A chronic lateral ankle instability intelligent diagnosis and treatment management system based on multidisciplinary collaboration

Abstract

The application provides a chronic lateral ankle instability intelligent diagnosis and treatment management system based on multidisciplinary cooperation, which comprises an intelligent diagnosis module, a multidisciplinary cooperative diagnosis and treatment module, a diagnosis and treatment monitoring module and a feedback optimization module; the intelligent diagnosis module is used for providing patient diagnosis basic characteristics in combination with individual pathological information of the patient; the multidisciplinary cooperative diagnosis and treatment module is used for realizing multidisciplinary cooperative individual diagnosis and treatment scheme making based on the patient diagnosis basic characteristics; the diagnosis and treatment monitoring module is used for monitoring the execution process of the individual diagnosis and treatment scheme; and the feedback optimization module is used for feeding back and optimizing the individual diagnosis and treatment scheme making in combination with diagnosis and treatment execution feedback information. The application generates individualized and self-iterative diagnosis and treatment decisions by taking multidisciplinary cooperation and result evaluation and optimization as cores, so that the curative effect and compliance are improved, the rehabilitation cycle is shortened, and the recurrence and complication risks are reduced under the premise of safety and compliance.

Description

Technical Field

[0001] This invention relates to the field of medical decision management system technology, and in particular to an intelligent diagnosis and treatment management system for chronic lateral ankle instability based on multidisciplinary collaboration. Background Technology

[0002] With increased participation in sports and an aging population, ankle sprains and their sequelae, chronic lateral instability, have a high incidence rate in clinical practice. Pain and functional limitations affect patients' ability to work and exercise. Current clinical pathways mostly revolve around classification and diagnosis based on imaging and physical examinations, with each department—such as orthopedics, rehabilitation, sports medicine, psychology, and traditional Chinese medicine acupuncture—developing its own treatment plan. However, there is insufficient collaboration between disciplines and fragmented processes: the timing and procedure of surgery, the intensity of postoperative rehabilitation stages, psychological interventions, and compliance management often rely on the experience of a single department. Interdisciplinary information sharing is not smooth, resulting in limited consistency and repeatability of protocols and significant differences in outcomes.

[0003] At the data level, imaging data, kinematic data, pain and function scales, and follow-up records come from diverse sources with varying dimensions, lacking a unified structured expression and comparable methods. Outpatient follow-ups are mostly discrete and low-frequency, making it difficult to reflect the quality of treatment plan implementation and the effectiveness of treatment at different stages in a timely manner. Existing intelligent tools mostly focus on single disciplines or single tasks, lacking the ability to manage the entire process for diseases like chronic ankle-related conditions that require long-term management, phased intervention, and multi-objective trade-offs. At the same time, publicly available decision support solutions mostly remain at the level of static rules or experience imitation, lacking a self-calibration mechanism based on real-world feedback, making it difficult to cope with real-world variables such as population differences, fluctuations in compliance, and resource constraints.

[0004] A review of publicly available technical solutions reveals that CN110021429A proposes a clinical intelligent decision support system for periarticular fractures, comprising an information input module, a diagnosis module, a decision module, and a case database. The diagnosis module includes a periarticular fracture classification module, which classifies periarticular fractures based on examination results from the information input module and clinical periarticular fracture classification standards. The classification of ankle fractures is based on the Lauge-Hansen classification and / or the Danis-Weber classification standards. This technical solution allows patients and their families to monitor and track their condition and treatment plans in real time. However, this solution focuses on diagnosis and decision support based on existing classification standards, lacking multidisciplinary collaborative individualized plan generation, interdisciplinary information exchange, and joint confirmation mechanisms. It struggles to simultaneously address multidimensional goals such as rehabilitation, psychological well-being, and functional reconstruction, remaining primarily at the classification-driven diagnostic level rather than providing comprehensive management. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of current methods by proposing an intelligent diagnosis and management system for chronic lateral ankle instability based on multidisciplinary collaboration.

[0006] The present invention adopts the following technical solution:

[0007] A multidisciplinary collaborative intelligent diagnosis and management system for chronic lateral ankle instability is disclosed. The system includes an intelligent diagnosis module, a multidisciplinary collaborative diagnosis and treatment module, a diagnosis and treatment monitoring module, and a feedback optimization module. The intelligent diagnosis module provides basic diagnostic characteristics of the patient by incorporating individualized pathological information. The multidisciplinary collaborative diagnosis and treatment module formulates individualized treatment plans based on the patient's basic diagnostic characteristics. The diagnosis and treatment monitoring module monitors the execution of the individualized treatment plan. The feedback optimization module optimizes the formulation of the individualized treatment plan by incorporating feedback information from the treatment execution process.

[0008] The intelligent diagnostic module includes a pathological image acquisition unit, a motion function acquisition unit, and a comprehensive feature extraction unit; the pathological image acquisition unit is used to acquire the patient's pathological image information; the motion function acquisition unit is used to acquire the patient's kinematic information; and the comprehensive feature extraction unit is used to extract basic diagnostic features from the pathological image information and kinematic information.

