Method for constructing prediction model of rheumatoid arthritis accompanied with brain cognitive impairment

Abstract

The invention discloses a prediction model construction method for rheumatoid arthritis accompanied with brain cognitive impairment, and belongs to the technical field of medical assistance, and the method specifically comprises the steps: synchronously collecting multi-source dynamic behavior data when a patient executes standardized joint activity training each time; vector superposition comparison under a three-dimensional coordinate system is carried out on the joint motion trajectory data, and the trajectory offset degree of each motion phase point is calculated; fractal dimension calculation is carried out on the moving path topological structure data, and path complexity is quantized; performing feature matrix construction on the facial micro-expression data, and calculating covariance structure change of an expression parameter matrix; and performing time domain alignment and causal analysis on the spatio-temporal trajectory offset event, the path topology degradation characteristic and the expression parameter variation signal, and when the three types of abnormalities present time coupling in two continuous training periods and meet a preset condition, generating early warning of cognitive impairment.

Description

Technical Field

[0001] The present invention relates to the field of medical assistance technology, and in particular to a method for constructing a prediction model for rheumatoid arthritis-associated brain cognitive dysfunction. Background Art

[0002] Rheumatoid arthritis (RA) is a common chronic autoimmune disease that seriously affects the joint function of patients, causing joint pain, swelling, and deformity, greatly reducing the quality of life of patients. With the deepening of research on RA, it is found that it not only affects the joints, but is also often accompanied by a variety of extra-articular manifestations. Brain cognitive dysfunction is one of them. Studies have shown that a considerable proportion of RA patients will experience varying degrees of cognitive impairment during the course of the disease, including memory loss, inattention, executive dysfunction, etc., which further increases the physical and mental burden of patients and has a serious impact on their daily life, social and work abilities.

[0003] In the prior art, research on RA-associated brain cognitive dysfunction has mostly focused on the diagnosis and treatment of the disease. Clinicians mainly rely on the patient's clinical manifestations, imaging examinations, and some neuropsychological tests to determine whether the patient has cognitive dysfunction, but these methods can often only make a diagnosis when the cognitive dysfunction is already more obvious, and lack early prediction methods. At present, joint movement training for RA patients is an important rehabilitation method to improve joint function. However, during the training process, monitoring of the risk of cognitive dysfunction that may occur in patients is relatively scarce. Although some studies have attempted to explore the association with cognitive function through a single data collection method, such as only monitoring joint movement data or only paying attention to changes in the patient's facial expressions, this single-dimensional data collection method is difficult to fully reflect the patient's complex physical state changes and cannot accurately predict the occurrence of cognitive dysfunction.

[0004] Existing technologies have obvious deficiencies in the early prediction of RA-associated brain cognitive dysfunction. There is a lack of systematic collection and comprehensive analysis of multi-source dynamic behavioral data during patients' daily rehabilitation training, making it difficult to capture early subtle signals of physical changes. Summary of the invention

[0005] The purpose of the present invention is to provide a method for constructing a prediction model for rheumatoid arthritis associated with brain cognitive dysfunction, and to solve the following technical problems:

[0006] Existing technologies for predicting brain cognitive dysfunction associated with RA have difficulty capturing early and subtle signals of physical changes.

[0007] The purpose of the present invention can be achieved through the following technical solutions:

[0008] A method for constructing a prediction model for rheumatoid arthritis complicated with cerebral cognitive dysfunction, comprising the following steps:

[0009] When a patient performs standardized joint activity training each time, multi-source dynamic behavior data is synchronously collected, including joint activity trajectory data, mobile path topological structure data, and facial micro-expression data;

[0010] Perform vector superposition comparison on the joint activity trajectory data in a three-dimensional coordinate system, calculate the trajectory deviation degree of each action phase point, and perform dynamic comparison with the patient's own historical baseline trajectory, and the historical baseline trajectory is generated from the training data before a set duration of the patient;

[0011] Calculate the fractal dimension of the mobile path topological structure data, quantify the path complexity, and analyze the change trend of the path fractal dimension value. The fractal dimension calculation uses the box counting method, and the complexity of the path is evaluated by counting the number of grids covered by the path;

