CN114376566A - Symmetry evaluation method for lower limb segments during hand load - Google Patents

Abstract

The invention relates to the technical field of weight-bearing gait analysis, in particular to a method for evaluating symmetry of lower limb segments during hand weight bearing, which comprises S1, data acquisition and processing; s2, data segmentation; s3, constructing a gait symmetry model; s4, carrying out normalization processing on the symmetry data, and calculating the symmetry data according to a threshold value method; and S5, establishing a symmetry grade under load and a consistency evaluation under load according to the symmetry. The method identifies and quantifies the difference of left and right leg signals by using a classification method based on statistical learning, and takes a quantification result as a first criterion for judging the symmetry; a dynamic time warping algorithm is introduced to solve the waveform shape similarity of the left leg time domain signal and the right leg time domain signal, and the waveform shape similarity is used as a second criterion for judging the symmetry; and (4) obtaining a symmetry trend by fitting left and right leg signals so as to further judge the symmetry.

Description

Symmetry evaluation method for lower limb segments during hand load

Technical Field

The invention relates to the technical field of load gait analysis, in particular to a symmetry evaluation method of a lower limb segment during hand load.

Background

Carrying a bag on the shoulder or holding a heavy object on the hand is a common upper limb weight-bearing mode, which may have an impact on lower limb gait, and the symmetry of gait is the primary manifestation of assessing gait. The symmetry is an important index for evaluating the gait, and has important research value in many fields. Clinical studies have found that gait symmetry can reveal recovery and pathological features in patients with motor and nervous system disorders. These include hemiplegia, pilot fractures, lower limb amputation, anterior cruciate ligament reconstruction, etc. Meanwhile, the gait symmetry can also predict the risk index of falling of the old. Relevant research of kinematics finds that the symmetry of lower limb gait greatly helps to detect the movement normative of professional athletes. Such as analyzing the symmetry of the sprint during run to adjust the speed during run-up. The study on the daily activities is also carried out at the same time, such as discussing whether the walking speed affects the gait symmetry, the swing amplitude of the arm affects the gait symmetry, and the shoe-wearing walking and barefoot walking affect the gait symmetry. For healthy people, the long-term gait asymmetry increases the oxygen and energy consumption of walking. It has been shown that gait asymmetry can also lead to osteoporosis in humans, creating more load on lower limb joints and increasing the risk of osteoarthritis. Therefore, the research on the gait symmetry has important significance in the fields of kinematics, medicine, biomechanics and the like.

Heavy walking, particularly heavy-weight walking by hands, is common among people such as office workers, students, and shoppers, and long-term heavy walking may cause problems such as muscle pain and damage to joints of lower limbs, and therefore, deep exploration of gait when walking under different load conditions is required. Ideally, the gait will be shifted in the same direction by a single load, and the degree of shift will tend to change in proportion to the increase of the load, but the degree of the rule will vary from subject to subject due to the difference in physical function and behavior habits of each individual. Therefore, a proper symmetry evaluation model is established, so that the gait symmetry of the testee under different load conditions can be more accurately evaluated, and whether the gait of the testee meets the symmetry rule under the load is further checked.

Lvstsky et al, a gait symmetry and consistency evaluation method based on multiple sensors, establishes a regression angle deviation model, analyzes the gait symmetry under load by analyzing the deviation angle of a linear regression line, but the method has low robustness.

Disclosure of Invention

The technical problem to be solved by the invention is as follows: aiming at the problems that unified gait symmetry and consistency analysis standards are lacked in the prior art and the robustness of the prior method is low, the gait symmetry under different load bearing states is utilized to establish a load bearing symmetry grade classification rule, and the movement consistency is measured through the gait symmetry of multiple experiments.