[0009] Furthermore, the multidisciplinary collaborative diagnosis and treatment module includes an initial plan generation unit, an interdisciplinary information interaction unit, and an individualized plan generation unit; the initial plan generation unit is used to generate an initial multidisciplinary collaborative diagnosis and treatment plan based on basic diagnostic features; the interdisciplinary information interaction unit is used to realize real-time information sharing among multidisciplinary physicians and support multidisciplinary physicians in modifying, adjusting, and confirming the initial multidisciplinary collaborative diagnosis and treatment plan; the individualized plan generation unit is used to generate an individualized diagnosis and treatment plan for the patient based on the multidisciplinary collaborative diagnosis and treatment plan confirmed by all multidisciplinary physicians.

[0010] Furthermore, the diagnosis and treatment monitoring module includes a process data acquisition unit, a rehabilitation indicator setting unit, and an effect comparison and analysis unit. The process data acquisition unit is used to collect the patient's diagnosis and treatment-related data during the execution of the individualized diagnosis and treatment plan. The diagnosis and treatment-related data includes kinematic parameters, physiological indicators, and psychological assessment data. The rehabilitation indicator setting unit is used to set the rehabilitation target indicators for the patient during the execution of the individualized diagnosis and treatment plan. The rehabilitation target indicators are pre-set by multidisciplinary physicians based on the patient's individual condition. The effect comparison and analysis unit is used to perform feature processing and comparative analysis on the patient's diagnosis and treatment-related data and rehabilitation target indicators to evaluate the execution effect of the individualized diagnosis and treatment plan and output the differences of various rehabilitation target indicators as monitoring results.

[0011] Furthermore, the initial treatment plan generation unit generates a set of candidate multidisciplinary collaborative treatment plans for the patient by combining the patient's basic diagnostic characteristics and based on the clinical guideline rule base and individualized rigid treatment plan constraints. Then, it calls a pre-established treatment plan evaluation model to evaluate each collaborative treatment plan in the set of multidisciplinary collaborative treatment plans and selects the collaborative treatment plan with the best comprehensive expected rehabilitation goal index evaluation result as the initial multidisciplinary collaborative treatment plan.

[0012] Furthermore, the initial scheme generation unit establishes the scheme evaluation model in the following manner:

[0013] S11: Obtain a case sample set through the historical case database. Each case sample in the case sample set includes the patient's basic diagnostic characteristics, the multidisciplinary collaborative treatment plan that has been implemented, and the evaluation results of rehabilitation target indicators. Each discipline dimension corresponds to the corresponding rehabilitation target indicator evaluation results.

[0014] S12: Establish a treatment plan evaluation model framework, which includes a patient feature input layer, a treatment plan parameter input layer, a shared feature layer, a feature interaction fusion layer, and a multidisciplinary result output layer. The patient feature input layer receives basic diagnostic features of the patient. The treatment plan parameter input layer receives structured parameters of the multidisciplinary collaborative treatment plan. The shared feature layer standardizes the input layer content and extracts patient-side shared representations and treatment plan-side shared representations. The feature interaction layer performs conditional modeling and information interaction between patient-side and treatment plan-side shared representations, achieving cross-modal fusion through feature concatenation, gating, and attention mechanisms to form a cross-disciplinary fusion representation for downstream prediction. The multidisciplinary result output layer includes multiple output sub-networks divided by discipline, each output sub-network outputting the expected rehabilitation target indicator evaluation result corresponding to a single discipline dimension based on the cross-disciplinary fusion representation.

[0015] S13: The protocol evaluation model is trained and optimized using a case sample set. The loss function during the training and optimization process satisfies:

[0016]

[0017] Among them, L total ω is the loss function during model training. i Let be the fusion weight for the i-th case sample, N be the total number of case samples in the current training batch, and K be the total number of subject types. Let the standardized training loss of the i-th case sample be the loss in the j-th subject dimension; satisfying:

[0018]

[0019] Among them, l ijLet L() be the mean squared error loss for the i-th case sample in the j-th subject dimension, and let L() be the mean squared error loss function. For the i-th case sample, output the rehabilitation target indicator assessment result for the j-th subject dimension of the model, y ij The evaluation result of the actual rehabilitation target indicator corresponding to the i-th case sample in the j-th subject dimension; μ j Let σ be the mean loss for the j-th subject dimension. j Let be the standard deviation of the loss for the j-th subject dimension;

[0020]

[0021] Wherein, λ is the weight sensitivity coefficient, used to adjust the overall impact of the loss in each discipline on the fusion weight, and is set through pre-experimentation; p is the aggregation index, used to control the sensitivity of the imbalance of the loss in each discipline to the adjustment of the fusion weight, and is set through pre-experimentation.

[0022] Furthermore, the feedback optimization module includes a result receiving and processing unit and a scheme optimization unit; the result receiving and processing unit is used to continuously receive treatment execution feedback information during system operation, and each sample information in the treatment execution feedback information includes: the patient's basic diagnostic characteristics in a single treatment process, the initial multidisciplinary collaborative treatment plan, the individualized treatment plan confirmed by multidisciplinary physicians, and the patient's treatment-related data; the scheme optimization unit updates and optimizes the scheme evaluation model based on the treatment execution feedback information.

