed on the predicted value of functional recovery progress, the risk level of edema worsening, the cumulative risk value of muscle fatigue and strain, and the feasibility fit. The multi-dimensional assessment indicators are transformed into a unified quantitative rehabilitation prediction score. By comprehensively considering the efficiency of functional recovery, risk control and individual adaptability, a quantitative basis is provided for the selection of candidate solutions. This effectively solves the problems of incomplete risk assessment and inability to dynamically optimize solutions in traditional techniques, and improves the safety and effectiveness of rehabilitation programs.

[0078] In one embodiment, S31, the candidate rehabilitation plan with the highest score is selected as the optimal rehabilitation plan using the following formula:

[0079]

[0080] in:

[0081] R * This indicates the optimal rehabilitation plan. R represents the set of candidate rehabilitation protocols. p Indicates the risk of pain worsening, R b Let P(·) represent the cumulative fatigue level, and let P(·) represent the predicted functional recovery progress, calculated using an LSTM network.

[0082]

[0083] Where, θ ROM(t) represents the joint range of motion vector at time t, μ strength (t) represents the muscle strength index at time t, a k Scheme A k Includes action type encoding, s k Scheme A k Strength parameters;

[0084] R e (·) indicates the risk level of worsening edema, and the calculation satisfies:

[0085]

[0086] Where ∈(t) represents the current rate of change of lymphedema volume, ∈ base Indicates the safety threshold for the current recovery stage, ||s k ||2 represents scheme A k The baseline strength parameters are w1 and w2, which represent clinical risk weight coefficients, and σ represents the Sigmoid function.

[0087] A d (·) represents the executability fit, calculated using the following formula:

[0088]

[0089] Among them, f f Scheme A k The training frequency, BMI represents body mass index, age represents patient age, and k1, k2 and k3 represent individual adaptive adjustment coefficients;

[0090] α, β, γ, δ, and η represent dynamic weighting coefficients, and α, β, and γ satisfy:

[0091]

[0092] Where, when ∈(t)>∈ threshold When (lymphedema threshold at each recovery stage), β = 1.0.

[0093] Specifically, from the set of candidate rehabilitation programs In this process, the optimal rehabilitation plan R is determined by maximizing the objective function. * Taking into account factors such as functional recovery progress, risk of edema worsening, feasibility and fit, risk of pain worsening, and fatigue accumulation level, a multi-dimensional balance is achieved through dynamic weighting coefficients. Risk of pain worsening R p Based on VAS score calculation, the likelihood of pain worsening is predicted; cumulative fatigue level R. bThe Borg index is used to assess the cumulative fatigue caused by training; the predicted functional recovery progress value P(·) is calculated using an LSTM network, with inputs including: the joint range of motion vector at time t (e.g., shoulder / elbow / wrist flexion / extension and rotation angles) θ. ROM (t), muscle strength indicators (such as electromyographic signal intensity) at time t μ strength (t), Scheme A k Includes action type encoding (e.g., passive activity, active training) a k Option A k Strength parameters (such as resistance value, number of repetitions) s k By learning from historical time-series data, the quantitative effect of the treatment on functional recovery is predicted, with an output range of [0,10]; the risk level of edema worsening is R. e (·) Risk is quantified based on the Sigmoid function σ, where ∈(t) represents the current rate of change in lymphedema volume, ∈ base This indicates the safety threshold for the current recovery stage (such as the edema control standard 1-7 days post-surgery), ||s k ||2 represents scheme A k The baseline intensity parameter (reflecting the overall training intensity value), w1 and w2 represent clinical risk weighting coefficients (set by medical guidelines to balance the influence of the absolute value of edema and the rate of change), and the risk value is mapped to [0,1] through the Sigmoid function σ; executability fit A d (·) The suitability of the scheme to individual patients is evaluated using logistic functions, f f Scheme A k The training frequency (e.g., number of training sessions per week), BMI (Body Mass Index, reflecting the patient's body type's adaptability to training), age (patient's age), and k1, k2, and k3 (individual adaptation adjustment coefficients, e.g., older patients have lower tolerance for high-frequency training) are used. A higher fit value indicates that the program is more suitable for the patient's physiological conditions and more feasible to implement. Dynamic weighting coefficients α, β, and γ are dynamically assigned weights based on the relative importance of joint range of motion and muscle strength. For example, early rehabilitation may focus more on restoring joint range of motion. When ∈(t)>∈ threshold When the lymphedema threshold for each recovery stage is β = 1.0, the higher the risk of lymphedema, the closer β is to 1, which increases the proportion of the risk term in the objective function and prioritizes controlling the deterioration of edema. When patients are over 40 years old, γ decreases with age, reducing the weight of high-intensity training to better suit the tolerance of elderly patients. The weight δ is determined by the level of postoperative complications; the more severe the postoperative complications, the higher the weight of pain risk. η is determined based on the rehabilitation stage, with increased weight for fatigue control in the later stages of rehabilitation (stages 3-4). Through the above formulas, this approach transforms functional recovery, edema risk, pain and fatigue risk, and individual suitability into a unified scoring system, avoiding the subjectivity of traditional approaches that rely on experience-based judgment. It adaptively adjusts the assessment focus based on the patient's real-time physiological data (such as changes in edema and muscle strength) and individual characteristics (age, BMI), enhancing the personalization of the approach. When lymphedema exceeds a safe threshold, the risk weight is automatically increased to prioritize rehabilitation safety and avoid secondary injury. Based on the LSTM's ability to learn from historical data, it accurately predicts the effectiveness of the approach, solving the problem of poor suitability caused by the one-size-fits-all approach of traditional approaches.

