Acupuncture skill objective assessment and scoring method based on force feedback time series data

Abstract

The application discloses a kind of based on force feedback timing data's acupuncture skill objective assessment and scoring method, belong to intelligent acupuncture teaching technical field, including the following steps: S1, three-dimensional force feedback timing original data under standard assessment action is collected;S2, original force feedback timing data is preprocessed;S3, four-dimensional multidimensional feature vector of time domain dynamics, frequency domain rhythm, time-frequency domain coupling and technique cooperativity is extracted from standardization timing data;S4, feature weight distribution model based on improved analytic hierarchy process is constructed, and the weight coefficient of each characteristic dimension and sub-feature is determined;S5, the standard deviation of expert standard value and each sub-feature is determined, establishes type fuzzy membership function single feature scoring model, and each sub-feature single score is calculated;S6, according to feature weight coefficient and single score, calculate comprehensive skill score, generate assessment analysis report.The application realizes the whole process, high precision, objective assessment of acupuncture skill, and is convenient for traditional chinese medical science acupuncture teaching.

Description

Technical Field

[0001] This invention relates to the field of intelligent acupuncture teaching technology, and in particular to an objective assessment and scoring method for acupuncture skills based on force feedback time-series data. Background Technology

[0002] Acupuncture is an important component of Traditional Chinese Medicine, and its therapeutic effects are closely related to the skill level of the practitioner. Currently, acupuncture skill assessments mainly rely on the subjective evaluation of senior experts, which has the following significant shortcomings: (1) High subjectivity: Different experts have different evaluation criteria, and the scoring results are easily affected by personal experience and subjective factors; (2) Difficulty in quantification: It is impossible to accurately quantify key technique parameters such as needle insertion speed, lifting and thrusting amplitude, and twisting frequency; (3) Unable to evaluate “deqi”: “Deqi” is the key to the efficacy of acupuncture, but traditional assessment methods cannot objectively evaluate whether the operator can accurately induce “deqi”. (4) Inefficient: It requires multiple experts to participate in the assessment at the same time, which is time-consuming and labor-intensive, and it is difficult to promote and apply it on a large scale.

[0003] In recent years, with the development of sensor and artificial intelligence technologies, some acupuncture skill assessment systems based on force feedback have emerged. However, most existing technologies only extract simple time-domain features (such as peak force and average force), failing to fully explore the rich information contained in the force feedback time-series data, especially the dynamic and rhythmic features related to "deqi" (the sensation of qi). Furthermore, the weight allocation in existing scoring models often employs subjective assignment methods, lacking scientific basis and failing to consider the influence of different acupoints and manipulation techniques on the importance of features, resulting in poor consistency between scoring results and expert evaluations.

[0004] Therefore, there is an urgent need to develop an assessment and scoring method that can comprehensively, objectively, and accurately evaluate acupuncture skills. Summary of the Invention

[0005] The purpose of this invention is to provide an objective assessment and scoring method for acupuncture skills based on force feedback time-series data, so as to achieve full-process, high-precision, and objective assessment of acupuncture skills and solve the core problems of traditional assessment methods such as strong subjectivity, inconsistent standards, and inability to quantify "deqi".

[0006] To achieve the above objectives, this invention provides an objective assessment and scoring method for acupuncture skills based on force feedback time-series data, comprising the following steps: S1. Construct an acupuncture force feedback data acquisition system to collect raw three-dimensional force feedback time sequence data under standard assessment movements; S2. Preprocess the original force feedback time series data to obtain standardized segmented time series data; S3. Extract multi-dimensional feature vectors from standardized time-series data, including time-domain dynamic features, frequency-domain rhythm features, time-frequency domain coupling features, and technique synergy features; S4. Construct a feature weight allocation model based on the improved analytic hierarchy process (AHP) and determine the weight coefficients of each feature dimension and sub-feature. S5. Determine the expert standard value and standard deviation of each sub-feature, establish a single-feature scoring model based on fuzzy membership function, and calculate the individual score of each sub-feature. S6. Calculate the comprehensive skill score based on the feature weight coefficients and individual scores, and generate an assessment analysis report.

[0007] Preferably, in S1, the acupuncture force feedback data acquisition system includes: a simulated human acupuncture model, a three-dimensional force sensor installed at the acupoints of the model, a data acquisition card, and a computer processing unit; the sampling frequency of the three-dimensional force sensor is not less than 1000Hz, the force measurement range is 0-50N, and the measurement accuracy is not less than 0.01N.

[0008] Preferably, in S2, the preprocessing of the original force feedback time series data specifically includes: S21. Wavelet thresholding denoising method is used to remove high-frequency noise and baseline drift from the original data; S22. Automatically identify and segment the three operation stages of needle insertion, needle movement, and needle withdrawal based on the force change rate threshold method; S23. Perform length normalization and amplitude standardization on the data of each stage; S24. An improved isolated forest algorithm is used to detect and correct operational anomalies.

