An immersive experience virtual reality system for orthopedic sports injury rehabilitation training

Abstract

The application discloses an immersive experience virtual reality system for orthopedic sports injury rehabilitation training, comprising: a synergy defect module for reading bone-muscle-ligament synergy characteristic parameters of a patient's injury site, coupling and arranging according to anatomical sites, action directions and stress relationships, and forming a bone-muscle-ligament synergy defect main graph; a primitive sequence module for extracting a leading defect chain and a protection constraint chain from the bone-muscle-ligament synergy defect main graph, combining repair and avoidance sequences, disassembling a rehabilitation target, and forming a joint training primitive sequence; and a fusion reference module for rewriting the joint training primitive sequence into an immersive scene execution bottom graph, extracting a primitive guide area parameter corresponding to a current virtual training primitive, and forming a consistent fusion reference; and the application realizes early identification of a patient's action preparation state and real-time adjustment of guide content.

Description

Technical Field

[0001] This invention relates to the field of orthopedic rehabilitation training technology, and specifically to an immersive virtual reality system for orthopedic sports injury rehabilitation training. Background Technology

[0002] Virtual reality rehabilitation training has evolved from early simple scenario presentations to digital rehabilitation methods that can provide real-time feedback, task-based training, and enhanced compliance. Especially in musculoskeletal rehabilitation, immersive scenarios are considered to help improve training participation, pain management, and functional recovery. At the same time, research on digital twins and personalized modeling for musculoskeletal diseases has begun to emphasize unifying imaging, kinematic, and force information into patient-specific models to support more accurate rehabilitation decisions.

[0003] Existing technologies still have two shortcomings: First, the training content of many virtual reality rehabilitation systems mainly relies on preset action templates or single functional indicators, which do not adequately depict the synergistic relationship between bony stability, muscle activation, ligament restriction, force distribution, and compensatory posture. This results in a low degree of matching between training objectives and the real state of the injured site, making it difficult to achieve refined training organization targeting specific injury mechanisms. Second, the scene execution, action recognition, and training evaluation in existing systems are mostly processed in a decentralized manner. The data generated during the execution process is difficult to continuously write back to the patient model and subsequent scene configurations, making it difficult to form a closed-loop update of the training process. Summary of the Invention

[0004] Purpose of the invention: The purpose of this invention is to provide an immersive virtual reality system for orthopedic sports injury rehabilitation training, which solves the problems of insufficient matching of the actual state of the injured site and lack of closed-loop updates in the training process.

[0005] Technical solution: The present invention provides an immersive virtual reality system for orthopedic sports injury rehabilitation training, comprising: a synergistic defect module, a primitive sequence module, a fusion reference module, an intent vector module, a guided execution module, and an evidence rewriting module; Among them, the synergistic defect module is used to read the musculoskeletal synergistic characteristic parameters of the patient's injury site, and couple and organize them according to anatomical location, direction of movement and force relationship to form a bone-muscle-ligament synergistic defect master map to characterize the synergistic defect relationship of bones, muscles and ligaments. The primitive sequence module is used to extract the dominant defect chain representing the main repair path and the constraint chain representing the safety boundary protection from the main bone-muscle-ligament collaborative defect map, and decompose the rehabilitation goal by combining the repair sequence and the avoidance sequence to form a joint training primitive sequence for guiding step-by-step training. The fusion reference module is used to convert the joint training primitive sequence into an immersive scene execution base map and extract the primitive guidance area parameters corresponding to the current virtual training primitive to form a consistent fusion reference for unifying the scene and guidance parameters; the intent vector module is used to collect the patient's action preparation feature signal using the consistent fusion reference and perform item-by-item correspondence and fusion with the primitive guidance area parameters to form an action intent probability vector for predicting the patient's action tendency. The guided execution module is used to rewrite the primitive guidance area parameters into a feedforward guided execution page based on the matching relationship between the action intent probability vector and the immersive scene execution base map, and drive the execution of the current virtual training primitive to form a collaborative execution evidence flow that records the entire execution process; The evidence write-back module is used to write back the collaborative execution evidence stream to the main graph of bone-muscle-ligament collaborative defects, and simultaneously revise the parameters of the joint training primitive sequence, immersive scene execution base map and primitive guidance area based on the write-back results.

[0006] Furthermore, the specific process of forming the bone-muscle-ligament collaborative defect master map by the collaborative defect module is as follows: Multi-source rehabilitation records of the patient's injury site are read, and anatomical structures and dynamic training information are integrated according to the direction of movement to form a collaborative mapping record set; based on the collaborative mapping record set, a correspondence is established for the bone-muscle-ligament collaborative defect features under the same anatomical location and direction of movement, forming a collaborative defect index set with directional constraints and force associations; the collaborative defect index set is expanded into a graphical structure according to anatomical location, direction of movement, and force changes to generate the bone-muscle-ligament collaborative defect master map.

[0007] Furthermore, the specific process of the primitive sequence module extracting the dominant defect chain and protective constraint chain and decomposing the rehabilitation goal is as follows: Defect candidate paths and constraint candidate paths are obtained from the bone-muscle-ligament collaborative defect main map to form a link candidate set; the dominant defect chain and protective constraint chain are selected from the link candidate set and written into the repair order and avoidance order respectively to form the basis for decomposing the rehabilitation goal.