[0023] The beneficial effects achieved by this invention are:

[0024] This invention employs a closed-loop design combining intelligent diagnosis, multidisciplinary collaborative planning, process monitoring, and feedback optimization. Centered on a treatment plan evaluation model with patient diagnostic baselines and treatment plan parameters as dual inputs, it first generates candidate treatment plans based on a clinical guideline rule base and individualized hard constraints. Then, it predicts and comprehensively selects the best plan based on multidisciplinary indicators, ensuring that the plan is both individualized and safe and feasible. During the training and operation of the treatment plan evaluation model, interdisciplinary loss standardization and sample fusion weights are introduced to achieve cross-disciplinary, same-scale, fair, and robust learning, reducing selection bias and the impact of outliers. Thus, while meeting adaptability and safety constraints, it improves the accuracy of plan selection and clinical compliance, shortens the recovery period, reduces the risk of relapse and complications, optimizes resource allocation, and significantly enhances the overall accuracy, collaboration, and success rate of the treatment process. Attached Figure Description

[0025] The invention will be further understood from the following description taken in conjunction with the accompanying drawings. The components in the drawings are not necessarily drawn to scale, but rather the emphasis is on illustrating the principles of the embodiments. In different views, the same reference numerals designate corresponding parts.

[0026] Figure 1 This is a schematic diagram of the overall modules of the present invention.

[0027] Figure 2 This is a schematic diagram of the workflow of the initial scheme generation unit of the present invention.

[0028] Figure 3 For this invention, under the condition that λ is set to 0.5, 1, and 2, the independent variable is... Changes in fusion weight ω i A schematic diagram of the influence function.

[0029] Figure 4 This is a schematic diagram of the workflow of the optimization unit in the present invention. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to its embodiments. It should be understood that the specific embodiments described herein are only for explaining the present invention and are not intended to limit the present invention. Other systems, methods, and / or features of this embodiment will become apparent to those skilled in the art after reviewing the following detailed description. It is intended that all such additional systems, methods, features, and advantages are included within this specification, are included within the scope of the present invention, and are protected by the appended claims. Further features of the disclosed embodiments are described in the following detailed description, and these features will be apparent from the following detailed description.

[0031] In the accompanying drawings of the embodiments of the present invention, the same or similar reference numerals correspond to the same or similar components. In the description of the present invention, it should be understood that if terms such as "upper," "lower," "left," "right," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or component referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, the terms used to describe positional relationships in the drawings are only for illustrative purposes and should not be construed as limiting the present patent. For those skilled in the art, the specific meaning of the above terms can be understood according to the specific circumstances.

[0032] In the embodiments of this invention, the letters i, j, c, and d are used only as counting variables or index symbols and have no specific, limited actual meaning.

[0033] Example 1:

[0034] like Figure 1As shown, this embodiment provides an intelligent diagnosis and management system for chronic lateral ankle instability based on multidisciplinary collaboration. The system includes an intelligent diagnosis module, a multidisciplinary collaborative diagnosis and treatment module, a diagnosis and treatment monitoring module, and a feedback optimization module. The intelligent diagnosis module is used to provide basic diagnostic characteristics of patients by combining individualized pathological information. The multidisciplinary collaborative diagnosis and treatment module is used to formulate individualized diagnosis and treatment plans based on the basic diagnostic characteristics of patients through multidisciplinary collaboration. The diagnosis and treatment monitoring module is used to monitor the execution process of the individualized diagnosis and treatment plan. The feedback optimization module is used to optimize the formulation of the individualized diagnosis and treatment plan by combining the feedback information of the diagnosis and treatment execution.

[0035] The intelligent diagnostic module includes a pathological image acquisition unit, a motion function acquisition unit, and a comprehensive feature extraction unit; the pathological image acquisition unit is used to acquire the patient's pathological image information; the motion function acquisition unit is used to acquire the patient's kinematic information; and the comprehensive feature extraction unit is used to extract basic diagnostic features from the pathological image information and kinematic information.

[0036] Specifically, the pathological image acquisition unit acquires pathological image information through a magnetic resonance imaging system, a computed tomography system, and an ultrasound imaging system. The pathological image information includes images of ankle ligaments, cartilage surface conditions, and bony structures.

[0037] Specifically, the motion function acquisition unit collects the patient's kinematic information through wearable sensors, including gait cycle parameters, weight distribution parameters, joint angle change parameters, etc.

[0038] Specifically, the comprehensive feature extraction unit extracts basic diagnostic features through a pre-defined pathological feature extraction model. The pathological feature extraction model includes a comprehensive analysis method based on machine learning algorithms, pattern recognition methods, and multimodal feature fusion technology. The basic diagnostic features include features such as ankle joint stability index, soft tissue injury grade, and kinematic abnormality pattern after vectorization.

[0039] The multidisciplinary collaborative diagnosis and treatment module includes an initial treatment plan generation unit, an interdisciplinary information interaction unit, and an individualized treatment plan generation unit. The initial treatment plan generation unit is used to generate an initial multidisciplinary collaborative diagnosis and treatment plan based on basic diagnostic features. The interdisciplinary information interaction unit is used to realize real-time information sharing among multidisciplinary physicians and to support multidisciplinary physicians in modifying, adjusting, and confirming the initial multidisciplinary collaborative diagnosis and treatment plan. The individualized treatment plan generation unit is used to generate an individualized treatment plan for the patient based on the multidisciplinary collaborative diagnosis and treatment plan confirmed by all multidisciplinary physicians.