[0094] In one embodiment, the method further includes:

[0095] S41, Calculate the dynamic deviation between the dynamic monitoring parameters of the upper limb functional feature set and the preset target parameters of the current stage rehabilitation program;

[0096] S42, based on the parameter deviation, uses an online learning mechanism to adjust the corresponding parameters of the rehabilitation plan generation model through a backpropagation algorithm.

[0097] Specifically, key physiological parameters of the current rehabilitation stage are collected in real time and compared with the target parameters preset in the current rehabilitation plan to generate deviations (such as joint range of motion deviation, Euclidean distance between the actual joint angle and the target angle). The weighted sum of squares of these deviations is used as the optimization objective to construct a loss function. Based on the gradient direction of the loss function with respect to the weight matrix, the priority weights of the model are adjusted with a preset learning rate. At the same time, an upper limit can be set for the parameter update magnitude (such as a single adjustment not exceeding 20% ​​of the original value). Through a real-time deviation backpropagation adjustment mechanism with multi-parameter fusion, dynamic closed-loop optimization of the rehabilitation plan is achieved.

[0098] In one embodiment, S51, the dynamic monitoring parameters include real-time angle values ​​of joint range of motion, instantaneous detection values ​​of muscle strength indicators, real-time volume change rate of lymphedema data, VAS pain score, and Borg fatigue index; the corresponding parameters of the rehabilitation program generation model are adjusted using an online learning mechanism through backpropagation algorithm, including the priority weight matrix of the movement type generation model and the mapping coefficients of the intensity-duration conversion matrix.

[0099] For example, the deviation between the dynamic monitoring parameters in the upper limb functional feature set (including real-time angle values ​​of joint range of motion, instantaneous detection values ​​of muscle strength indicators, real-time volume change rate of lymphedema data, VAS pain score, and Borg fatigue index) and the preset target parameters of the current stage rehabilitation program is calculated. This deviation reflects the difference between the patient's actual rehabilitation progress and the expected goal. Based on this deviation, the corresponding parameters of the rehabilitation program generation model are adjusted using an online learning mechanism and a backpropagation algorithm. Specifically, this includes adjusting the priority weight matrix of the movement type generation model (e.g., dynamically adjusting the recommended weights of passive / active movements based on the patient's muscle strength recovery) and optimizing the mapping coefficients of the intensity-duration conversion matrix (e.g., adjusting the conversion relationship between training intensity and duration based on the lymphedema change rate). By leveraging real-time data feedback and dynamic updates to model parameters, the rehabilitation program achieves adaptive optimization: continuously refining the matching degree between the program generation model and the patient's individual characteristics, thus solving the problem that traditional programs cannot be dynamically adjusted with the rehabilitation process; through iterative learning of the backpropagation algorithm, the model gradually optimizes the action type recommendation strategy and intensity configuration logic, improving the program's responsiveness to changes in the patient's physiological state, avoiding low rehabilitation efficiency or secondary injury risks caused by static programs, and achieving precise and dynamic postoperative rehabilitation management.