[0008] Furthermore, the primitive sequence module forms a joint training primitive sequence as follows: based on the rehabilitation goal decomposition criteria, the rehabilitation goal is decomposed into multiple training primitives to form a primitive candidate sequence; the primitive candidate sequence is interlocked and revised, retaining the training primitives that are consistent with the dominant defect chain and do not break the protective constraint chain, and rewriting the training primitives that conflict with the protective constraint chain but have a promoting effect on the dominant defect chain as transitional training primitives, and merging them to form a joint training primitive sequence.

[0009] Furthermore, the fusion reference module is used to form a consistent fusion reference by: expanding the attributes of each training primitive in the joint training primitive sequence according to its anatomical position, action direction, force range, interactive feedback, switching conditions, and backtracking conditions, and converting it into an immersive scene execution base map; extracting the current primitive scene page from the immersive scene execution base map, and solving the guidance field strength based on factors such as reference trajectories, entry degree, fit degree, proximity degree, conflict degree, and deviation degree in the current primitive scene page, to form a primitive guidance area parameter set.

[0010] Furthermore, the intent vector module is used to collect action preparation feature signals as follows: based on the consistency fusion reference and the sequential relationship of the current virtual training primitive, it determines the associated training primitives that have switching and backtracking relationships with the current virtual training primitive, forming a candidate primitive competition table; based on the candidate primitive competition table, it determines the action preparation acquisition conditions and collects the patient's action preparation feature signals to form an action preparation feature signal sequence.

[0011] Furthermore, the intent vector module is used to form the action intent probability vector by: matching the action preparation feature signal sequence with the primitive guidance area parameters item by item to form a candidate alignment evidence sequence; performing consistency fusion on the candidate alignment evidence sequence, and classifying the candidate alignment evidence corresponding to the current virtual training primitive and the associated training primitive into candidate intent components; merging the candidate intent components according to the order of the candidate primitives in the candidate primitive competition table to generate the action intent probability vector.

[0012] Furthermore, the guided execution module is used to form a collaborative execution evidence flow as follows: Based on the matching relationship between the action intent probability vector and the immersive scene execution base map, the method of rewriting the parameters of the current primitive guidance area is determined to form a parameter rewriting relationship; the parameters of the primitive guidance area are rewritten according to the parameter rewriting relationship to generate a feedforward guided execution page; the feedforward guided execution page is used to drive the current virtual training primitive and continuously acquire the execution status; the feedforward guided execution page is rewritten in a linked manner according to the execution status to form a continuous execution status; the continuous execution status is collected in chronological order to form a collaborative execution evidence flow.

[0013] Furthermore, the evidence write-back module writes the collaborative execution evidence flow back to the main graph of bone-muscle-ligament collaborative defects as follows: the collaborative execution evidence flow is unfolded in chronological order and mapped to the main graph of bone-muscle-ligament collaborative defects, the joint training primitive sequence, the immersive scene execution base map, and the primitive guidance area parameters, forming a dual ledger alignment table; the records in the dual ledger alignment table that correspond to the same graph node, the same training primitive, the same scene layer, and the same parameter group are bound to form a federated write-back unit set.

[0014] Furthermore, the evidence write-back module revises the parameters of the joint training primitive sequence, the immersive scenario execution base map, and the primitive guidance area as follows: For each federated write-back unit in the federated write-back unit set, the topological consensus write-back index is solved based on the consistency between its internal real execution state and reference state, the coherence of primitive continuation, and the degree of conflict, and the corresponding write-back direction and write-back intensity are determined to form a write-back decision table; based on the write-back decision table, the parameters of the joint training primitive sequence, the immersive scenario execution base map, and the primitive guidance area are revised synchronously.

[0015] Beneficial Effects: Compared with the prior art, the present invention has the following significant advantages: By extracting the dominant defect chain and the protective constraint chain and forming a joint training primitive sequence, the present invention achieves the synergistic decomposition of rehabilitation goals between repair advancement and risk avoidance, thereby improving the pertinence and safety of orthopedic sports injury rehabilitation training; by forming a motion intention probability vector and rewriting the primitive guidance area parameters into a feedforward guidance execution page, the present invention achieves the early identification of the patient's motion preparation state and the real-time adjustment of the guidance content, thereby improving the responsiveness, guidance accuracy and individualized adaptation ability of immersive training. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the system of the present invention; Figure 2 This is a flowchart illustrating the generation of the main image of bone-muscle-ligament synergistic defects according to the present invention.

[0017] Figure 3 This is a flowchart illustrating the formation of a joint training primitive sequence according to the present invention.

[0018] Figure 4 This is a flowchart illustrating the generation of collaborative execution evidence flow according to the present invention. Detailed Implementation

[0019] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0020] like Figure 1-4 As shown, this embodiment of the invention provides an immersive virtual reality system for orthopedic sports injury rehabilitation training, comprising: The collaborative defect module is used to read the musculoskeletal synergistic characteristic parameters of the patient's injury site, and to couple and organize them according to anatomical location, direction of movement and force relationship to form a main image of musculoskeletal synergistic defects.

[0021] Read the multi-source rehabilitation records corresponding to the patient's injury site, and integrate the anatomical structure information and dynamic training information according to the direction of movement to form a collaborative mapping record set.