[0040] Furthermore, the diagnosis and treatment monitoring module includes a process data acquisition unit, a rehabilitation indicator setting unit, and an effect comparison and analysis unit. The process data acquisition unit is used to collect the patient's diagnosis and treatment-related data during the execution of the individualized diagnosis and treatment plan. The diagnosis and treatment-related data includes kinematic parameters, physiological indicators, and psychological assessment data. The rehabilitation indicator setting unit is used to set the rehabilitation target indicators for the patient during the execution of the individualized diagnosis and treatment plan. The rehabilitation target indicators are pre-set by multidisciplinary physicians based on the patient's individual condition. The effect comparison and analysis unit is used to perform feature processing and comparative analysis on the patient's diagnosis and treatment-related data and rehabilitation target indicators to evaluate the execution effect of the individualized diagnosis and treatment plan and output the differences of various rehabilitation target indicators as monitoring results.

[0041] Furthermore, the initial treatment plan generation unit generates a set of candidate multidisciplinary collaborative treatment plans for the patient by combining the patient's basic diagnostic characteristics and based on the clinical guideline rule base and individualized rigid treatment plan constraints. Then, it calls the pre-established treatment plan evaluation model to evaluate each collaborative treatment plan in the set of multidisciplinary collaborative treatment plans and selects the collaborative treatment plan with the best comprehensive expected rehabilitation goal index evaluation result as the initial multidisciplinary collaborative treatment plan.

[0042] Specifically, the clinical guideline rule base covers consensus on the diagnosis and treatment of chronic lateral ankle instability, rules for surgical procedure selection, rehabilitation stages and intensity boundaries, and the scope of application for psychological intervention and acupuncture treatment; the individualized rigid program constraints are an inviolable set of constraints generated based on the patient's diagnostic characteristics, which includes, but is not limited to: postoperative time window and stage goals, rehabilitation stages and upper and lower limits of training intensity, allowable weight-bearing levels and requirements for brace / protective gear use, pain scores and joint range of motion, etc.

[0043] Furthermore, such as Figure 2 , Figure 3 As shown, the initial scheme generation unit establishes the scheme evaluation model in the following manner:

[0044] S11: Obtain a case sample set through the historical case database. Each case sample in the case sample set includes the patient's basic diagnostic characteristics, the multidisciplinary collaborative treatment plan that has been implemented, and the evaluation results of rehabilitation target indicators. Each discipline dimension corresponds to the corresponding rehabilitation target indicator evaluation results.

[0045] S12: Establish a treatment plan evaluation model framework, which includes a patient feature input layer, a treatment plan parameter input layer, a shared feature layer, a feature interaction fusion layer, and a multidisciplinary result output layer. The patient feature input layer receives basic diagnostic features of the patient. The treatment plan parameter input layer receives structured parameters of the multidisciplinary collaborative treatment plan. The shared feature layer standardizes the input layer content and extracts patient-side shared representations and treatment plan-side shared representations. The feature interaction layer performs conditional modeling and information interaction between patient-side and treatment plan-side shared representations, achieving cross-modal fusion through feature concatenation, gating, and attention mechanisms to form a cross-disciplinary fusion representation for downstream prediction. The multidisciplinary result output layer includes multiple output sub-networks divided by discipline, each output sub-network outputting the expected rehabilitation target indicator evaluation result corresponding to a single discipline dimension based on the cross-disciplinary fusion representation.

[0046] S13: The protocol evaluation model is trained and optimized using a case sample set. The loss function during the training and optimization process satisfies:

[0047]

[0048] Among them, L total ω is the loss function during model training. i Let be the fusion weight for the i-th case sample, N be the total number of case samples in the current training batch, and K be the total number of subject types. Let the standardized training loss of the i-th case sample be the loss in the j-th subject dimension; satisfying:

[0049]

[0050] Among them, L ij Let L() be the mean squared error loss for the i-th case sample in the j-th subject dimension, and let L() be the mean squared error loss function. For the i-th case sample, output the rehabilitation target indicator assessment result for the j-th subject dimension of the model, y ij The evaluation result of the actual rehabilitation target indicator corresponding to the i-th case sample in the j-th subject dimension; μ j Let σ be the mean loss for the j-th subject dimension. j Let be the standard deviation of the loss for the j-th subject dimension;

[0051]

[0052] Wherein, λ is the weight sensitivity coefficient, used to adjust the overall impact of the loss in each discipline on the fusion weight, and is set through pre-experimentation; p is the aggregation index, used to control the sensitivity of the imbalance of the loss in each discipline to the adjustment of the fusion weight, and is set through pre-experimentation.