[0012] Construct a feature matrix for the facial micro-expression data, and calculate the covariance structure change of the expression parameter matrix. The feature matrix includes data in three dimensions: the contraction frequency of the frontalis muscle, the opening and closing rate of the eyelids, and the tremor amplitude of the orbicularis oris muscle;

[0013] Based on the processed data, establish a multi-dimensional abnormal association model, perform time domain alignment and causal analysis on spatio-temporal trajectory deviation events, path topological degradation features, and expression parameter variation signals. When the three types of abnormalities show time coupling in two consecutive training cycles and meet the preset conditions, generate an early warning for cognitive dysfunction.

[0014] As a further solution of the present invention: The collection of multi-source dynamic behavior data includes:

[0015] Real-time monitor the spatio-temporal feature data of the large joint activities of the limbs through a wearable joint angle sensor, and record the joint activity angle, movement speed, and acceleration;

[0016] Continuously record the topological structure data of the patient's movement path through a plantar pressure distribution sensor and an indoor positioning device, including gait characteristics, path length, number of turning points, and spatial positioning accuracy;

[0017] Capture the dynamic change data of the patient's facial micro-expression through a head-mounted miniature camera, and focus on recording the contraction frequency of the frontalis muscle, the opening and closing rate of the eyelids, and the tremor amplitude of the orbicularis oris muscle.

[0018] As a further solution of the present invention: The standardized joint activity training includes a pre-set repetitive composite action sequence, and the action sequence is designed to simulate functional activities in daily life, including upper limb extension, lower limb walking, and balance control. The execution duration, intensity, and environmental conditions of each action sequence are kept consistent.

[0019] As a further solution of the present invention: The dynamic comparison of joint movement trajectory data specifically includes:

[0020] Perform vector superposition comparison in a three-dimensional coordinate system for each joint movement trajectory within the current training cycle and the patient's own historical baseline trajectory, calculate the trajectory deviation degree of each action phase point, and when the spatial positioning deviation at the end of elbow flexion exceeds the preset dynamic threshold for three consecutive trainings, trigger an abnormal movement pattern mark; the preset dynamic threshold is dynamically adjusted according to the patient's age, disease course, and joint movement ability.

[0021] As a further solution of the present invention: The mobile path topological structure analysis uses the fractal dimension calculation method to quantify the complexity of the patient's movement trajectory in the same training environment. When the path fractal dimension value drops by more than the threshold compared to the previous training and is accompanied by an abnormal increase in the number of turning points, it is determined as a feature of degraded spatial navigation ability; the abnormal increase in the number of turning points refers to the number of turning points exceeding a% of the historical average, where a is a set value.

[0022] As a further solution of the present invention: The dynamic analysis of facial micro-expressions captures three biometric parameters, namely the contraction frequency of the frontalis muscle, the opening and closing rate of the eyelids, and the tremor amplitude of the orbicularis oris muscle, by establishing an expression feature matrix exclusive to the patient when performing complex action sequences. When a significant change occurs in the covariance structure of the expression parameter matrix, an abnormal emotion expression signal is generated; the significant change refers to the eigenvalue of the covariance matrix changing by more than b% of the historical average, where b is a set value.

[0023] As a further solution of the present invention: Through vector superposition comparison in a three-dimensional coordinate system, calculate the difference between the patient's current joint movement trajectory and the historical baseline trajectory to obtain the trajectory deviation degree, and quantify the deviation degree in units of standard deviation σ; use the fractal dimension calculation method to quantify the complexity of the patient's movement path, calculate the difference between the current path fractal dimension value and the historical baseline value to obtain the path fractal dimension reduction; calculate the coefficient of variation of the dynamic change data of facial micro-expressions to obtain the fluctuation degree of the expression parameters, where the coefficient of variation is the ratio of the standard deviation to the mean; when the values of the trajectory deviation degree, the path fractal dimension reduction, and the fluctuation degree of the expression parameters all exceed the normal range, mark them as spatio-temporal trajectory deviation events, path topological degradation features, and expression parameter variation signals. Align the spatio-temporal trajectory deviation events, path topological degradation features, and expression parameter variation signals on the time axis, and analyze whether the three types of abnormalities show temporal coupling within two consecutive training cycles. The temporal coupling refers to the time interval between the occurrences of the three types of abnormalities within the same training cycle not exceeding the set duration. When the three types of abnormalities meet the determination conditions and show temporal coupling within two consecutive training cycles, an early warning of cognitive dysfunction is generated.