[0100] In one embodiment, the method further includes:

[0101] S61 uses a three-dimensional skeletal motion model to visualize the upper limb functional feature set, generating a multi-dimensional rehabilitation progress report that includes joint range of motion change curves, muscle strength recovery heat maps, edema volume comparison maps, pain trend curves, and fatigue heat maps.

[0102] S62 sets the visualization refresh frequency according to the update cycle of the upper limb functional feature set, and overlays and compares the multi-dimensional rehabilitation progress report with the phased rehabilitation goals based on the refresh frequency to generate a visual comparison result of rehabilitation progress.

[0103] Specifically, a three-dimensional skeletal motion model is used to visualize the upper limb functional feature set. Joint range of motion, muscle strength indicators, and lymphedema data are transformed into intuitive visual elements, generating a multi-dimensional rehabilitation progress report that includes curves showing changes in joint range of motion, a heatmap of muscle strength recovery, a comparison chart of edema volume, a pain trend curve, and a fatigue heatmap. This report presents the patient's rehabilitation status graphically. The visualization refresh frequency is set according to the update cycle of the upper limb functional feature set, synchronizing the latest monitoring data in real time. The multi-dimensional rehabilitation progress report is overlaid and compared with the phased rehabilitation goals, generating a visual comparison of rehabilitation progress. This allows the deviation between the patient's rehabilitation progress and the preset goals to be presented visually. This mechanism uses 3D visualization technology and dynamic comparison display to achieve intuitive presentation and real-time tracking of rehabilitation data: multi-dimensional charts integrate joint activity, muscle strength, edema, pain and fatigue trend data, solving the problem of insufficient quantitative display of rehabilitation status in traditional programs and providing doctors with accurate assessment basis; the dynamic comparison function based on refresh frequency makes the difference between rehabilitation progress and stage goals quantifiable and visualized, assisting in the dynamic adjustment of clinical decisions, while improving patients' awareness and compliance with the rehabilitation process, and realizing a visualized closed loop of personalized rehabilitation management.

[0104] The aforementioned method for assisting upper limb functional rehabilitation after breast cancer surgery utilizes a wearable sensor array to collect real-time data on the patient's upper limb joint range of motion, muscle strength indicators, lymphedema, VAS pain score, and Borg fatigue index, fusing these data to generate a functional feature set. This allows for precise quantification of the patient's individual physiological state. The rehabilitation process is scientifically divided into four stages: 1-7 days post-surgery, 8-14 days, 15 days-3 months, and 3-6 months post-surgery. Based on the functional feature set, a preset mapping mechanism generates dynamic, phased rehabilitation goals, enabling dynamic management of the rehabilitation process. Furthermore, a machine learning-based rehabilitation program generation model is employed, combining phased goals with individual patient characteristics. Data and functional feature sets generate candidate rehabilitation plans, improving the adaptability of the plans to individual differences and postoperative complications. A predictive model based on multi-dimensional assessment is used to quantitatively score and select the optimal plan, comprehensively considering factors such as predicted functional recovery progress, risk of worsening edema, risk of cumulative muscle pain and fatigue, and individual feasibility, thus controlling the risk of secondary injury. By calculating the deviation between dynamic monitoring parameters and preset target parameters, an online learning mechanism is used to adjust model parameters, achieving adaptive optimization of the rehabilitation plan. A multi-dimensional rehabilitation progress report is generated using a three-dimensional skeletal motion model and overlaid with and compared with staged goals, enabling intuitive presentation and real-time tracking of rehabilitation data. The organic combination of these technologies enables personalized and dynamic management of upper limb rehabilitation after breast cancer surgery, significantly improving the adaptability and effectiveness of rehabilitation plans and solving the technical problems of insufficient personalization, lack of dynamic optimization capabilities, and incomplete risk assessment in traditional rehabilitation plans.

[0105] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0106] Based on the same inventive concept, this application also provides a post-mastectomy upper limb function rehabilitation assistive system for implementing the above-described post-mastectomy upper limb function rehabilitation assistive method. The solution provided by this system is similar to the implementation scheme described in the above method. Therefore, the specific limitations of one or more embodiments of the post-mastectomy upper limb function rehabilitation assistive system provided below can be found in the above-described limitations of the post-mastectomy upper limb function rehabilitation assistive method, and will not be repeated here.