[0053] This approach constructs a protocol evaluation model with patient diagnostic baseline characteristics and protocol parameters as dual inputs. It comprehensively evaluates and selects candidate multidisciplinary collaborative treatment protocols, thereby obtaining the optimal initial multidisciplinary collaborative treatment protocol while meeting adaptability and safety constraints. By introducing interdisciplinary loss standardization and sample fusion weights during model training, it achieves fair comparison and robust learning of cross-disciplinary indicators at the same scale, suppresses the influence of noise and biased samples, overcomes the limitations of traditional single-disciplinary evaluation, scientifically quantifies the comprehensive potential of protocols, provides objective evidence for clinical decision-making, and helps select the individualized protocol most likely to achieve the best overall recovery outcome for the patient. This significantly improves the accuracy, synergy, and success rate of the treatment process.

[0054] Example 2:

[0055] This embodiment should be understood to include at least all the features of any of the foregoing embodiments, and to further improve upon them;

[0056] This embodiment provides an intelligent diagnosis and management system for chronic lateral ankle instability based on multidisciplinary collaboration. The system includes an intelligent diagnosis module, a multidisciplinary collaborative diagnosis and treatment module, a diagnosis and treatment monitoring module, and a feedback optimization module. The intelligent diagnosis module is used to provide basic diagnostic characteristics of patients by combining individualized pathological information. The multidisciplinary collaborative diagnosis and treatment module is used to formulate individualized diagnosis and treatment plans based on the basic diagnostic characteristics of patients through multidisciplinary collaboration. The diagnosis and treatment monitoring module is used to monitor the execution process of the individualized diagnosis and treatment plan. The feedback optimization module is used to optimize the formulation of the individualized diagnosis and treatment plan by combining the feedback information of the diagnosis and treatment execution.

[0057] Furthermore, the feedback optimization module includes a result receiving and processing unit and a scheme optimization unit; the result receiving and processing unit is used to continuously receive treatment execution feedback information during system operation, and each sample information in the treatment execution feedback information includes: the patient's basic diagnostic characteristics in a single treatment process, the initial multidisciplinary collaborative treatment plan, the individualized treatment plan confirmed by multidisciplinary physicians, and the patient's treatment-related data; the scheme optimization unit updates and optimizes the scheme evaluation model based on the treatment execution feedback information;

[0058] Furthermore, such as Figure 4 As shown, the specific workflow of the scheme optimization unit is as follows:

[0059] S21: During system operation, continuously collect treatment execution feedback information. When the sample information in the treatment execution feedback information reaches the preset number of update samples, use the sample information in the current treatment execution feedback information as the update sample set to update and optimize the solution evaluation model.

[0060] S22: For each piece of information in the updated sample set, calculate its residual difference across each subject dimension:

[0061]

[0062] Where, r ij For the residual difference of the i-th updated sample in the j-th subject dimension, The rehabilitation target indicator for the i-th updated sample in the j-th subject dimension is obtained by updating the diagnosis and treatment-related data in the sample. The model outputs rehabilitation target indicators for the i-th updated sample in the j-th subject dimension, which are obtained by inputting the patient's basic diagnostic characteristics in the updated sample and the individualized treatment plan confirmed by multidisciplinary physicians into the current plan evaluation model;

[0063] S23: Apply standardized loss to the residual differences for each discipline dimension, satisfying:

[0064]

[0065] in, Let the residual difference of the i-th updated sample be the standardized difference along the j-th subject dimension. This represents the mean residual difference in the j-th subject dimension of the updated sample set. This represents the standard deviation of the residual differences along the j-th subject dimension in the updated sample set. Let ρ() be the standardized loss function for the i-th updated sample in the j-th subject dimension, and let ρ() be the Huber loss.

[0066] S24: Iteratively update the current scheme evaluation model using the following update loss function:

[0067]

[0068] Among them, L update For the incremental update loss function of the scheme evaluation model, N new This represents the number of samples in the current updated sample set. The update weight for the i-th update sample satisfies:

[0069]

[0070] Among them, D disc D represents the discrete set of parameters for the treatment plan in the updated sample; cont This is a continuous set of parameters for the treatment plan in the updated sample; Let be the initial value of the multidisciplinary collaborative treatment plan for the i-th update sample in the discrete parameter dimension d. Let be the value of the individualized treatment plan for the i-th updated sample after confirmation by multidisciplinary physicians on the discrete parameter dimension d, and let 1[] be the indicator function, which takes 1 when the condition is met and 0 otherwise; Let the initial multidisciplinary collaborative treatment plan value be taken for the i-th updated sample along the continuous parameter dimension c. Let range be the value of the individualized treatment plan for the i-th updated sample, confirmed by multidisciplinary physicians, along the continuous parameter dimension c. c The normalization interval width for the continuous parameter dimension c is set through prior experiments; norm() is the normalization processing function, which can be scaled using the theoretical upper limit.

[0071] This solution uses the residual difference between actual rehabilitation target indicators and model-predicted rehabilitation target indicators as an optimization signal during system operation, and performs batch-standardized loss processing on residuals from each discipline to achieve cross-disciplinary, same-scale, and anomaly-resistant model correction. By converting the degree of doctor modifications to treatment plans in the samples into update weights, samples with greater doctor modifications receive higher learning strength, thereby quickly absorbing first-hand revision experience in actual use and correcting historical biases. Through a batch-triggered incremental update mechanism, the predictive accuracy and calibration of the treatment plan evaluation model are continuously improved, providing more individualized and collaborative treatment plan decisions while meeting adaptability and safety constraints.