[0024] As a further solution of the present invention: Design a cognitive-motor coupling degree evaluation index, which is weighted and composed of three dimensions: action trajectory fidelity, environmental adaptation coefficient, and emotional feedback sensitivity. When the index value shows a monotonically decreasing trend for three consecutive training cycles and the slope exceeds the preset critical value, start a hierarchical early warning mechanism, and dynamically adjust the determination threshold of each abnormal parameter according to the patient's age, disease course, and previous cognitive evaluation results; the hierarchical early warning mechanism includes three levels: low risk, medium risk, and high risk.

[0025] Advantages of the present invention:

[0026] By synchronously collecting multi-source dynamic behavior data such as joint movement trajectories, mobile path topologies, and facial micro-expressions, and through processing such as vector superposition comparison, fractal dimension calculation, and feature matrix construction, the present invention establishes a multi-dimensional abnormal association model. It can accurately capture abnormalities such as spatio-temporal trajectory deviation, path topology degradation, and expression parameter variation at an early stage, generate an early warning of cognitive dysfunction based on time coupling, and also designs a cognitive-motor coupling degree evaluation index and a hierarchical early warning mechanism, which can dynamically adjust the determination threshold, and is of great significance for the early prediction, intervention, and disease management of rheumatoid arthritis accompanied by brain cognitive dysfunction, effectively improving the quality of life of patients. Brief Description of the Drawings

[0027] The present invention will be further described below with reference to the drawings.

[0028] Figure 1 is a schematic flowchart of the present invention. Detailed Embodiments

[0029] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0030] Please refer to Figure 1 as shown, the present invention is a method for constructing a prediction model for rheumatoid arthritis accompanied by brain cognitive dysfunction:

[0031] Example 1

[0032] 1. When the patient performs standardized joint movement training each time, use advanced and diverse data collection technologies to synchronously collect multi-source dynamic behavior data, including joint movement trajectory data, mobile path topology data, and facial micro-expression data.

[0033] In terms of collecting joint movement trajectory data, a wearable joint angle sensor is used. This sensor has high-precision real-time monitoring capabilities and can comprehensively track the spatio-temporal characteristic data of the movement trajectories of the large joints of the limbs. It can not only accurately record the joint movement angles, down to every degree of change, but also precisely measure the movement speed, capture the subtle fluctuations in speed during joint movement, as well as the dynamic changes in acceleration, thus fully presenting the movement state of the joint at each movement stage.

[0034] For collecting mobile path topological structure data, a method of collaborative operation between a plantar pressure distribution sensor and an indoor positioning device is adopted. The plantar pressure distribution sensor can continuously and accurately record the pressure distribution of each part of the sole during the patient's walking process, and then analyze gait characteristics, such as step size, walking frequency, and the transfer of the pressure center of gravity when both feet touch the ground. At the same time, the indoor positioning device continuously tracks the patient's movement path through advanced positioning algorithms, accurately obtains the path length, and can accurately measure both the straight-line walking distance and the actual length of the curved movement. The number of turning points is also accurately recorded to reflect the direction change of the patient during movement. In addition, the spatial positioning accuracy can reach the millimeter level, ensuring that the subtle position changes of the patient's movement in the indoor space can be captured, and comprehensively quantifying the topological structure data of the patient's movement path.