[0022] The specific process involves using the patient's injury site as the search criterion, comparing the descriptions of the injury site location in each record of the multi-source rehabilitation records one by one, filtering out multi-source rehabilitation records that match the patient's injury site, and reading the anatomical structure information related to the injury site location and the dynamic training information related to the movement direction one by one. At the same time, content that can characterize the injury site location, structural state, movement changes, and movement direction is extracted. Then, the injury site location and structural state in the anatomical structure information are matched with the movement changes in the dynamic training information according to the movement direction. For example, if the anatomical structure information records the structural state of a certain injury site, and the dynamic training information records the movement changes of the same injury site in the corresponding movement direction, then the structural state and the movement changes in the corresponding movement direction are written into the same record, so that the anatomical structure information in the same movement direction and the dynamic training information in the same movement direction form a correspondence. The matched content is then written into the same record, and multiple matched records are collected to form a collaborative mapping record set.

[0023] Based on the co-mapping record set, a correspondence is established according to the musculoskeletal co-defect characteristics of the same anatomical location and the same direction of movement, forming a co-defect index set with directional constraints and force associations.

[0024] The specific process involves reading anatomical structure information and dynamic training information from each record in the co-mapping record set. The location of the injury site, bone state, muscle state, and ligament state corresponding to the same anatomical position are then listed side-by-side with the changes in movement, force, and posture corresponding to the same direction of movement. From the listed content, musculoskeletal synergistic defect features that reflect the joint changes of bones, muscles, and ligaments are extracted. The musculoskeletal synergistic defect features corresponding to the same anatomical position and the same direction of movement are written into the same index record. At the same time, the direction of movement is written into the corresponding position as directional constraint content, and the changes in force are written into the corresponding position as force association content. The multiple index records are then aggregated to form a co-defect index set.

[0025] The synergistic defect index set is expanded into a graphical structure according to anatomical location, direction of movement, and force changes to form a main map of bone-muscle-ligament synergistic defects.

[0026] The specific process involves merging the index records in the synergistic defect index set according to their anatomical locations into corresponding location entries. For example, multiple index records corresponding to the same ankle joint location are merged into the ankle joint location entry, and multiple index records corresponding to the same knee joint location are merged into the knee joint location entry. Then, the movement directions recorded in each index record are linked together in the order of association between anatomical locations, and the force changes corresponding to each movement direction are linked together along the change relationship under the same movement direction. This allows the musculoskeletal synergistic defect characteristics under the same anatomical location to fall into a unified graph structure expression along with the directional constraint content and force association content under the same movement direction. Subsequently, the association between anatomical locations, the correspondence between movement directions, and the connection between force changes are all expanded into the graph structure. The graph structure of the synergistic defect index set is expanded to form the main graph of musculoskeletal synergistic defects.

[0027] The primitive sequence module is used to extract the dominant defect chain and protective constraint chain from the main map of bone-muscle-ligament synergistic defects, and decompose the rehabilitation goals by combining the order of repair and the order of avoidance to form a joint training primitive sequence.

[0028] Defect candidate paths and constraint candidate paths are obtained from the main defect graph of bone-muscle-ligament synergy, forming a link candidate set.

[0029] The specific process is as follows: based on the anatomical location association, movement direction correspondence, and force change relationship within the main image of bone-muscle-ligament synergistic defect, the extended networks related to the bone-muscle-ligament synergistic defect characteristics are retrieved one by one. The associated networks that are continuously distributed along the anatomical location and can characterize the evolution trend of bone-muscle-ligament synergistic defect characteristics are merged into defect candidate paths. Then, the associated networks that are continuously distributed along the movement direction and force change and can characterize the restriction trend of directional constraint content and force-related content are merged into constraint candidate paths. Finally, the defect candidate paths and constraint candidate paths are loaded into the same candidate set according to the corresponding anatomical location, movement direction, and force change, forming a link candidate set.

[0030] The dominant defect chain and the protective constraint chain are selected from the candidate chain set, and the dominant defect chain and the protective constraint chain are written into the repair priority order and the avoidance priority order, respectively, to form the basis for decomposing the rehabilitation target.

[0031] The specific process involves verifying the anatomical location, direction of motion, and force changes of each defect candidate path and constraint candidate path in the candidate link set, one by one. The musculoskeletal synergistic defect features corresponding to each defect candidate path are then listed along the path's extension order. The coverage, continuity, and concentration of the musculoskeletal synergistic defect features carried by each defect candidate path are compared to screen out the defect candidate paths that can centrally represent the main evolutionary trend of musculoskeletal synergistic defect features, and these are established as the dominant defect chains. Simultaneously, the content corresponding to the direction of motion of the constraint candidate paths is also analyzed. The order of force changes along the path is listed, and then the coverage, continuity, and degree of restriction concentration of the directional constraint content and force-related content carried by each constraint candidate path are compared. The constraint candidate paths that can centrally represent the direction of the directional constraint content and force-related content restriction are selected and established as protective constraint chains. The order of the dominant defect chain along the link is loaded into the repair sequence, and the order of the protective constraint chain along the link is loaded into the avoidance sequence. The repair sequence and the avoidance sequence are combined into the rehabilitation goal decomposition basis to form the rehabilitation goal decomposition basis.