The technical scheme adopted by the invention is as follows: a method for evaluating the symmetry of a lower limb segment during hand weight bearing comprises the following steps:

s1, data acquisition and processing: acquiring acceleration data of left and right leg sensors respectively, synthesizing signals of the acceleration data in three-axis directions, and filtering the synthesized acceleration signals;

the data collected by the acceleration sensor are acceleration signals in the three-axis direction, and in order to weaken the direction error caused by the installation of the sensor, the three-axis acceleration signals are subjected to amplitude scalar operation to synthesize three-axis resultant acceleration signals; formula (1) is the calculation process of the amplitude of the resultant acceleration:

in the formula, A_xIndicating an acceleration signal in the front-rear direction, A_yIndicating an acceleration signal in the left-right direction, A_zIndicating an acceleration signal in the up-down direction.

Filtering useless signals and mixed noise signals in the acceleration signals through a Savitzky-Golay filter, wherein the Savitzky-Golay filter is a filter for filtering and denoising based on local polynomial least square fitting in a time domain;

Savitzky-Golay filtering removes noise while preserving the shape of the signal.

S2, data segmentation: the method comprises the following specific steps of taking a local minimum point of three-axis resultant acceleration as a period division point, wherein the local minimum point corresponds to the moment when a sole contacts the ground, and the period division comprises the following specific steps:

s21, detecting all minimum value points of the acceleration signal through a peak value detection method;

s22, with the number of sample points included in the normal gait cycle as a constraint, searching the next sample point forward and backward, and deleting the rest extreme points;

s23, because a complete cycle can not be formed at the beginning and the end, an error calibration point can be generated, the range of the amplitude of the extreme point and the amplitude of the average extreme point is set as a constraint, and redundant cycle division points are filtered;

s3, constructing a gait symmetry model: judging the gait symmetry direction and calculating the gait symmetry of the combined acceleration signal which is divided by the S2;

further, the gait cycles of the left leg and the right leg are classified twice by a nonlinear classification method, and the gait difference of the left leg and the right leg is quantified by using the accuracy;

judging the gait symmetry direction: the gait combined acceleration contains information such as acting force and speed change, and the acting force is directly related to the gait deviation, so that the deviation direction of the gait can be distinguished by analyzing the combined acceleration of the left leg and the right leg; comparing the fixed integral value of the gait acceleration signals of the left leg and the right leg; if the constant integral value of the left leg acceleration signal is larger than that of the right leg acceleration signal, the gait is considered to be deviated to the left; if the constant integral value of the right leg acceleration signal is larger than that of the left leg acceleration signal, the gait is considered to be deviated to the right; a direction of "+" represents a left bias and a direction of "-" represents a right bias.

Walking is a complex behavioral action and comprises a plurality of potential nonlinear information, and the information can be analyzed to effectively identify the symmetry of gait; the gait symmetry is identified by using a classification method based on statistical learning, the degree of the gait symmetry is quantified by classification accuracy, and if the classification accuracy is high, the difference between left and right leg signals is large, so that the left and right leg signals are easy to distinguish; if the classification accuracy is low, the difference of the left leg signal and the right leg signal is small and difficult to distinguish; by applying a classification method based on statistical learning, more nonlinear features can be identified, and the method has good robustness and interpretability; in order to improve the reliability and generalization capability of the model, common features such as an average value, a variance, a maximum value, skewness and the like are adopted, and Decision Trees (DT) are used for carrying out classification processing on left and right leg signals;

the DT algorithm is a data-based inductive inference classification method, which can establish a decision tree for classification judgment by performing inference learning on disordered sample data in a decision table form without excessive field background knowledge and a fixed diagnosis model, and extract clear and intuitive classification rules from the decision tree; common decision tree algorithms are: ID3, C4.5, and CART; in contrast to ID3 and C4.5, CART can be classified and regression analysis can be performed, but CART can only generate binary trees, while ID3 and C4.5 can be multi-branched, and CART is more suitable for large samples, ID3 and C4.5 are suitable for small samples; c4.5 is the inheritance and improvement of ID3, and solves the problems that ID3 can not process continuous variables (only process classified variables) and is sensitive to missing values (Table 1), so the C4.5 decision tree method is adopted in the invention.