[0035] The dynamic changes in facial micro-expressions are captured by a head-mounted micro-camera. This camera focuses on the key facial expression muscle groups of the patient, and mainly records the contraction frequency of the frontalis muscle to reflect changes such as tension and anxiety in the patient's emotional state; precisely monitors the opening and closing rate of the eyelids, as the rapid or slow opening and closing of the eyelids can often convey different emotional reactions and attention levels of the patient; and the tremor amplitude of the orbicularis oris muscle, as the subtle tremors of the orbicularis oris muscle can reflect potential emotional fluctuations or changes in the patient's muscle control ability. Through the careful collection of these dynamic facial micro-expression data, rich data support is provided for analyzing the patient's emotional state and potential cognitive changes.

[0036] 2.1 Processing and analysis of joint movement trajectory data:

[0037] For the collected joint movement trajectory data, advanced mathematical analysis methods are used to perform vector superposition and comparison in a three-dimensional coordinate system to deeply explore the accuracy of the patient's joint movement. Specifically, first calculate the trajectory deviation degree of each movement phase point, which can accurately reflect the deviation degree of the patient's current joint movement trajectory compared with the standard or expected trajectory. To more accurately evaluate the changes in the patient's joint movement, the currently calculated trajectory deviation degree is dynamically compared with the patient's own historical baseline trajectory. The historical baseline trajectory here is determined by collecting the patient's training data before a set time period, and it represents the relatively stable joint movement pattern of the patient in the past period.

[0038] In actual operation, the joint movement trajectories within the current training cycle and the historical baseline trajectories are placed in a three-dimensional coordinate system for vector superposition comparison. This comparison method in the three-dimensional coordinate system can comprehensively and accurately capture the movement changes of joints in space, avoiding the possible information loss in two-dimensional analysis. After calculating the trajectory deviation degrees of each action phase point, the spatial positioning deviation at the end of elbow flexion is focused on. When this deviation exceeds the preset dynamic threshold for three consecutive trainings, the system will automatically trigger the marking of abnormal movement patterns. The triggering of this marking is of great significance as it can timely remind medical staff to pay attention to the possible joint movement abnormalities of patients. The preset dynamic threshold is not fixed but is dynamically adjusted according to the patient's age, disease course, and joint movement ability. Patients in different age groups have different physiological structures and movement abilities of joints; the length of the disease course will also affect the degree of joint damage and recovery; the joint movement ability varies from person to person. Therefore, dynamically adjusting the preset threshold can ensure that the judgment of abnormal movement patterns is more accurate and reasonable, in line with the actual situation of each patient.

[0039] It is worth mentioning that the standardized joint movement training carried out by patients includes pre-set repetitive composite action sequences. These action sequences are ingeniously designed to simulate functional activities in daily life, such as upper limb extension, which is necessary for actions like picking up objects and opening doors in daily life; lower limb walking, which is the most basic way of movement; and balance control, which is crucial for maintaining body stability and avoiding falls. Moreover, to ensure the comparability of data and the accuracy of analysis results, the execution duration, intensity, and environmental conditions of each action sequence are kept consistent. In this way, the data collected in different training cycles can more truly reflect the changes in the patient's joint movement ability, providing a reliable basis for subsequent analysis and judgment.

[0040] 2.2 Mobile path topological structure data processing and analysis:

[0041] For the collected mobile path topological structure data, a method of calculating the fractal dimension is used to quantify the path complexity and deeply analyze the changing trend of the path fractal dimension value. The fractal dimension is a mathematical concept used to describe the irregularity degree of complex geometric objects. In this study, calculating the fractal dimension can more accurately evaluate the complexity of the patient's mobile path. The box-counting method is specifically adopted, which is a classic and effective method for calculating the fractal dimension. Its principle is to place the patient's mobile path on a plane composed of grids and evaluate the complexity of the path by counting the number of grids covered by the path. The more grids the path covers, the more complex the path is, and the higher the fractal dimension value; conversely, the simpler the path is, the lower the fractal dimension value.

[0042] During the actual analysis process, the complexity of the patient's movement trajectory in the same training environment is quantified. When the path fractal dimension value drops by more than the threshold compared to the previous training and is accompanied by an abnormal increase in the number of turning points, it is determined as a characteristic of spatial navigation ability degradation. An abnormal increase in the number of turning points means that the number of turning points exceeds a% (a is a set value) of the historical average. Spatial navigation ability is an important ability for humans to locate, plan routes, and reach destinations in the environment. For patients with rheumatoid arthritis, the degradation of spatial navigation ability may be related to brain cognitive dysfunction. Therefore, by monitoring the changes in the path fractal dimension value and the number of turning points, the potential changes in the patient's spatial navigation ability can be detected in a timely manner, providing important clues for early warning and intervention.