[0032] Based on the criteria for decomposing rehabilitation goals, the rehabilitation goals are broken down into training primitives, forming a candidate sequence of primitives.

[0033] The specific process is as follows: based on the rehabilitation goal decomposition criteria, the contents corresponding to the dominant defect chain and the contents corresponding to the protective constraint chain in the rehabilitation goals are merged item by item. The action direction, force changes and advancement levels of the rehabilitation goals are segmented according to the order of repair and avoidance, so that each content can correspond to an independent training primitive. Then, each training primitive is arranged one by one according to the order relationship already loaded in the rehabilitation goal decomposition criteria. The training primitives that can inherit the previous training primitive and extend the next training primitive are sequentially loaded into the same sequence to form a primitive candidate sequence.

[0034] Interlocked revisions are performed on the candidate primitive sequence. Training primitives that are consistent with the dominant defect chain and do not break the protection constraint chain are retained, and training primitives that conflict with the protection constraint chain but have a promoting effect on the dominant defect chain are rewritten as transitional training primitives, forming a joint training primitive sequence.

[0035] The specific process involves unfolding each training primitive in the candidate sequence according to its original order, listing the corresponding action direction, force change, and propulsion position, and comparing them item by item with the link content of the dominant defect chain that already has the repair sequence loaded. This verifies whether the training primitive can positively propel the musculoskeletal synergistic defect characteristics pointed to by the dominant defect chain. Simultaneously, the same training primitive is compared item by item with the link content of the protection constraint chain that already has the avoidance sequence loaded, verifying whether the training primitive will touch the directional constraint content and force association content corresponding to the protection constraint chain. After two-way comparison, training primitives that are consistent with the dominant defect chain and do not break the protection constraint chain are retained in the primitive candidate list. In the sequence, training primitives that conflict with the protective constraint chain but still contribute to the dominant defect chain are read from the original training primitives. The action direction and force changes corresponding to the conflict position are reordered into milder connecting content, transforming the read training primitives into transitional training primitives. The remaining training primitives and the rewritten transitional training primitives are then merged, concatenated, and grouped according to the order in the primitive candidate sequence. This ensures that the ending position of the previous training primitive is connected to the starting position of the next training primitive and the transitional training primitive, and that the advancing content of the next training primitive and the transitional training primitive still extends around the dominant defect chain without exceeding the scope defined by the protective constraint chain, thus forming a joint training primitive sequence.

[0036] The fusion reference module is used to rewrite the joint training primitive sequence into an immersive scene execution base map and extract the primitive guidance area parameters corresponding to the current virtual training primitive to form a consistent fusion reference.

[0037] The training primitives in the joint training primitive sequence are expanded according to anatomical location, action direction, force range, interactive feedback, switching conditions, and rollback conditions, and rewritten as an immersive scene execution base map.

[0038] The specific process involves reading each training primitive in the joint training primitive sequence into its anatomical location, action direction, force range, interactive feedback, switching condition, and rollback condition. The anatomical location is then mapped to its corresponding location in the immersive scene, the action direction to its corresponding direction, the force range to its corresponding constraint, the interactive feedback to its corresponding feedback, the switching condition to its corresponding switching, and the rollback condition to its corresponding rollback. Finally, each mapped content is unfolded and loaded into the same scene structure according to the order in the joint training primitive sequence, thus forming the immersive scene execution base map.

[0039] Extract the current primitive scene page from the immersive scene execution base map, and solve the guidance field strength based on the current primitive scene page to form the primitive guidance area parameter set.

[0040] The specific process involves retrieving the anatomical location, action direction, force range, interactive feedback, switching conditions, and rollback conditions corresponding to the current training primitive from the immersive scene execution base map. All scene content corresponding to the current training primitive is then merged into the same page, and the current primitive scene page is analyzed. The corresponding positions on the current primitive scene page are used to determine the location range of the starting and target guidance areas; the corresponding directions are used to determine the action direction of the passage guidance area; the corresponding constraint content is used to determine the force constraints of the passage guidance area; the corresponding feedback content is used to determine the interactive feedback of the starting and target guidance areas; the corresponding switching content is used to determine the switching conditions of the target guidance area; and the corresponding rollback content is used to determine the rollback conditions of the rollback guidance area. Finally, the content corresponding to the starting, passage, target, and rollback guidance areas in the guidance field corresponding to the current training primitive is aggregated into the same parameter set, forming the primitive guidance area parameter set.

[0041] The expression for solving the guiding field intensity is: ; in, The value represents the guiding field intensity at the location point, x represents the location point to be determined in the current primitive scene page, T represents the action advancement duration corresponding to the current primitive scene page, and t represents the continuous time variable during the action advancement process. This represents the position of the reference guide trajectory at time t, and α represents the position decay coefficient. This indicates the entry level corresponding to the initial guidance region at time t. This indicates the degree of alignment of the traffic guidance area with the direction at time t. This indicates the degree of approach of the target guidance area at time t. This indicates the degree of conflict corresponding to the range of forces at time t. This indicates the degree of deviation corresponding to the backoff condition at time t. Let t represent the degree of fluctuation corresponding to the interactive feedback at time t, and exp represent the natural exponential function.

[0042] It should be noted that this is calculated from the reference guide trajectory and the allowed deviation range in the current primitive scene page.