Table 1 the general DT algorithm is as follows:

further, a DTW (dynamic Time warping) algorithm is adopted to calculate the waveform similarity of the synthetic acceleration signals;

by adopting a nonlinear classification method, a plurality of pieces of hidden gait information can be searched, but the characteristics still cannot completely replace time domain signals, and the gait symmetry is obtained by introducing waveform similarity of time domain signals of left and right legs because the amplitude of the gait signals can be influenced by considering that the small error in the installation of the acceleration sensor can influence the amplitude of the gait signals, but the change trend of the gait signals is not easily influenced; because the time lengths of the gait signals of the left leg and the right leg are not completely equal, the gait signal of one leg needs to be waring and distorted on a time axis to align the two sequences; DTW is an effective method for warping distortion, and similarity between sequences is calculated by shortening and extending time sequences;

the DTW algorithm carries out a specific calculation process of waveform similarity calculation on the acceleration signals of the left leg and the right leg, wherein the time sequences of the left leg and the right leg are assumed to be A ═ a respectively₁,a₂,...,a_i,...,a_n}，B＝{b₁,b₂,...,b_j,...,b_mAn n x m matrix is constructed, and the element d (a) in the matrix_i,b_j) Represents a_iAnd b_jEuclidean distance of two points, d (a)_i,b_j) The smaller, the higher the similarity; in an n × m matrix, the continuum of elements constitutes a path, which is assumed to be a regular path, denoted by W:

W＝(ω₁,ω₂,...,ω_k)(max(n,m)≤k≤m+n-1) (2)

wherein, w_kAnd k is the k element of W, m is the number of left leg signal points, and n is the number of right leg signal points.

Finding the path with the shortest accumulative distance from a plurality of paths of the regular paths meeting the constraint conditions:

wherein, K is the number of path points and is used for compensating the regular paths W with different lengths.

The initial condition is set to D (1,1) ═ D (a)₁,b₁) (ii) a Starting from two sequence starting points (1,1), obtaining a minimum accumulated value D (n, m) through iterative calculation of equations (2) and (3), wherein the accumulated value is the shortest accumulated distance DTW (A, B) of the time sequences A and B.

The gait cycle of the left leg and the right leg is classified twice by utilizing a nonlinear classification method, and the gait difference of the left leg and the right leg is quantified by utilizing the accuracy as a first criterion for judging the symmetry; in addition, in order to resist errors in installation of the acceleration sensor and fully extract time domain information, DTW is introduced to obtain waveform similarity of time domain signals of the left leg and the right leg, and the waveform similarity is used as a second criterion for judging symmetry;

s4, carrying out normalization processing on the symmetry size data, and averaging according to a threshold value method if the normalized symmetry size data is lower than a threshold value to obtain averaged symmetry size data; if the sum of the acceleration signals is higher than the threshold value, carrying out symmetry magnitude trend analysis on the sum of the acceleration signals segmented in the step S2 to obtain symmetry magnitude data;

normalizing the first criterion and the second criterion in order to make the two criteria in the same dimension; if the difference of the normalized values is small, the symmetry results obtained by the first judgment data and the second judgment data are basically consistent, and the average value of the symmetry results is taken as a final result; if the difference of the normalized values is large, the first judgment data and the second judgment data are considered to conflict, the symmetry trend of the left leg and the right leg is obtained by fitting the acceleration signals of the left leg and the right leg, and the two judgment criteria are further judged, so that a more appropriate criterion is selected as a final result;

obtaining the symmetry trend of the left leg and the right leg, namely resampling the gait cycles of the left leg and the right leg after cycle division to obtain the same 100 points; obtaining 100 two-dimensional points by taking the left leg data as x and the right leg data as y, and fitting the two-dimensional points to obtain a deviation angle to judge the trend of gait symmetry; the specific steps of solving the offset angle obtained by fitting the left leg data and the right leg data are as follows:

suppose that

And

respectively represent the left and right gait data of the experimenter under the loading state U, and,

and

the method is to obtain a new vector by means of mean value removal on the basis of original data, and is defined as follows:

wherein i represents the ith sample point of the experimenter S in the U group weight bearing mode; 1,2, 100 × N, S1, 2, 5, U1, 2, 5; n is the total number of cycles of the experimenter S under the loading state U; constructing a linear regression model by taking the left gait data as independent variables and the right gait data as dependent variables:

wherein, beta_0S,UAnd beta_1S,URespectively, the intercept and the regression coefficient of the experimenter S in the U group loading mode, epsilon_S,UIs a random error term;

is a predicted gait acceleration value; taking a single gait pattern of a single subject as an example, equation (6) is simplified as follows:

Y_i＝β₀+β₁L_i+e (7)

wherein, Y_iAs right side gait data, L_iLeft side gait data, e residual term, regression coefficient beta₁And intercept beta₀Respectively representing offset and amplitude difference;

removing strong influence points in the data points, namely points with strong influence on the result, and fitting the data points; wherein, the closer the slope of the fitting straight line is to 1, the more symmetrical the gait of the left leg and the right leg is; the experiment quantifies the level of symmetry using the rotation angle radian theta between the fitted straight line and the reference line with the slope of 1; the calculation formula of the rotation angle radian θ is as follows:

wherein k is₁To fit the slope of the line, θ ∈ [0, + ∞]A smaller theta represents better symmetry.

S5, establishing a symmetry grade under load and a consistency evaluation under load according to the symmetry;

symmetry rating under load: ideally, the gait will be deviated in the same direction by the load on one side, and the deviation degree will show a proportional change trend along with the increase of the load degree; however, in actual cases, the degree of expression varies depending on the difference in physical function of each subject;

difference in gait symmetry of subject under two different load-bearing conditions D_ijTo represent; d_ijIs defined as D_ij＝S_Gi-S_Gj,i、j∈{1,2,3,4,5}，i≠j；

Wherein S is_GiIs the gait symmetry of the action Gi, S_GjGait symmetry for action Gj; i D_ijL is S_GiRelative to S_GjThe size of the offset, the sign representing S_GiRelative to S_GjThe direction of the offset; "+" represents a left bias and "-" represents a right bias, then if D is_ijIf the value is greater than 0, the action Gi is deviated to the left relative to the gait of the action Gj; if D is_ijIf the value is less than 0, the gait of the action Gi deviates to the right relative to the gait of the action Gj;

evaluation of the symmetry level of gait under load: whether the symmetrical offset directions of loads on different sides are consistent or not, whether the symmetrical offset magnitude of loads on the same side is in direct proportion to the load degree or not and whether the symmetrical offset direction is consistent with the load direction or not; dividing 4 symmetry levels from high to low according to a given rule, wherein the symmetry levels are A, B, C and D; listing the results of right-hand-held heavy walking, left-hand-held heavy walking and overall symmetry evaluation (table III); taking the overall evaluation as an example, the specific steps of gait symmetry grading are listed:

(1) if D is₁₂×D₁₃> 0 and D₁₄×D₁₅If the value is more than 0, the symmetry grade is marked as C, otherwise, the value is D;

(2) if D satisfies C on the premise of symmetry level₁₃＞D₁₂And D is₁₅＞D₁₄The symmetry level is designated as B;

(3) on the premise that the symmetry level meets B, if D₁₂< 0 and D₁₃＜0、D₁₄> 0 and D ₁₅0, and the symmetry grade is marked as A;

"a" and "B" are considered close to ideal with good symmetry, while "C" level and "D" are considered inconsistent with ideal with poor symmetry;

TABLE 2 symbol representation

Evaluation of consistency under load: the gait consistency refers to the change degree of the gait symmetry of a subject after multiple experiments under the same condition, and the smaller the change degree is, the better the gait consistency is represented;

firstly, arranging the symmetries under 5 different load conditions from small to large, analyzing the arrangement conditions of three experiments in sequence, and then respectively calculating the arrangement difference of the symmetries under each load condition under the three experiments; this gap is denoted herein by the letter R; if R is equal to 0, the action consistency is considered to be excellent; if R is equal to 1 or 2, the action consistency is considered to be better; if R is equal to 3 or 4, the action consistency is considered to be poor; for example, if the symmetry ranks of a subject are 2, 3, and 4 in three experiments without load, the difference between the maximum rank and the minimum rank is 2, and the action consistency is considered to be better when walking without load.