[0043] 2.3 Facial micro-expression data processing and analysis:

[0044] When processing facial micro-expression data, the information contained therein is deeply mined by constructing a feature matrix. This feature matrix includes data in three dimensions: the contraction frequency of the frontalis muscle, the opening and closing rate of the eyelids, and the tremor amplitude of the orbicularis oris muscle. These three biometric parameters can reflect the patient's emotional state and potential cognitive changes from different perspectives. Changes in the contraction frequency of the frontalis muscle may be related to emotions such as tension and anxiety in the patient; the speed of the opening and closing of the eyelids can reflect the patient's degree of attention concentration and emotional arousal level; the tremor amplitude of the orbicularis oris muscle may reflect changes in the patient's emotional fluctuations or muscle control ability.

[0045] In the specific analysis process, by establishing an exclusive expression feature matrix for the patient, the changes in these three biometric parameters of the patient during the execution of complex action sequences are mainly captured. Then, the covariance structure change of the expression parameter matrix is calculated. The covariance structure can reflect the correlation and change trend between different biometric parameters. When the covariance structure of the expression parameter matrix undergoes a significant change, the system will generate an abnormal emotional expression signal. Here, a significant change means that the eigenvalue change of the covariance matrix exceeds b% (b is a set value) of the historical average. The generation of the abnormal emotional expression signal can timely remind medical staff to pay attention to the possible emotional and cognitive problems of the patient, providing a basis for further evaluation and intervention. Through in-depth analysis of facial micro-expression data, the patient's psychological and cognitive state can be understood from a new perspective, providing more comprehensive information for the early prediction of rheumatoid arthritis accompanied by brain cognitive dysfunction.

[0046] 3. Based on these processed data, a multi-dimensional abnormal association model is established. The core of this model is to perform time-domain alignment and causal analysis on spatio-temporal trajectory deviation events, path topology degradation characteristics, and expression parameter mutation signals.

[0047] For joint movement trajectory data, the present invention precisely calculates the difference between the patient's current joint movement trajectory and the historical baseline trajectory by means of vector superposition comparison in a three-dimensional coordinate system. In this three-dimensional coordinate system, the movement of the joint in space can be comprehensively and accurately reflected. Through meticulous comparison, a key index, namely the trajectory deviation degree, is obtained. To more intuitively measure the deviation degree, the present invention quantifies it in units of the standard deviation σ. As a commonly used statistic, the standard deviation can clearly show the degree of dispersion of data relative to the mean, so that the present invention can more accurately judge the degree to which the current joint movement trajectory deviates from the historical baseline.

[0048] For mobile path topological structure data, a fractal dimension calculation method is adopted to quantify the complexity of the patient's mobile path. First, calculate the fractal dimension value of the current path, and then compare it with the historical baseline value to obtain the reduction in the path fractal dimension. This reduction can reflect the change in the complexity of the patient's mobile path over time. If the reduction is large, it may imply problems with the patient's spatial navigation ability or motor coordination.

[0049] For facial micro-expression dynamic change data, the coefficient of variation is calculated to obtain the degree of fluctuation of the expression parameters. The coefficient of variation is defined as the ratio of the standard deviation to the mean. It eliminates the influence of the magnitude of the data mean and can more objectively reflect the relative degree of dispersion of the data. The degree of fluctuation of the expression parameters can, to a certain extent, reflect the change in the patient's emotional stability and mental state.

[0050] When the three values of the trajectory deviation degree, the reduction in the path fractal dimension, and the degree of fluctuation of the expression parameters all exceed the normal range, the present invention respectively marks them as spatio-temporal trajectory deviation events, path topological degradation characteristics, and expression parameter variation signals. These marks are the basis for subsequent analysis and early warning, and can help the present invention quickly identify possible abnormal situations.