[0043] Align the primitive guide area parameter set with the current primitive scene page to form a consistent fusion reference.

[0044] The specific process involves comparing the starting guidance area, the passage guidance area, the target guidance area, and the backtracking guidance area in the primitive guidance area parameter set with their corresponding positions, directions, constraints, feedback, switching, and backtracking content in the current primitive scene page. The starting guidance area is assigned to its corresponding position in the current primitive scene page, the passage guidance area to its corresponding direction, the target guidance area to its corresponding feedback content, and the backtracking guidance area to its corresponding backtracking content. Finally, all parameters in the primitive guidance area parameter set and scene content in the current primitive scene page are merged into the same reference expression according to the same training primitive's arrangement. This ensures a one-to-one correspondence between the primitive guidance area parameter set and the current primitive scene page in terms of position, direction, constraints, feedback, switching, and backtracking, forming a consistent and integrated reference.

[0045] The intent vector module is used to collect action preparation feature signals using consistency fusion references, and then perform consistency fusion with the primitive guidance area parameters item by item to form an action intent probability vector.

[0046] Based on the consistency fusion reference and the sequential relationship of the current virtual training primitive, training primitives that have switching and backtracking relationships with the current virtual training primitive are identified, forming a candidate primitive competition table.

[0047] The specific process is as follows: Consistency fusion refers to the alignment content of the parameter set of the primitive guidance area corresponding to the current virtual training primitive and the current primitive scene page. Combining the arrangement order relationship between the current virtual training primitive, the preceding position training primitive, and the following position training primitive, the training primitive located in the joint training primitive sequence that is in the preceding position of the current virtual training primitive and the training primitive located in the following position of the current virtual training primitive are searched. The training primitives that have a switching relationship with the current virtual training primitive and the training primitives that have a backsliding relationship with the current virtual training primitive are determined. The training primitives with the switching relationship and the training primitives with the backsliding relationship with the current virtual training primitive are collected to form a candidate primitive competition table.

[0048] Based on the candidate primitive competition table, the action preparation acquisition conditions are determined, and action preparation feature signals are acquired to form an action preparation feature signal sequence.

[0049] The specific process is as follows: the candidate primitive competition table records training primitives that have switching and rollback relationships with the current virtual training primitive, thereby determining the action preparation acquisition conditions and specifying the acquisition time, acquisition site, acquisition type, and acquisition frequency of the action preparation feature signals. The acquisition start time is determined according to the acquisition time, the body acquisition position is determined according to the acquisition site, the signal type is determined according to the acquisition type, and the acquisition interval is determined according to the acquisition frequency. When the acquisition start time is reached, the body acquisition position acquires action preparation feature signals according to the signal type and acquisition interval. The acquired action preparation feature signals are arranged in the acquisition order to form an action preparation feature signal sequence.

[0050] The action preparation feature signal sequence is matched one by one with the primitive guide region parameters in the consistency fusion reference to form a candidate alignment evidence sequence.

[0051] The specific process involves matching the action preparation feature signals in the action preparation feature signal sequence with the contents corresponding to the start guidance area, passage guidance area, target guidance area, and backtracking guidance area in the consistency fusion reference according to the acquisition order, binding the action preparation feature signals with the contents of the primitive guidance area parameter set as candidate alignment evidence, and arranging the candidate alignment evidence in the corresponding order to form a candidate alignment evidence sequence.

[0052] The candidate alignment evidence sequence is fused with consistency, and the candidate alignment evidence corresponding to the current virtual training primitive and the associated training primitive is integrated into a candidate intent component.

[0053] The specific process is as follows: the candidate alignment evidence in the candidate alignment evidence sequence is matched with the primitive guidance area parameters in the consistency fusion reference item by item according to the collection order and the corresponding order. Based on the switching relationship and backoff relationship recorded in the candidate primitive competition table, the matched candidate alignment evidence is divided into candidate alignment evidence corresponding to the current virtual training primitive and candidate alignment evidence corresponding to the associated training primitive. The candidate alignment evidence corresponding to the current virtual training primitive and the candidate alignment evidence corresponding to the associated training primitive are respectively organized into candidate intent components corresponding to the current virtual training primitive and the associated training primitive.

[0054] The candidate intent components are merged according to the order of the candidate primitives in the candidate primitive competition table to form an action intent probability vector.

[0055] The specific process is as follows: the candidate intention components are listed one by one according to the order of the candidate primitives recorded in the candidate primitive competition table; the candidate intention components are written into the corresponding positions according to the order of the candidate primitives in the candidate primitive competition table; and all the candidate intention components that have been written into the corresponding positions are arranged from beginning to end along the order of the candidate primitive competition table and gathered into the same vector expression to form the action intention probability vector.

[0056] The guided execution module is used to rewrite the primitive guidance area parameters into feedforward guided execution pages based on the matching relationship between the action intent probability vector and the immersive scene execution base map, and drive the current virtual training primitive to form a collaborative execution evidence flow.

[0057] Based on the matching relationship between the action intent probability vector and the immersive scene execution base map, the rewriting method of the primitive guidance area parameters corresponding to the current virtual training primitive is determined, and the parameter rewriting relationship is formed.