[0072] Example 3:

[0073] This embodiment uses a national-level sports rehabilitation medicine center as the application scenario. This center is among the first batch of approved multidisciplinary collaborative rehabilitation demonstration bases in China, treating over 15,000 sports injury patients annually, of whom approximately 8.6% are chronic lateral ankle instability. To verify the practical application effect of this system, this embodiment selects 120 confirmed patients as the pilot group. The average age of the patients is 26 years old, with athletes accounting for 25% and the general population accounting for 75%. The pilot period is set at 12 months, covering three stages: preoperative assessment, postoperative rehabilitation, and follow-up management. The system adopts a distributed deployment mode, with front-end data acquisition devices wirelessly connecting to the hospital's data center, and the back-end supported by a cluster server for computing and analysis.

[0074] The pathology imaging unit uses a Siemens three-feature magnetic resonance imaging system, capable of imaging ligaments, cartilage, and bony structures at a resolution of 0.5 mm; CT acquisition uses a Philips 64-slice spiral scanner, capable of sub-millimeter-level three-dimensional reconstruction, displaying joint space narrowing and osteophyte formation. Ultrasound acquisition is performed using a Mindray high-frequency probe, with a real-time resolution of 0.3 mm, supporting ligament laxity detection under dynamic weight-bearing conditions;

[0075] The motion function acquisition unit uses the Noraxon multichannel electromyography system and Vicon three-dimensional motion capture system from Norway. The gait cycle error is controlled within 0.1 degrees and the weight distribution error is not more than 1%. In the gait test, the varus angle of the affected side of the test patients fluctuated by 35% more than that of the healthy side. The average activation delay of the peroneus brevis muscle was 150 milliseconds, while that of the healthy control group was less than 50 milliseconds, and the difference was significant.

[0076] The integrated feature extraction unit uses a deep learning-based ResNet image analysis network and a random forest kinematic feature model to work together, which can transform imaging and kinematic information into feature indicators such as ankle stability index, soft tissue injury grade, and joint dysfunction pattern. The system was backtested on 120 patient samples and the diagnostic accuracy reached 95.2%, which is a significant improvement compared to 84.5% of single imaging diagnosis.

[0077] In the initial treatment plan generation, the multidisciplinary collaborative diagnosis and treatment module calls the built-in clinical guideline rule base and individualized constraints to generate candidate plans. The rule base covers surgical indications, rehabilitation training intensity grading, psychological intervention process and indications for traditional Chinese medicine acupuncture treatment. The system automatically generates an average of 4.2 candidate plans per patient in the pilot group. After scoring by the plan evaluation model, the optimal plan is selected as the initial treatment path.

[0078] The interdisciplinary information exchange unit is deployed on the hospital's multidisciplinary consultation platform, supporting simultaneous online modifications of treatment plans by orthopedics, rehabilitation medicine, sports medicine, psychology, and traditional Chinese medicine departments. Information exchange latency is less than 100 milliseconds, ensuring real-time communication among doctors. In actual use, the multidisciplinary team can complete a consensus on a patient's treatment plan in an average of 30 minutes, reducing efficiency by approximately 46% compared to traditional face-to-face joint outpatient clinics.

[0079] After the doctor confirms the individualized treatment plan, the system outputs the final treatment plan. The system can automatically generate a patient rehabilitation stage table and a schedule task sheet based on the final treatment plan. Taking a patient six weeks after surgery as an example, the individualized plan clarifies the weight-bearing level increase curve, joint range of motion target, frequency of psychological intervention and rhythm of traditional Chinese medicine acupuncture intervention.

[0080] The process data acquisition of the diagnosis and treatment monitoring module is completed through a wearable inertial measurement unit and a smart foot pressure insole; the inertial measurement unit has a sampling frequency of 200 Hz, and the pressure insole has a resolution of 16 points per square centimeter, which can record the patient's gait cycle in real time; the patient's data is uploaded to the cloud in real time during rehabilitation training, with a delay of no more than two seconds; the rehabilitation indicators are set in advance by the doctor, including a pain score reduction of more than 50%, an increase in ankle joint range of motion of more than 20 degrees, and an increase in the distance of the six-minute walking test of 30%, etc.; the effect comparison and analysis unit calculates the difference based on an adaptive threshold algorithm and outputs a graphical trend curve.

[0081] In this embodiment, the feedback optimization module adopts an incremental online learning method. The system continuously receives data samples from patients during their rehabilitation process, and triggers a model update when the cumulative feedback reaches fifty cases. During the optimization process, the extent of the doctor's adjustments to the plan is quantified as a weighting factor and participates in the retraining of the plan evaluation model. After running continuously for twelve months, the average deviation between the model's predicted rehabilitation goal achievement rate and the actual results decreased from the initial 14.3% to 6.8%, and the accuracy was significantly improved.

[0082] Regarding the implementation results, among 120 pilot patients, the system management group of this embodiment was compared with the traditional single-discipline treatment group. The results showed that the average recovery period of the system management group was shortened by three weeks, and the relapse rate was reduced to 8.5%, while that of the control group was 15.7%. The average patient compliance score of the system management group was 9.1 (out of 10), which was significantly higher than that of the control group (7.8). At the same time, the compliance with psychological intervention increased by more than 20%, and the patient satisfaction score reached 96.3%.