The invention has the beneficial effects that:

1. the assessment method provides a new guidance scheme for the recovery state of clinical patients;

2. the method has lower requirement on data precision and higher robustness;

3. compared with a gait symmetry and consistency evaluation method model based on multiple sensors and using the same data set, the method of the invention has the advantages that the shank gait consistency index is 5 percent higher;

4. the index P for measuring the difference between the control group and the experimental group is 0.006 which is far less than 0.05; therefore, the gait symmetry under different loading conditions is significantly different, and the gait symmetry indexes of the shank and the thigh under the loading condition are both 12.5 percent higher.

5. Identifying and quantifying left and right leg signal differences by using a statistical learning-based classification method, and taking a quantification result as a first criterion for judging the symmetry; a dynamic time warping algorithm is introduced to solve the waveform shape similarity of the left leg time domain signal and the right leg time domain signal, and the waveform shape similarity is used as a second criterion for judging the symmetry; and (4) obtaining a symmetry trend by fitting left and right leg signals so as to further judge the symmetry.

Drawings

FIG. 1 is a box-shaped view of gait symmetry of a prior art lower leg under different loads;

FIG. 2 is a weighted gait assessment framework of the invention;

FIG. 3 is a data pre-processing process of the present invention;

FIG. 4 is a symmetry model construction of the present invention;

FIG. 5 is a DTW distortion mapping curve of the present invention;

FIG. 6 is a left and right leg acceleration fit of the present invention;

FIG. 7 the results of the calf movement consistency analysis of the present invention.

Detailed Description

The invention will be further described with reference to the accompanying drawings and examples, which are simplified schematic drawings and illustrate only the basic structure of the invention in a schematic manner, and therefore only show the structures relevant to the invention.

FIG. 1 is a box diagram of the gait symmetry of the lower leg under different loads in the prior art, and it can be seen that the difference of the gait symmetry of the lower leg under different load conditions;

a method for evaluating the symmetry of a lower limb segment during hand weight bearing comprises the following steps:

as shown in fig. 2-4, S1, data acquisition and processing: respectively acquiring acceleration data of a large leg sensor and a small leg sensor of a left leg and a right leg, synthesizing signals of the acceleration data in the three-axis direction, and filtering the synthesized acceleration signals;

s2, data segmentation: taking a local minimum value point of the three-axis resultant acceleration as a period division point, wherein the local minimum value point corresponds to the moment when the sole contacts the ground;

in the embodiment, 8 volunteers with age range of 21-25 years, average height of 1.71m and average height of 75.6kg are included, and each person has no gait disorder and is more accustomed to using the right hand; each volunteer wears an acceleration inertial sensor in the middle of the thigh and the middle of the calf to carry out five gait modes, namely no-load walking, right-hand load walking of 2.5kg, right-hand load walking of 5kg, left-hand load walking of 2.5kg and left-hand load walking of 5kg, which are respectively marked as G1, G2, G3, G4 and G5, each group of experiments are repeated three times, and the experiments sample 3-axis linear acceleration at a constant rate of 250 hz.

S3, constructing a gait symmetry model: judging the gait symmetry direction and calculating the gait symmetry of the combined acceleration signal which is divided by the S2;

further, the gait cycles of the left leg and the right leg are classified twice by a nonlinear classification method, and the gait difference of the left leg and the right leg is quantified by using the accuracy;

classifying the periodic signals of the left leg and the right leg by using a C4.5 decision tree, wherein the classification accuracy is shown in a table 3; for the same subject, if the classification accuracy of the left leg and the right leg is low under the condition of no load, the classification accuracy of the subject is obviously improved when the left hand and the right hand bear loads, which indicates that the symmetry of the subject is obviously deteriorated due to the influence of the loads of the left hand and the right hand; if the classification accuracy of the left leg and the right leg is higher under the condition of no load, the classification accuracy is obviously reduced when one side bears the load, and the classification accuracy is improved when the other side bears the load, because the load on one side improves the symmetry of the gait, and the load on the other side obviously deteriorates the symmetry of the gait; as can be seen from Table 3, the results obtained correspond to the trend of symmetry under load; therefore, it is reasonable to use the accuracy obtained by classification using DT as the first criterion.