[0058] The specific process is as follows: based on the distribution of each candidate intent component in the action intent probability vector, and comparing it with the anatomical position, action direction, force range, interactive feedback, switching conditions, and backtracking conditions corresponding to the current virtual training primitive in the immersive scene execution base map, the degree of correspondence between the action tendency pointed to by the action intent probability vector and the current virtual training primitive in the immersive scene execution base map is determined. Then, the degree of correspondence is implemented in the starting guidance area, passage guidance area, target guidance area, and backtracking guidance area in the primitive guidance area parameters. The content, order, and corresponding position of the primitive guidance area parameters that should be adjusted under the current virtual training primitive are determined. The determined adjustment content and corresponding position are loaded into the same relational expression to form a parameter transposition relationship.

[0059] The parameters of the primitive guidance area are rewritten according to the parameter rewriting relationship to form the feedforward guidance execution page.

[0060] The specific process is as follows: according to the parameter rewriting relationship, the contents corresponding to the starting guide area, the passage guide area, the target guide area, and the fallback guide area in the primitive guide area parameters are compared item by item with the adjustment contents, order, and corresponding positions already specified in the parameter rewriting relationship. The contents that need to be changed are moved into the corresponding positions, and the contents that need to be retained are loaded into their original positions. Then, the rewritten starting guide area, passage guide area, target guide area, and fallback guide area are collected according to the arrangement of contents corresponding to the current virtual training primitive and form a feedforward guide execution page.

[0061] The feedforward guided execution page is used to drive the current virtual training primitive, while continuously acquiring the execution state and rewriting the feedforward guided execution page according to the execution state to form a continuous execution state.

[0062] The specific process is as follows: when using the feedforward guidance execution page to drive the current virtual training primitive, the start guidance area, passage guidance area, target guidance area, and backtrack guidance area already loaded in the feedforward guidance execution page are continuously applied to the corresponding advancement process of the current virtual training primitive according to the arrangement of the content of the current virtual training primitive. During the advancement process, the execution state represented by the changes in action, force, posture, and advancement is continuously collected. Then, the execution state is compared with the start guidance area, passage guidance area, target guidance area, and backtrack guidance area in the feedforward guidance execution page one by one. The content that needs to be adjusted is rewritten in conjunction with the execution state, so that the feedforward guidance execution page keeps changing synchronously during the continuous advancement process. The continuously changed feedforward guidance execution page and the continuously changing execution state are connected together in time sequence to form a continuous execution state.

[0063] The continuous execution status is collected in chronological order to form a collaborative execution evidence flow.

[0064] The specific process involves converging the changes in the feedforward guidance execution page, the current virtual training primitives, and the execution state changes recorded sequentially in the continuous execution state, and then linking the corresponding content at each moment into the same continuous record according to the time sequence. This ensures that the changes in the feedforward guidance execution page, the current virtual training primitives, and the execution state changes within the same time frame remain consistent. Finally, all the consistent records are grouped into a unified streaming expression to form a collaborative execution evidence stream.

[0065] The evidence write-back module is used to write back the collaborative execution evidence stream to the main graph of bone-muscle-ligament collaborative defects, and simultaneously revise the parameters of the joint training primitive sequence, the immersive scene execution base map, and the primitive guidance area.

[0066] The collaborative execution evidence flow is unfolded in chronological order and mapped to the bone-muscle-ligament collaborative defect main graph, the joint training primitive sequence, the immersive scene execution base graph, and the primitive guidance area parameters, forming a dual ledger alignment table.

[0067] The specific process involves sequentially displaying the changes in the feedforward guidance execution page, the current virtual training primitives, and the changes in execution status recorded in the collaborative execution evidence stream along the time sequence. Then, based on the content corresponding to each moment, the corresponding nodes and relationships in the musculoskeletal synergistic defect main graph, the training primitives and their positions in the joint training primitive sequence, the corresponding scene content in the immersive scene execution base map, and the corresponding parameter content in the primitive guidance area parameters are respectively linked to the same time position. The linked content at the same time position is then written into the corresponding positions, creating a one-to-one alignment relationship between the collaborative execution evidence stream and the musculoskeletal synergistic defect main graph, the joint training primitive sequence, the immersive scene execution base map, and the primitive guidance area parameters in the timeline. The aligned content is then compiled into a table, forming a dual-ledger alignment table.

[0068] Bind records in the dual ledger pairing table that correspond to the same graph node, the same training primitive, the same scene layer, and the same parameter group to form a federated write-back unit set.

[0069] The specific process involves verifying each record in the dual ledger alignment table that points to the same graph node, the same training primitive, the same scene layer, and the same parameter group. Records with consistent time positions and related content are grouped into the same aggregation unit. This ensures that the corresponding node content in the bone-muscle-ligament synergistic defect main graph, the corresponding training primitive content in the joint training primitive sequence, the corresponding scene layer content in the immersive scene execution base graph, and the corresponding parameter group content in the primitive guidance area parameters are all within the same aggregation unit and linked in parallel. Then, all the aggregation units that have been linked in parallel are assembled into a unified set according to the arrangement content in the dual ledger alignment table, forming a federated write-back unit set.

[0070] For each federated write-back unit in the federated write-back unit set, the topological consensus write-back index is solved, and the corresponding write-back direction and write-back intensity are determined to form a write-back decision table.