[0107] In one exemplary embodiment, such as Figure 2 As shown, a post-mastectomy upper limb functional rehabilitation assistive system is provided, comprising:

[0108] The data acquisition and fusion module 101 is used to collect the patient's upper limb function data and fuse it to generate an upper limb function feature set. The upper limb function data includes joint range of motion, muscle strength index, lymphedema data, VAS pain score and Borg fatigue index.

[0109] The phase goal generation module 102 is used to divide the rehabilitation phase into four stages: 1-7 days post-surgery, 8-14 days post-surgery, 15 days-3 months post-surgery, and 3-6 months post-surgery. Based on the upper limb functional feature set, it generates corresponding phased rehabilitation goals using a preset mapping mechanism.

[0110] The rehabilitation plan generation module 103 is used to generate candidate rehabilitation plans based on phased rehabilitation goals and using a machine learning-based rehabilitation plan generation model.

[0111] The evaluation and selection module 104 is used to generate a rehabilitation prediction score by using a prediction model based on multi-dimensional evaluation, combining the upper limb functional feature set and candidate rehabilitation programs, and select the candidate rehabilitation program with the highest score as the optimal rehabilitation program.

[0112] In one embodiment, the rehabilitation plan generation module 103 is further configured to:

[0113] Obtain individual patient data, including age, body mass index, and postoperative complication level;

[0114] By using a motion type generation model, combined with phased rehabilitation goals and individual patient data, a set of candidate motion types that match the current rehabilitation stage is generated through feature matching.

[0115] Based on muscle strength indicators, lymphedema data, VAS pain scores, and Borg fatigue index from the upper limb functional feature set, baseline intensity parameters for each candidate movement type are determined through a training intensity prediction model. The baseline intensity parameters include unit movement resistance value and maximum safe repetition range.

[0116] Based on the phased rehabilitation goals, the baseline intensity parameters are mapped to training duration configurations using a preset intensity-duration conversion matrix. The duration configurations include the duration of a single training session and the training frequency.

[0117] A set of candidate rehabilitation plans is generated by dynamically combining candidate action type sets, associated baseline intensity parameters, and their mapped duration configurations.

[0118] In one embodiment, the evaluation and selection module 104 is further configured to:

[0119] Based on the movement type, intensity parameters, and training duration configuration of candidate rehabilitation programs, a predictive model is used in conjunction with the joint range of motion and muscle strength indicators of the upper limb functional feature set to generate a predicted value of functional recovery progress.

[0120] Based on the lymphedema data of the upper limb functional feature set, the historical trends of VAS pain scores and Borg fatigue index, and the training intensity parameters of candidate rehabilitation programs, the risk level of edema worsening and the cumulative risk value of muscle pain and fatigue were calculated.

[0121] By combining patient age, body mass index, and training frequency configuration of candidate rehabilitation programs in individual patient data, the feasibility and suitability of the training programs are evaluated.

[0122] The predicted value of functional recovery progress, the risk level of edema worsening, the cumulative risk value of muscle pain and fatigue, and the feasibility fit are normalized and weighted to generate a rehabilitation prediction score.

[0123] In one embodiment, the evaluation and selection module 104 is further configured to select the candidate rehabilitation plan with the highest score as the optimal rehabilitation plan using the following formula:

[0124]

[0125] in:

[0126] R *This indicates the optimal rehabilitation plan. R represents the set of candidate rehabilitation protocols. p Indicates the risk of pain worsening, R b Let P(·) represent the cumulative fatigue level, and let P(·) represent the predicted functional recovery progress, calculated using an LSTM network.

[0127]

[0128] Where, θ ROM (t) represents the joint range of motion vector at time t, μ strength (t) represents the muscle strength index at time t, a k Scheme A k Includes action type encoding, s k Scheme A k Strength parameters;

[0129] R e (·) indicates the risk level of worsening edema, and the calculation satisfies:

[0130]

[0131] Where ∈(t) represents the current rate of change of lymphedema volume, ∈ base Indicates the safety threshold for the current recovery stage, ||s k ||2 represents scheme A k The baseline strength parameters are w1 and w2, which represent clinical risk weight coefficients, and σ represents the Sigmoid function.