[0083] Regarding the substitution of technical methods, image acquisition can be replaced by a combination of MRI and computed tomography with ultrasound and low-dose X-rays. Although the imaging accuracy decreases, the cost is lower, making it suitable for grassroots applications. Motion function acquisition can be replaced by single-camera AI gait analysis, simplifying hardware deployment but reducing accuracy by about 8% to 10% in complex motion recognition. The scheme evaluation model can be replaced by a deep neural network with a multi-classification model based on support vector machines, which has a faster inference speed but limited cross-disciplinary feature fusion capabilities. The feedback optimization module can be replaced by batch offline updates instead of online updates, which is simpler to implement but less adaptable to real-time conditions than online methods.

[0084] In summary, this embodiment, by introducing high-precision imaging equipment, kinematic acquisition systems, and interdisciplinary information platforms into a real medical scenario, and combining quantitative data and comparative analysis, verifies the advantages of the system in shortening the treatment cycle, improving compliance, and reducing recurrence rates. It provides a complete and feasible technical path for the intelligent, individualized, and multidisciplinary collaborative management of chronic lateral ankle instability.

[0085] The content disclosed above is only a preferred and feasible embodiment of the present invention, and is not intended to limit the scope of protection of the present invention. Therefore, all equivalent technical changes made based on the content of the present invention specification and drawings are included within the scope of protection of the present invention. Furthermore, the elements therein can be updated as technology develops.