[0051] Align the marked spatio-temporal trajectory deviation events, path topological degradation characteristics, and expression parameter variation signals on the time axis. This step is like arranging different puzzle pieces in chronological order so that the present invention can more clearly observe the temporal relationship between them. Then, analyze whether these three types of abnormalities show temporal coupling within two consecutive training cycles. Temporal coupling means that the time interval between the occurrences of these three types of abnormalities within the same training cycle does not exceed a set duration. This means that when these three types of abnormalities are closely related in time, they are likely to have some internal causal relationship. When the three types of abnormalities meet the judgment conditions and show temporal coupling within two consecutive training cycles, the system will generate an early warning of cognitive impairment. This early warning mechanism based on multi-dimensional abnormal association and temporal coupling can greatly improve the accuracy and reliability of early prediction of cognitive impairment.

[0052] To more comprehensively evaluate the cognitive and motor states of patients, the present invention designs an evaluation index for cognitive-motor coupling degree. This index is weighted and composed of three dimensions: action trajectory fidelity, environmental adaptation coefficient, and emotional feedback sensitivity. Action trajectory fidelity reflects the accuracy and stability of a patient's movements when performing joint activity training; the environmental adaptation coefficient embodies the patient's adaptability in different training environments; and emotional feedback sensitivity measures the degree of association between the patient's emotional expression and motor behavior. These three dimensions are interrelated and jointly reflect the coupling situation of the patient's cognitive and motor functions.

[0053] When the value of the cognitive-motor coupling degree evaluation index shows a monotonically decreasing trend for three consecutive training cycles and the decline slope exceeds a preset critical value, the present invention will activate a hierarchical warning mechanism. This mechanism is divided into three levels: low risk, medium risk, and high risk according to the degree of risk. At the same time, to make the warning more accurate and personalized, the present invention will dynamically adjust the judgment thresholds of each abnormal parameter according to the patient's age, disease course, and previous cognitive evaluation results. Patients of different ages have different physiological functions, the length of the disease course will affect the development degree of the disease, and previous cognitive evaluation results can reflect the patient's previous cognitive state. By comprehensively considering these factors, it can ensure that the hierarchical warning mechanism is more scientific and reasonable, and provide more timely and effective intervention measures for patients.

[0054] Example Two

[0055] When a patient repeatedly performs a grasping action, the three-dimensional motion trajectory reconstruction technology is used to detect the unnecessary turning-back phenomenon that appears in the finger movement path. When the number of turn-backs exceeds twice the standard deviation of the historical baseline value, it is determined as an abnormal action plan caused by working memory decline; at the same time, micro-expression features are extracted from the facial video stream. When it is detected that the response delay of the facial expression muscle group exceeds 300 milliseconds and is accompanied by a 40% decrease in the amplitude of the electromyogram activity around the eyes, it is marked as a state of insufficient compensation for emotional cognitive function; for gait data, through the spectral analysis of the migration trajectory of the plantar pressure center, the abnormal periodic fluctuation of the step length caused by the imbalance of attention resource allocation is identified, and this feature has a significant correlation with the rhythm disorder of the parietal alpha wave in the electroencephalogram signal;

[0056] In the joint movement preparation stage, a quantitative evaluation model of the formation efficiency of motor intention is established through the non-linear fitting of the rising slope of the beta wave energy in the pre-motor cortex and the action initiation delay time. When the slope decreases by 20% and the delay time extends by 500 milliseconds, the risk of developing executive dysfunction within the next three months is predicted to increase to 65%. During the action execution process, the dynamic correlation coefficient between the degree of mu rhythm suppression in the cerebellar region and the joint movement speed fluctuation is calculated. If this coefficient is continuously lower than 0.6 for three consecutive days, it is determined that the neuro-motor coordination function enters the compensatory period. After the task is completed, the joint decay curve of the prefrontal theta wave power reconstruction speed and the task objective achievement accuracy is analyzed. When the synchronous decline rate of both exceeds 0.5% per day, an early warning signal is triggered.