[0071] The specific process involves comparing each federated write-back unit in the federated write-back unit set item by item. This includes verifying the content of the main graph corresponding to the bone-muscle-ligament synergy defects, the training primitive content corresponding to the joint training primitive sequence, the scene layer content corresponding to the immersive scene execution base graph, and the parameter group content corresponding to the primitive guidance area parameters within each federated write-back unit. The consistency, deviation, and correlation between the contents within the same federated write-back unit are then integrated into the same solution process to obtain the topological consensus write-back index. Based on the deviation and correlation trends represented by the topological consensus write-back index, the write-back direction and write-back intensity corresponding to each federated write-back unit are determined. The federated write-back unit, topological consensus write-back index, write-back direction, and write-back intensity are then sequentially listed in the same table to form a write-back decision table.

[0072] The expression for solving the topological consensus write-back metric is as follows: ; in, Indicates the first The topology consensus write-back metrics corresponding to each federated write-back unit Indicates the federal write-back unit number, Indicates the first The cumulative duration of evidence corresponding to each federal write-back unit Represents continuous time variables in the process of evidence accumulation. Indicates the first The mismatch suppression coefficient between the actual execution state and the reference execution state in each federated write-back unit. Indicates time Next The actual execution state vector corresponding to each federated write-back unit Indicates time Next The reference execution state vector corresponding to each federated write-back unit Indicates time Next The main federal writeback unit corresponds to one federal writeback unit Figure 1 To a certain extent, Indicates time Next Consistency of primitive continuation corresponding to each federal write-back unit Indicates time Next The degree of conflict corresponding to each federal write-back unit Indicates time Next The degree of interaction fluctuation corresponding to each federal write-back unit.

[0073] It should be noted that, It is the first The allowable deviation range between the actual execution state vector and the reference execution state vector in each federated write-back unit is calculated. It is obtained by aggregating the collaborative execution evidence stream content already bound in the federal write-back unit; It is obtained by aggregating the corresponding reference ledger content in the federal write-back unit.

[0074] Based on the write-back decision table, the parameters of the joint training primitive sequence, the immersive scene execution base map, and the primitive guidance area are revised synchronously.

[0075] The specific process is as follows: based on the write-back direction and write-back intensity corresponding to each federated write-back unit listed in the write-back decision table, the corresponding training primitives and their arrangement positions in the joint training primitive sequence are reordered, the corresponding scene content in the immersive scene execution base map is swapped, and the corresponding parameter content in the primitive guidance area parameters is rewritten. This ensures that the joint training primitive sequence, the immersive scene execution base map, and the primitive guidance area parameters maintain consistent changes at the corresponding positions pointed to by the same federated write-back unit. All the reordered, swapped, and rewritten content is then aggregated into a unified revision result according to the arrangement content in the write-back decision table.

[0076] In summary, this invention achieves the synergistic decomposition of rehabilitation goals between repair advancement and risk avoidance by extracting the dominant defect chain and protective constraint chain and forming a joint training primitive sequence, thereby improving the pertinence and safety of orthopedic sports injury rehabilitation training. Furthermore, by forming a motion intention probability vector and rewriting the primitive guidance area parameters into a feedforward guidance execution page, it enables early identification of the patient's motion preparation state and real-time adjustment of the guidance content, thereby improving the responsiveness, guidance accuracy, and individualized adaptation capabilities of immersive training.

[0077] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An immersive virtual reality system for orthopedic sports injury rehabilitation training, characterized in that, include: Collaborative defect module, primitive sequence module, fusion reference module, intent vector module, guided execution module, and evidence write-back module; Among them, the synergistic defect module is used to read the musculoskeletal synergistic characteristic parameters of the patient's injury site, and couple and organize them according to anatomical location, direction of movement and force relationship to form a bone-muscle-ligament synergistic defect master map to characterize the synergistic defect relationship of bones, muscles and ligaments. The primitive sequence module is used to extract the dominant defect chain representing the main repair path and the constraint chain representing the safety boundary protection from the main bone-muscle-ligament collaborative defect map, and decompose the rehabilitation goal by combining the repair sequence and the avoidance sequence to form a joint training primitive sequence for guiding step-by-step training. The fusion reference module is used to convert the joint training primitive sequence into an immersive scene execution base map and extract the primitive guidance area parameters corresponding to the current virtual training primitive to form a consistent fusion reference for unifying the scene and guidance parameters; the intent vector module is used to collect the patient's action preparation feature signal using the consistent fusion reference and perform item-by-item correspondence and fusion with the primitive guidance area parameters to form an action intent probability vector for predicting the patient's action tendency. The guided execution module is used to rewrite the primitive guidance area parameters into a feedforward guided execution page based on the matching relationship between the action intent probability vector and the immersive scene execution base map, and drive the execution of the current virtual training primitive to form a collaborative execution evidence flow that records the entire execution process; The evidence write-back module is used to write back the collaborative execution evidence stream to the main graph of bone-muscle-ligament collaborative defects, and simultaneously revise the parameters of the joint training primitive sequence, immersive scene execution base map and primitive guidance area based on the write-back results.