[0132] A d (·) represents the executability fit, calculated using the following formula:

[0133]

[0134] Among them, f f Scheme A k The training frequency, BMI represents body mass index, age represents patient age, and k1, k2 and k3 represent individual adaptive adjustment coefficients;

[0135] α, β, γ, δ, and η represent dynamic weighting coefficients, and α, β, and γ satisfy:

[0136]

[0137] Where, when ∈(t)>∈ threshold When (lymphedema threshold at each recovery stage), β = 1.0.

[0138] In one embodiment, the system further includes a synchronization optimization module for:

[0139] Calculate the dynamic deviation between the dynamic monitoring parameters of the upper limb functional feature set and the preset target parameters of the current stage rehabilitation program;

[0140] Based on the parameter deviation, the corresponding parameters of the rehabilitation plan generation model are adjusted through backpropagation algorithm using an online learning mechanism.

[0141] In one embodiment, the synchronous optimization module dynamically monitors parameters including real-time angle values ​​of joint range of motion, instantaneous detection values ​​of muscle strength indicators, real-time volume change rate of lymphedema data, VAS pain score, and Borg fatigue index; and uses an online learning mechanism to adjust the corresponding parameters of the rehabilitation program generation model through backpropagation algorithm, including the priority weight matrix of the movement type generation model and the mapping coefficients of the intensity-duration conversion matrix.

[0142] In one embodiment, the system further includes a visualization module for:

[0143] The upper limb functional feature set is visualized by using a three-dimensional skeletal motion model to generate a multi-dimensional rehabilitation progress report that includes joint range of motion change curves, muscle strength recovery heat maps, edema volume comparison maps, pain trend curves, and fatigue heat maps.

[0144] The visualization refresh frequency is set according to the update cycle of the upper limb functional feature set, and the multi-dimensional rehabilitation progress report is overlaid and compared with the phased rehabilitation goals based on the refresh frequency to generate a visual comparison result of rehabilitation progress.

[0145] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of the above-described method for upper limb functional rehabilitation assistive system after breast cancer surgery.

[0146] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0147] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0148] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A method for assisting in the rehabilitation of upper limb function after breast cancer surgery, characterized in that, The method includes: The patient's upper limb function data were collected and fused to generate an upper limb function feature set. The upper limb function data included joint range of motion, muscle strength index, lymphedema data, VAS pain score and Borg fatigue index. The rehabilitation phase is divided into four stages: 1-7 days post-surgery, 8-14 days post-surgery, 15 days-3 months post-surgery, and 3-6 months post-surgery. Based on the upper limb functional feature set, corresponding phased rehabilitation goals are generated using a preset mapping mechanism. Based on the phased rehabilitation goals, a candidate rehabilitation plan is generated using a machine learning-based rehabilitation plan generation model. A rehabilitation prediction score is generated by using a prediction model based on multidimensional assessment, combined with the upper limb functional feature set and candidate rehabilitation programs, and the candidate rehabilitation program with the highest score is selected as the optimal rehabilitation program.

2. The method according to claim 1, characterized in that, Based on the phased rehabilitation goals, the generation of candidate rehabilitation plans using a machine learning-based rehabilitation plan generation model includes: Acquire individual patient data, including age, body mass index, and postoperative complication level; Using an action type generation model, combined with the phased rehabilitation goals and individual patient data, a set of candidate action types that match the current rehabilitation stage is generated through feature matching. Based on the upper limb functional feature set, the baseline intensity parameters of each candidate movement type are determined by the training intensity prediction model. The baseline intensity parameters include the unit movement resistance value and the maximum safe repetition range. Based on the phased rehabilitation goals, the baseline intensity parameters are mapped to training duration configurations using a preset intensity-duration conversion matrix. The duration configurations include the duration of a single training session and the training frequency. A set of candidate rehabilitation plans is generated by dynamically combining the candidate action type set, the associated baseline intensity parameters, and their mapped duration configurations.

3. The method according to claim 2, characterized in that, The process of generating a rehabilitation prediction score using a prediction model based on multi-dimensional assessment, combined with the upper limb functional feature set and candidate rehabilitation programs, includes: Based on the movement type, intensity parameters, and training duration configuration of the candidate rehabilitation program, a predicted value for functional recovery progress is generated by using a prediction model combined with the joint range of motion and muscle strength indicators of the upper limb functional feature set. Based on the lymphedema data of the upper limb functional feature set, the historical trends of VAS pain score and Borg fatigue index, and the training intensity parameters of the candidate rehabilitation program, calculate the risk level of edema worsening and the cumulative risk value of muscle pain and fatigue. By combining the patient's age, body mass index, and training frequency configuration of the candidate rehabilitation programs in the individual patient data, the feasibility and suitability of the training programs are evaluated. The predicted value of functional recovery progress, the risk level of edema worsening, the cumulative risk value of muscle pain and fatigue, and the feasibility fit are normalized and weighted to generate a rehabilitation prediction score.