[0057] The joint movement data for 90 consecutive days are divided by week, and trend features such as the decay rate of the maximum joint flexion angle per week and the gradient change of the coefficient of variation of repetitive actions are extracted, and a multivariate regression model is established with the concentration of IL-6 inflammatory factors in the cerebrospinal fluid during the same period. At the same time, the delta wave enhancement event synchronized with the joint swelling cycle is isolated from the electroencephalogram signal, and the time-delay correlation between it and the change in the functional connection strength of the hippocampal region is analyzed through cross-wavelet transform. When it is found that the delta wave energy predicts a decrease in hippocampal connection strength 3 days in advance, an early warning index for the spread of inflammation to the central nervous system is established. Finally, the weight distribution of behavioral characteristics and neural indicators is integrated to generate a three-dimensional prediction map including the retention period of spatial orientation ability, the probability of episodic memory decline, and the risk index of pathological tremor.

[0058] According to the dynamic adjustment of the movement trajectory sampling scheme based on the spatial layout of the patient's residence, auxiliary positioning beacons are deployed in complex areas such as the kitchen and corridor to improve the ecological validity of spatial cognitive assessment. For the high-incidence period of joint stiffness from 3 to 5 am, a high-density electroencephalogram monitoring mode is activated to capture the associated characteristics of pathological slow waves during sleep and the deterioration of motor function after waking up. A compensatory ability development curve exclusive to the patient is established. When a new abnormal behavior pattern is detected, a supplementary training plan including virtual reality scene reconstruction and neural feedback intensity adjustment is automatically generated.

[0059] The above has described in detail an embodiment of the present invention, but the content described is only a preferred embodiment of the present invention and cannot be considered as limiting the scope of implementation of the present invention. All equivalent changes and improvements made according to the scope of the application of the present invention should still fall within the scope covered by the patent of the present invention.

Claims

1. A method for constructing a prediction model for rheumatoid arthritis associated with brain cognitive dysfunction, characterized in that: The following steps are involved: Each time the patient performs standardized joint motion training, multi-source dynamic behavior data is synchronously collected, including joint motion trajectory data, movement path topology data, and facial micro-expression data; Perform vector superposition comparison on the joint motion trajectory data in a three-dimensional coordinate system, calculate the trajectory deviation of each action phase point, and dynamically compare it with the patient's own historical baseline trajectory, which is generated by the training data before the patient sets the duration; The fractal dimension of the mobile path topological structure data is calculated to quantify the path complexity and analyze the change trend of the path fractal dimension value. The fractal dimension calculation adopts the box counting method to evaluate the complexity of the path by counting the number of grids covered by the path. Constructing a feature matrix for facial micro-expression data and calculating the covariance structure change of the expression parameter matrix, wherein the feature matrix includes data of three dimensions: frontalis muscle contraction frequency, eyelid opening and closing rate, and orbicularis oris muscle tremor amplitude; Based on the processed data, a multidimensional abnormal association model is established to perform time domain alignment and causal analysis on the spatiotemporal trajectory deviation events, path topology degradation characteristics, and expression parameter variation signals. When the three types of abnormalities show temporal coupling within two consecutive training cycles and meet the preset conditions, an early warning of cognitive dysfunction is generated.

2. The method for constructing a prediction model for rheumatoid arthritis associated with brain cognitive dysfunction according to claim 1, characterized in that: The collecting of multi-source dynamic behavior data includes: Wearable joint angle sensors are used to monitor the spatiotemporal characteristics of the movement trajectory of the large joints of the limbs in real time, and record the joint movement angle, movement speed and acceleration; The plantar pressure distribution sensor and indoor positioning device continuously record the topological structure data of the patient's movement path, including gait characteristics, path length, number of turning points and spatial positioning accuracy; The dynamic changes in the patient's facial micro-expressions are captured through a head-mounted miniature camera, with a focus on recording the contraction frequency of the frontalis muscle, the eyelid opening and closing rate, and the tremor amplitude of the orbicularis oris muscle.