TABLE 3 Classification accuracy

Note that P_G1Is the classification accuracy in normal walking, P_G2The classification accuracy, P, for a right-hand load of 2.5kg_G3The classification accuracy, P, for a weight of 5kg on the right hand_G4The classification accuracy when the left hand bears 2.5kg of load, P_G5For the classification accuracy at a left-handed weight of 5kg, Sub1 to Sub8 are the subject numbers;

further, a DTW algorithm is adopted to calculate the waveform similarity of the synthetic acceleration signals; as shown in fig. 5, through warping distortion, the DTW algorithm can find corresponding similar points on the left leg signal and the right leg signal, the connecting line of the left leg and the right leg represents the corresponding points of the left leg and the right leg distortion, and the similarity of the left leg and the right leg signal in the same gait phase is compared, so that the similarity result is more convincing.

The two main criteria are important bases for judging the gait symmetry, but in practical application, due to errors in installation positions and the like, the situation that one criterion can well show the gait symmetry and the other criterion is greatly influenced by the errors occurs; aiming at the situation, the gait symmetry trend is obtained by fitting the left and right leg signals to further judge the symmetry magnitude, the symmetry trend is calculated as shown in figure 6, a dotted line represents a fitting line under all data points, a triangle represents a strong influence point, a black line represents the fitting line after the strong influence point is removed, the slope of the fitting straight line of the black line is approximately equal to 1, and the gait symmetry of the left and right legs is represented.

S4, carrying out normalization processing on the symmetry size data, and averaging according to a threshold value method if the normalized symmetry size data is lower than a threshold value to obtain averaged symmetry size data; if the sum of the acceleration signals is higher than the threshold, performing symmetry magnitude trend analysis on the sum of the acceleration signals segmented in the step S2 to obtain symmetry magnitude data, wherein the threshold is 0.3 in the embodiment;

the gait symmetry under each load condition is obtained by combining the gait offset direction and the gait symmetry, the statistic results of the symmetry of the crus are listed, and the statistic results are shown in table 4:

TABLE 4 statistical table of symmetry

Note that_G1Is the gait symmetry during normal walking, S_G2Gait symmetry of right-hand load 2.5kg, S_G3Is the gait symmetry of a right-hand load of 5kg, S_G4Gait symmetry of 2.5kg for left-handed load, S_G5The gait symmetry is when the left hand bears the weight of 5 kg;

the variance estimation is carried out on the gait symmetry of 5 different load conditions as shown in the table 5, and the corresponding P value is less than 0.05; therefore, the gait symmetry under different load conditions is significantly different.

Table 5 model test results

And SS: the sum of squares; df: degree of freedom; MS: mean square error; f: ratio of inter-group mean square to intra-group mean square P: index for measuring difference between control group and experimental group

S5, establishing a symmetry grade under load and a consistency evaluation under load according to the symmetry;

as shown in fig. 7, analysis of the lower leg segment showed that the gait consistency of the calf segment reached 82.5% while the gait consistency of the thigh segment reached 90%. Gait will exhibit different motor congruencies for different weight bearing patterns. For the G1 group walking without weight bearing, 87.5% and 100% of the subjects on the calf and thigh, respectively, showed better consistency; for the G2 group, which was 2.5kg weight bearing on the left hand, 87.5% and 87.5% of subjects on the calf and thigh, respectively, showed better consistency; for the G3 group, which was 5kg left-handed weighted, 62.5% and 97.5% of subjects on the calf and thigh, respectively, showed better consistency; for the G4 group with a 2.5kg right hand weight, 87.5% and 67.5% of subjects on the calf and thigh, respectively, showed better consistency; for the G5 group with a weight of 5kg on the right hand, 87.5% and 100% of the subjects on the calf and thigh, respectively, showed better consistency; it can be seen that the subject's consistency is optimal without weight bearing; the subject had the worst consistency when the right hand was loaded with 2.5 kg.