Claims

1. A chronic lateral ankle instability intelligent diagnosis and treatment management system based on multidisciplinary collaboration, characterized in that, The system comprises an intelligent diagnosis module, a multidisciplinary collaborative diagnosis and treatment module, a diagnosis and treatment monitoring module and a feedback optimization module; the intelligent diagnosis module is used to provide patient diagnosis basic features in combination with individualized pathological information of the patient; The multidisciplinary collaborative diagnosis and treatment module is used to realize multidisciplinary collaborative individualized diagnosis and treatment scheme making based on the patient diagnosis basic features; the diagnosis and treatment monitoring module is used to monitor the execution process of the individualized diagnosis and treatment scheme; and the feedback optimization module is used to feedback optimize the making of the individualized diagnosis and treatment scheme in combination with diagnosis and treatment execution feedback information; The intelligent diagnosis module comprises a pathological image acquisition unit, a motor function acquisition unit and a comprehensive feature extraction unit; the pathological image acquisition unit is used to acquire pathological image information of the patient; the motor function acquisition unit is used to acquire kinematic information of the patient; and the comprehensive feature extraction unit is used to extract diagnosis basic features from the pathological image information and the kinematic information; The multidisciplinary collaborative diagnosis and treatment module comprises an initial scheme generation unit, a cross-disciplinary information interaction unit and an individualized scheme generation unit; the initial scheme generation unit is used to generate an initial multidisciplinary collaborative diagnosis and treatment scheme based on the diagnosis basic features; the cross-disciplinary information interaction unit is used to realize real-time information sharing among multidisciplinary physicians and support change adjustment and confirmation of the initial multidisciplinary collaborative diagnosis and treatment scheme by the multidisciplinary physicians; and the individualized scheme generation unit is used to generate the multidisciplinary collaborative diagnosis and treatment scheme confirmed by all the multidisciplinary physicians as the individualized diagnosis and treatment scheme of the patient; The initial scheme generation unit generates a candidate multidisciplinary collaborative diagnosis and treatment scheme set related to the patient in combination with the diagnosis basic features of the patient according to a clinical guideline rule base and individualized hard scheme constraints, and then calls a pre-established scheme evaluation model to evaluate each collaborative diagnosis and treatment scheme in the multidisciplinary collaborative diagnosis and treatment scheme set, and selects a collaborative diagnosis and treatment scheme with the optimal comprehensive expected rehabilitation target index evaluation result as the initial multidisciplinary collaborative diagnosis and treatment scheme; The initial scheme generation unit specifically establishes the scheme evaluation model in the following manner: S11: Obtain a case sample set from a historical case library, wherein each case sample in the case sample set comprises patient diagnosis basic features, an executed multidisciplinary collaborative diagnosis and treatment scheme and a rehabilitation target index evaluation result; and each discipline dimension corresponds to a corresponding rehabilitation target index evaluation result; S12: a scheme evaluation model framework is established, the scheme evaluation model framework comprising a patient feature input layer, a scheme parameter input layer, a shared feature layer, a feature interaction fusion layer, and a multidisciplinary result output layer; the patient feature input layer is used to receive patient diagnosis basic features; the scheme parameter input layer is used to receive structured parameters of the multidisciplinary collaborative diagnosis and treatment scheme; the shared feature layer is used to standardize the content of the input layer, and extract patient-side shared representation and scheme-side shared representation respectively; the feature interaction layer is used to conditionally model and interact information between the patient-side shared representation and the scheme-side shared representation, to realize cross-modal fusion through feature splicing, gating, and attention mechanism, and to form a cross-disciplinary fusion representation for downstream prediction; the multidisciplinary result output layer comprises a plurality of output sub-networks divided according to disciplines, and each output sub-network outputs an expected rehabilitation target index evaluation result of a corresponding single discipline dimension based on the cross-disciplinary fusion representation; S13: the scheme evaluation model is trained and optimized using a case sample set, and a loss function of the training and optimization process satisfies: ； wherein, is a loss function in the model training process, is a fusion weight of the th case sample, is the total number of case samples in the current training batch, is the total number of subject types, is the standardized training loss of the th case sample in the th subject dimension; satisfies: ； ； in, For the first The first case sample was in the Mean squared error loss across each subject dimension Let the mean squared error loss function be . For the first The first case sample was in the The model outputs rehabilitation target indicator assessment results for each subject dimension. For the first The first case sample was in the Evaluation results of actual rehabilitation goal indicators corresponding to each subject dimension; For the first The mean loss across each subject dimension, For the first Standard deviation of loss for each subject dimension; ； wherein, is a weight sensitivity coefficient, used to adjust the overall influence of the loss on each subject on the fusion weight, set by pre-experiment; is a polymerization index, used to control the sensitivity of the imbalance of the loss of each subject to the adjustment of the fusion weight, set by pre-experiment; The feedback optimization module comprises a result receiving and sorting unit and a scheme optimization unit; the result receiving and sorting unit is used to continuously receive diagnosis and treatment execution feedback information in the system running process, and each sample information in the diagnosis and treatment execution feedback information comprises patient diagnosis basic features in a single diagnosis and treatment process, an initial multidisciplinary collaborative diagnosis and treatment scheme, an individualized diagnosis and treatment scheme confirmed by multidisciplinary physicians, and diagnosis and treatment related data of the patient; the scheme optimization unit updates and optimizes the scheme evaluation model based on the diagnosis and treatment execution feedback information; The specific workflow of the scheme optimization unit is as follows: S21: continuously collect diagnosis and treatment execution feedback information in the system running process, and when the sample information in the diagnosis and treatment execution feedback information reaches a preset update sample quantity, use the sample information in the current diagnosis and treatment execution feedback information as an update sample set to update and optimize the scheme evaluation model; S22: for each sample information in the update sample set, calculate the residual difference in each discipline dimension: ； in, For the first The updated sample is in the first Residual differences across discipline dimensions For the first The updated sample is in the first Rehabilitation target indicators across various disciplines were obtained by updating treatment-related data in the sample. For the first The updated sample is in the first The model outputs rehabilitation target indicators across multiple disciplines, which are obtained by inputting the patient's basic diagnostic characteristics and the individualized treatment plan confirmed by multidisciplinary physicians into the current plan evaluation model. S23: standardize the residual difference in each discipline dimension to satisfy: ； ； in, For the first The updated sample is in the first Standardized residual differences across various disciplinary dimensions This is the first batch of updated sample sets. Mean residual variance across each disciplinary dimension This is the first batch of updated sample sets. Standard deviation of residual variance across each subject dimension For the first The updated sample is in the first Standardized loss functions across various disciplines Handling Huber losses; S24: iteratively update the current scheme evaluation model by updating the loss function as follows: ； wherein, is the incremental update loss function of the scheme evaluation model, is the number of samples in the current update sample set, is the update weight of the i-th update sample, satisfying: ； in, This is the set of discrete parameters for the treatment plan in the updated sample; This is a continuous set of parameters for the treatment plan in the updated sample; For the first Update samples in discrete parameter dimension The initial values ​​of the multidisciplinary collaborative treatment plan, For the first Update samples in discrete parameter dimension The values ​​of the individualized treatment plan confirmed by multidisciplinary physicians are as follows. This is an indicator function that takes the value 1 if the condition is true and 0 otherwise. For the first Update samples in continuous parameter dimensions The initial values ​​of the multidisciplinary collaborative treatment plan, For the first Update samples in continuous parameter dimensions The values ​​of the individualized treatment plan confirmed by multidisciplinary physicians are as follows. For continuous parameter dimensions The normalized interval width is set through prior experiments; This is a normalization function that can be scaled using a theoretical upper limit.

2. The intelligent diagnosis and treatment management system for chronic lateral ankle instability based on multidisciplinary collaboration according to claim 1, characterized in that, The diagnosis and treatment monitoring module comprises a process data acquisition unit, a rehabilitation index setting unit, and an effect comparison and analysis unit; the process data acquisition unit is used to acquire diagnosis and treatment related data of the patient in the process of executing the individualized diagnosis and treatment scheme, the diagnosis and treatment related data comprising kinematic parameters, physiological indexes, and psychological evaluation data; the rehabilitation index setting unit is used to set rehabilitation target indexes of the patient in the process of executing the individualized diagnosis and treatment scheme, the rehabilitation target indexes being pre-set by multidisciplinary physicians according to the individual condition of the patient; the effect comparison and analysis unit is used to perform feature processing and comparison analysis on the diagnosis and treatment related data of the patient and the rehabilitation target indexes to evaluate the execution effect of the individualized diagnosis and treatment scheme, and output the differences in each rehabilitation target index as a monitoring result.

Images