3. The method for constructing a prediction model for rheumatoid arthritis associated with brain cognitive dysfunction according to claim 2, characterized in that: The standardized joint motion training includes a pre-set repetitive compound action sequence designed to simulate functional activities in daily life, including upper limb extension, lower limb walking and balance control. The execution duration, intensity and environmental conditions of each action sequence remain consistent.

4. The method for constructing a prediction model for rheumatoid arthritis associated with brain cognitive dysfunction according to claim 3, characterized in that: The dynamic comparison of joint motion trajectory data specifically includes: The vector superposition and comparison of the joint activity trajectory in the current training cycle with the patient's own historical baseline trajectory in the three-dimensional coordinate system are performed, and the trajectory deviation of each action phase point is calculated. When the spatial positioning deviation at the end of elbow flexion exceeds the preset dynamic threshold for three consecutive training sessions, the abnormal movement pattern mark is triggered; the preset dynamic threshold is dynamically adjusted according to the patient's age, disease course and joint mobility.

5. The method for constructing a prediction model for rheumatoid arthritis associated with brain cognitive dysfunction according to claim 4, characterized in that: The mobile path topological structure analysis adopts a fractal dimension calculation method to quantify the complexity of the patient's movement trajectory in the same training environment. When the path fractal dimension value decreases by more than a threshold compared with the previous training and is accompanied by an abnormal increase in the number of turning points, it is determined to be a feature of spatial navigation ability degradation; the abnormal increase in the number of turning points refers to the number of turning points exceeding a% of the historical average, where a is a set value.

6. The method for constructing a prediction model for rheumatoid arthritis associated with brain cognitive dysfunction according to claim 5, characterized in that: The facial micro-expression dynamic analysis establishes a patient-specific expression feature matrix, focusing on capturing three biometric characteristic parameters: frontalis muscle contraction frequency, eyelid opening and closing rate, and orbicularis oris muscle tremor amplitude when executing complex action sequences, and generates an abnormal emotion expression signal when the covariance structure of the expression parameter matrix changes significantly; the significant change refers to the change in the eigenvalue of the covariance matrix exceeding b% of the historical average, where b is a set value.

7. The method for constructing a prediction model for rheumatoid arthritis associated with brain cognitive dysfunction according to claim 6, characterized in that: By means of vector superposition comparison in a three-dimensional coordinate system, the difference between the patient's current joint movement trajectory and the historical baseline trajectory is calculated to obtain the trajectory deviation, and the degree of deviation is quantified in units of standard deviation σ; the complexity of the patient's movement path is quantified by using a fractal dimension calculation method, and the difference between the current path fractal dimension value and the historical baseline value is calculated to obtain the path fractal dimension reduction; the degree of fluctuation of the expression parameters is obtained by calculating the coefficient of variation of the dynamic change data of facial micro-expressions, and the coefficient of variation is the ratio of the standard deviation to the mean; When the values ​​of trajectory deviation, path fractal dimension decrease and expression parameter fluctuation degree all exceed the normal range, they are marked as spatiotemporal trajectory deviation events, path topology degradation features and expression parameter variation signals. The spatiotemporal trajectory deviation events, path topology degradation features and expression parameter variation signals are aligned on the time axis to analyze whether the three types of anomalies show temporal coupling in two consecutive training cycles. The temporal coupling means that the time interval between the occurrence of the three types of anomalies in the same training cycle does not exceed the set duration. When the three types of anomalies meet the judgment conditions and show temporal coupling in two consecutive training cycles, an early warning of cognitive dysfunction is generated.

8. The method for constructing a prediction model for rheumatoid arthritis associated with brain cognitive dysfunction according to claim 1, characterized in that: A cognitive-motor coupling evaluation index is designed, which is composed of three weighted dimensions: motion trajectory fidelity, environmental adaptation coefficient, and emotional feedback sensitivity. When the index value shows a monotonically decreasing trend for three consecutive training cycles and the slope exceeds the preset critical value, the graded early warning mechanism is activated, and the judgment threshold of each abnormal parameter is dynamically adjusted according to the patient's age, disease course, and previous cognitive assessment results; the graded early warning mechanism includes three levels: low risk, medium risk, and high risk.

Images