4. The method according to claim 1, characterized in that, The selection of the candidate rehabilitation plan with the highest score as the optimal rehabilitation plan is achieved through the following formula: in: R * This indicates the optimal rehabilitation plan. R represents the set of candidate rehabilitation protocols. p Indicates the risk of pain worsening, R b Let P(·) represent the cumulative fatigue level, and let P(·) represent the predicted functional recovery progress, calculated using an LSTM network. Where, θ ROM (t) represents the joint range of motion vector at time t, μ strength (t) represents the muscle strength index at time t, a k Scheme A k Includes action type encoding, s k Scheme A k Strength parameters; R e (·) indicates the risk level of worsening edema, and the calculation satisfies: Where ∈(t) represents the current rate of change of lymphedema volume, ∈ base Indicates the safety threshold for the current recovery stage, ||s k ||2 represents scheme A k The baseline strength parameters are w1 and w2, which represent clinical risk weight coefficients, and σ represents the Sigmoid function. A d (·) represents the executability fit, calculated using the following formula: Among them, f f Scheme A k The training frequency, BMI represents body mass index, age represents patient age, and k1, k2 and k3 represent individual adaptive adjustment coefficients; α, β, γ, δ, and η represent dynamic weighting coefficients, and α, β, and γ satisfy: Where, when ∈(t)>∈ threshold When (lymphedema threshold at each recovery stage), β = 1.

0.

5. The method according to claim 1, characterized in that, The method further includes: Calculate the dynamic deviation between the dynamic monitoring parameters of the upper limb functional feature set and the preset target parameters of the current stage rehabilitation plan; Based on the parameter deviation, the corresponding parameters of the rehabilitation plan generation model are adjusted using an online learning mechanism and a backpropagation algorithm.

6. The method according to claim 2, characterized in that, The dynamic monitoring parameters include real-time angle values ​​of joint range of motion, instantaneous detection values ​​of muscle strength indicators, real-time volume change rate of lymphedema data, VAS pain score, and Borg fatigue index. The corresponding parameters of the rehabilitation program generation model are adjusted using an online learning mechanism and a backpropagation algorithm, including the priority weight matrix of the action type generation model and the mapping coefficients of the intensity-duration conversion matrix.

7. The method according to claim 1, characterized in that, The method further includes: The upper limb functional feature set is visualized using a three-dimensional skeletal motion model to generate a multi-dimensional rehabilitation progress report that includes joint range of motion change curves, muscle strength recovery heatmaps, edema volume comparison charts, pain trend curves, and fatigue heatmaps. The visualization refresh frequency is set according to the update cycle of the upper limb functional feature set, and the multi-dimensional rehabilitation progress report and the phased rehabilitation goals are overlaid and compared based on the refresh frequency to generate a visual comparison result of rehabilitation progress.

8. A rehabilitation assistive system for upper limb function after breast cancer surgery, characterized in that, The system includes: The data acquisition and fusion module is used to collect the patient's upper limb function data and fuse it to generate an upper limb function feature set. The upper limb function data includes joint range of motion, muscle strength index, lymphedema data, VAS pain score and Borg fatigue index. The phase goal generation module is used to divide the rehabilitation phase into four stages: 1-7 days post-surgery, 8-14 days post-surgery, 15 days-3 months post-surgery, and 3-6 months post-surgery. Based on the upper limb functional feature set, it generates corresponding phased rehabilitation goals using a preset mapping mechanism. The rehabilitation plan generation module is used to generate candidate rehabilitation plans based on the phased rehabilitation goals and using a machine learning-based rehabilitation plan generation model. The evaluation and selection module is used to generate a rehabilitation prediction score by using a prediction model based on multi-dimensional evaluation, combining the upper limb functional feature set and candidate rehabilitation programs, and select the candidate rehabilitation program with the highest score as the optimal rehabilitation program.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images