TABLE 6 gait uniformity test results

In light of the foregoing description of the preferred embodiment of the present invention, many modifications and variations will be apparent to those skilled in the art without departing from the spirit and scope of the invention. The technical scope of the present invention is not limited to the content of the specification, and must be determined according to the scope of the claims.

Claims (6)

1. A method for assessing symmetry of a lower limb segment under hand weight, comprising the steps of:

s1, data acquisition and processing: respectively acquiring sensor acceleration data of a left leg and a right leg, synthesizing signals of the acceleration data in three-axis directions, and filtering the synthesized acceleration signals;

s2, data segmentation: taking local minimum value points of the combined acceleration signal as cycle division points, and detecting all minimum value points of the combined acceleration signal through peak value detection; taking the number of sample points contained in the gait cycle as a constraint, searching the next sample point forward and backward, and deleting the rest extreme points; taking the range of the amplitude of the extreme point and the amplitude of the average extreme point as a constraint, and filtering redundant cycle division points;

s3, constructing a gait symmetry model: judging the gait symmetry direction and calculating the gait symmetry of the combined acceleration signal which is divided by the S2;

s4, normalization: carrying out normalization processing on the data of the symmetry size, and averaging if the data of the symmetry size is lower than a threshold value according to a threshold value method to obtain the data of the symmetry size; if the sum of the acceleration signals is higher than the threshold value, carrying out symmetry magnitude trend analysis on the sum of the acceleration signals segmented in the step S2 to obtain symmetry magnitude data;

and S5, establishing a symmetry grade under load and a consistency evaluation under load according to the symmetry data.

2. The method of claim 1 for assessing symmetry of a lower limb segment during hand weight bearing, wherein: the filtering employs a Savitzky-Golay filter.

3. The method of claim 1 for assessing symmetry of a lower limb segment during hand weight bearing, wherein: the gait symmetry direction judgment is that the gait is deviated to the left by comparing the fixed integral value of the gait acceleration signals of the left leg and the right leg, if the fixed integral value of the left leg combined acceleration signal is larger than the fixed integral value of the right leg combined acceleration signal, the gait is deviated to the right, and if the fixed integral value of the right leg combined acceleration signal is larger than the fixed integral value of the left leg combined acceleration signal, the gait is deviated to the right.

4. The method of claim 1 for assessing symmetry of a lower limb segment during hand weight bearing, wherein: the gait symmetry magnitude calculation comprises:

performing secondary classification on gait cycles of the left leg and the right leg by a nonlinear classification method, and quantifying gait difference of the left leg and the right leg by using accuracy;

and performing waveform similarity calculation on the synthesized acceleration signals by adopting a DTW algorithm.

5. The method of claim 4 for assessing symmetry of a lower limb segment during hand-weight bearing, wherein: the nonlinear classification method is a C4.5 decision tree in a DT method, and classification processing is carried out on left and right leg signals by utilizing the C4.5 decision tree.

6. The method of claim 4 for assessing symmetry of a lower limb segment during hand-weight bearing, wherein: the waveform similarity calculation includes:

let the left and right leg time series be A ═ a₁,a₂,...,a_i,...,a_n}，B＝{b₁,b₂,...,b_j,...,b_mAn n x m matrix is constructed, and the element d (a) in the matrix_i,b_j) Represents a_iAnd b_jThe Euclidean distance of two points;

in an n × m matrix, the continuum of elements constitutes a path, which is assumed to be a regular path, denoted by W:

W＝(ω₁,ω₂,...,ω_k)，(max(n,m)≤k≤m+n-1) (2)

wherein, w_kRepresenting the kth element of W, wherein m is the number of left leg signal points, and n is the number of right leg signal points;

finding the path with the shortest accumulative distance from a plurality of paths of the regular paths meeting the constraint conditions:

k is the number of path points and is used for compensating regular paths W with different lengths;

the initial condition is set to D (1,1) ═ D (a)₁,b₁) (ii) a Starting from two sequence starting points (1,1), obtaining a minimum accumulated value D (n, m) through iterative calculation of equations (2) and (3), wherein the accumulated value is the shortest accumulated distance DTW (A, B) of the time sequences A and B.

Images