2. The immersive virtual reality system for orthopedic sports injury rehabilitation training according to claim 1, characterized in that, The specific process of forming the bone-muscle-ligament collaborative defect master map by the collaborative defect module is as follows: Read the multi-source rehabilitation records of the patient's injured area and integrate anatomical structures and dynamic training information according to the direction of movement to form a collaborative mapping record set; based on the collaborative mapping record set, establish corresponding relationships for bone-muscle-ligament collaborative defect features under the same anatomical location and direction of movement, forming a collaborative defect index set with directional constraints and force associations; expand the collaborative defect index set according to anatomical location, direction of movement, and force changes to generate the bone-muscle-ligament collaborative defect master map.

3. The immersive virtual reality system for orthopedic sports injury rehabilitation training according to claim 1, characterized in that, The specific process of extracting the dominant defect chain and protective constraint chain and decomposing rehabilitation goals by the primitive sequence module is as follows: Defect candidate paths and constraint candidate paths are obtained from the bone-muscle-ligament collaborative defect main map to form a link candidate set; the dominant defect chain and protective constraint chain are selected from the link candidate set and written into the repair priority order and avoidance priority order respectively to form the basis for decomposing rehabilitation goals.

4. The immersive virtual reality system for orthopedic sports injury rehabilitation training according to claim 3, characterized in that, The primitive sequence module forms a joint training primitive sequence as follows: According to the rehabilitation goal decomposition criteria, the rehabilitation goal is decomposed into multiple training primitives to form a primitive candidate sequence; the primitive candidate sequence is interlocked and revised, retaining the training primitives that are consistent with the dominant defect chain and do not break the protective constraint chain, and rewriting the training primitives that conflict with the protective constraint chain but have a promoting effect on the dominant defect chain as transitional training primitives, and merging them to form a joint training primitive sequence.

5. The immersive virtual reality system for orthopedic sports injury rehabilitation training according to claim 4, characterized in that, The fusion reference module is used to form a consistent fusion reference. Specifically, it expands the attributes of each training primitive in the joint training primitive sequence according to its anatomical location, action direction, force range, interactive feedback, switching conditions, and rollback conditions, and converts them into an immersive scene execution base map. Extract the current primitive scene page from the immersive scene execution base map, and solve the guidance field strength based on factors such as reference trajectory, entry degree, fit degree, proximity degree, conflict degree and deviation degree in the current primitive scene page to form the primitive guidance area parameter set.

6. The immersive virtual reality system for orthopedic sports injury rehabilitation training according to claim 1, characterized in that, The intent vector module is used to collect action preparation feature signals as follows: Based on the consistency fusion reference and the successive relationship of the current virtual training primitive, it determines the associated training primitives that have switching and backtracking relationships with the current virtual training primitive, and forms a candidate primitive competition table. The conditions for acquiring action preparation were determined based on the candidate primitive competition table, and the action preparation feature signals of the patients were acquired to form a sequence of action preparation feature signals.

7. The immersive virtual reality system for orthopedic sports injury rehabilitation training according to claim 6, characterized in that, The intent vector module is used to form an action intent probability vector. Specifically, it maps the action preparation feature signal sequence to the primitive guidance area parameters one by one to form a candidate alignment evidence sequence. It performs consistency fusion on the candidate alignment evidence sequence and organizes the candidate alignment evidence corresponding to the current virtual training primitive and the associated training primitive into candidate intent components. The candidate intent components are merged according to the order of the candidate primitives in the candidate primitive competition table to generate an action intent probability vector.

8. The immersive virtual reality system for orthopedic sports injury rehabilitation training according to claim 1, characterized in that, The guided execution module is used to form a collaborative execution evidence stream as follows: Based on the matching relationship between the action intent probability vector and the immersive scene execution base map, the method of rewriting the parameters of the current primitive guidance area is determined to form a parameter rewriting relationship; the parameters of the primitive guidance area are rewritten according to the parameter rewriting relationship to generate a feedforward guided execution page; the feedforward guided execution page is used to drive the current virtual training primitive and continuously acquire the execution status; the feedforward guided execution page is rewritten in a linked manner according to the execution status to form a continuous execution status; the continuous execution status is collected in chronological order to form a collaborative execution evidence stream.

9. The immersive virtual reality system for orthopedic sports injury rehabilitation training according to claim 8, characterized in that, The process by which the evidence write-back module writes back the collaborative execution evidence stream to the main graph of bone-muscle-ligament collaborative defects is as follows: the collaborative execution evidence stream is unfolded in chronological order and mapped to the main graph of bone-muscle-ligament collaborative defects, the joint training primitive sequence, the immersive scene execution base map and the primitive guidance area parameters, respectively, to form a dual ledger alignment table; Bind records in the dual ledger pairing table that correspond to the same graph node, the same training primitive, the same scene layer, and the same parameter group to form a federated write-back unit set.

10. An immersive virtual reality system for orthopedic sports injury rehabilitation training according to claim 9, characterized in that, The specific revision of the joint training primitive sequence, immersive scenario execution base map, and primitive guidance area parameters in the evidence write-back module is as follows: For each federated write-back unit in the federated write-back unit set, the topological consensus write-back index is solved based on the consistency between its internal real execution state and reference state, the coherence of primitive continuation, and the degree of conflict, and the corresponding write-back direction and write-back intensity are determined to form a write-back decision table. Based on the write-back decision table, the parameters of the joint training primitive sequence, the immersive scene execution base map, and the primitive guidance area are revised synchronously.

Images