[0009] Preferably, in S3, the time-domain dynamic features include: average needle insertion speed, peak needle insertion force, force rise time, and needle insertion stability during the needle insertion phase; lifting and thrusting amplitude, lifting and thrusting frequency, twisting angle, twisting frequency, force fluctuation standard deviation, maximum force value, and minimum force value during the needle movement phase; and average needle withdrawal speed, needle withdrawal time, and residual force value during the needle withdrawal phase. Among these, needle insertion stability is defined as the average value of the sliding window standard deviation of the force signal during the needle insertion phase, with a window size of 50 sampling points.

[0010] Preferably, in S3, the frequency domain rhythm features are extracted by fast Fourier transform, including the main frequency, the main frequency band energy ratio, and the harmonic distortion degree of the needle-walking stage; wherein the harmonic distortion degree is the ratio of the sum of the energy of the first 3 harmonics to the fundamental frequency energy, and the main frequency band energy ratio is the proportion of the energy within the range of the main frequency ± 0.5 Hz to the total energy.

[0011] Preferably, in S3, the time-frequency domain coupling features are extracted using Morlet wavelet continuous wavelet transform, including the gas characteristic index. GFI The calculation formula is as follows: ; in, These are continuous wavelet transform coefficients. and This refers to the start and end times of the needle-carrying phase. and This is the characteristic frequency band of the gas extraction process. The sampling frequency.

[0012] Preferably, in S3, the synergistic features of the manipulation are extracted through cross-wavelet transform, including the lifting-twist synergistic index. CSI The calculation formula is as follows: ; in, These are the continuous wavelet transform coefficients of the thrust-direction force signal. These are the continuous wavelet transform coefficients of the twisting direction force signal. These are the amplitude coefficients of the cross wavelet transform.

[0013] Preferably, in S4, the improved analytic hierarchy process specifically includes: S41. Construct a hierarchical model that includes a target layer, a criterion layer, and an indicator layer; S42. Construct a fuzzy judgment matrix using the 0.1-0.9 scaling method and transform it into a fuzzy consistency matrix; S43. Calculate the initial subjective weights of each level of elements based on the fuzzy consistency matrix; S44. Combining expert scoring data, the entropy weight method is used to dynamically adjust the weights to obtain the basic weight coefficients. S45. Based on the type of acupoints to be assessed and the operating techniques, the basic weight coefficients are adaptively adjusted to obtain the final comprehensive weight coefficients.

[0014] Preferably, in S5, the fuzzy membership function is divided into three types according to the feature type: The optimal form feature is achieved using a symmetric Gaussian function: ; The larger the better characteristic uses a raised half-Gaussian function: ; The smaller the better feature, the more likely it is to use a reduced half-Gaussian function: ; In the formula, For the examinee i In the feature dimension, the th j Individual scores of sub-features For the examinee i In the feature dimension, the th jThe actual extracted feature values ​​of each sub-feature For the first i In the feature dimension, the th j Expert standard values ​​for individual features For the first i In the feature dimension, the th j Standard deviation of expert ratings for individual characteristics.

[0015] Preferably, in S6, the formula for calculating the comprehensive skills score is: ; in, For comprehensive skills assessment, For the first i The combined weight of each feature dimension, For the first i The number of sub-features contained in each dimension. For the first i In the dimension of the first j The overall weight of individual features.

[0016] Therefore, the beneficial effects of the above-mentioned objective assessment and scoring method for acupuncture skills based on force feedback time-series data in this invention are as follows: (1) The present invention is based entirely on objective force feedback time series data for assessment and scoring, avoiding the influence of subjective factors of experts. The scoring results have high consistency and repeatability, and can accurately distinguish acupuncture operators of different levels.

[0017] (2) This invention extracts features from four dimensions: time domain, frequency domain, time-frequency domain and synergy of techniques, which comprehensively reflects the mechanical characteristics, rhythmic characteristics, qi-delivery characteristics and synergistic characteristics of acupuncture operation, and can provide a comprehensive evaluation of acupuncture skills.

[0018] (3) This invention innovatively proposes the Qi Deqi Feature Index (GFI) and the Lifting-Twisting Synergy Index (CSI), realizing the objective quantitative evaluation of the two core acupuncture concepts of "Qi Deqi" and "manual synergy". At the same time, it proposes an improved hierarchical analysis method, which combines subjective experience, objective data and dynamic adjustment mechanism to realize the scientific allocation of feature weights.

[0019] (4) This invention is simple to operate and highly automated, which can greatly improve the assessment efficiency and reduce the assessment cost. It can be widely used in acupuncture teaching in Chinese medicine colleges, physician qualification examination and international acupuncture training and certification.

[0020] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0021] Figure 1This is a schematic diagram illustrating the steps of an embodiment of the objective assessment and scoring method for acupuncture skills based on force feedback time-series data according to the present invention. Detailed Implementation

[0022] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0023] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0024] Example 1: like Figure 1 As shown, this invention provides an objective assessment and scoring method for acupuncture skills based on force feedback time-series data, comprising the following steps: S1. Construct an acupuncture force feedback data acquisition system to collect raw three-dimensional force feedback time-series data under standard assessment movements, as detailed below: First, an acupuncture force feedback data acquisition system was constructed. This system includes a simulated human acupuncture model, three-dimensional force sensors installed at the acupoints on the model, a data acquisition card, and a computer processing unit. The three-dimensional force sensor adopts a high-precision strain gauge sensor with a sampling frequency set to 2000Hz, a force measurement range of 0-50N, and a measurement accuracy of 0.005N. It can simultaneously acquire force signals in three directions: X (twisting direction), Y (lifting and inserting direction), and Z (perpendicular needle insertion direction).

[0025] A three-dimensional force sensor is installed below the acupoint on a simulated human body model using a 5mm thick, 25±5 Shore A medical-grade silicone elastic medium. The sensor's force-bearing surface is parallel to the acupoint surface, ensuring lossless force signal transmission when the acupuncture needle is inserted vertically. Disposable sterile acupuncture needles with a diameter of 0.30mm and a length of 40mm are used uniformly for assessment. System calibration is performed before each assessment: zero-point calibration is performed under no-load conditions, and three-point sensitivity calibration is performed using standard weights (1N, 5N, 10N, 20N). The calibration error must be less than 0.02N. The data acquisition card provides microsecond-level timestamps to ensure synchronous acquisition of the three-dimensional force signal.

[0026] Before the assessment, the examinee was required to complete acupuncture procedures at designated acupoints according to standard operating procedures, including the three stages of needle insertion, manipulation, and withdrawal. The data acquisition system collected the raw three-dimensional force feedback timing data in real time during the operation and transmitted it to the computer processing unit for storage and subsequent processing.

[0027] S2. Preprocess the original force feedback time series data to obtain standardized segmented time series data.

[0028] The raw force feedback data contains a large amount of high-frequency noise, baseline drift, and operational anomalies, requiring preprocessing to improve data quality. Preprocessing specifically includes the following sub-steps: S21. Wavelet Thresholding Denoising: The original data is decomposed into 5 levels using the db4 wavelet. High-frequency coefficients are processed using a soft thresholding method, followed by wavelet reconstruction to remove high-frequency noise and baseline drift. The formula for calculating the soft thresholding function is: in, For the first j Layer k Wavelet coefficients, The threshold is calculated using the following formula: , The standard deviation of noise. N Data length. Noise standard deviation. The median absolute deviation method is used for estimation, and the calculation formula is as follows: ,in, These are the high-frequency coefficients of the first-level wavelet decomposition.

[0029] S22. Automatic Segmentation of Operation Stages: Based on the force change rate threshold method, the system automatically identifies and segments the operation into three stages: needle insertion, needle movement, and needle withdrawal. Specifically, it calculates the first derivative of the force signal; when the derivative exceeds a set needle insertion threshold... T At time 1, the needle insertion phase is considered to have begun; when the derivative is less than the set needle movement threshold... T 2. When the force signal enters a stable fluctuation state, it is determined that the needle-feeding stage has begun; when the derivative is less than the set needle-out threshold... T 3. When the force signal rapidly decreases to near zero, the needle withdrawal stage is considered to have begun. The stage segmentation threshold adopts an adaptive method combining a base threshold and an acupoint correction coefficient. The base threshold is... T 1 = 0.5 N / s , T 2 = 0.1 N / s , T 3 = -0.5 N / sThe correction factor for acupoints on the head and face is 0.8, and the correction factor for acupoints on the trunk is 1.2.

[0030] S23. Data Standardization: The data at each stage undergoes length normalization and amplitude normalization. Length normalization uses linear interpolation, standardizing the input stage to 1000 sampling points, the output stage to 60000 sampling points (corresponding to a 30-second output time), and the output stage to 500 sampling points. Amplitude normalization uses Z-score normalization, calculated using the following formula: in, The original data, This represents the mean of the data for this period. This represents the standard deviation of the data for this period.

[0031] S24. Operational Anomaly Detection and Correction: An improved Isolation Forest algorithm is used to detect anomalies during the operation process. This algorithm introduces mechanical constraints of acupuncture operation, using the range, rate of change, and duration of force signal variation as auxiliary features for anomaly judgment. The improved Isolation Forest algorithm introduces three mechanical constraints: ① The force value at a single sampling point does not exceed 50N; ② The absolute value of the force change rate does not exceed 100N / s; ③ The duration of the anomaly does not exceed 10ms. Points that simultaneously meet two or more constraints are judged as anomalies. For detected anomalies, linear interpolation is used for correction, and the calculation formula is: in, This is the force signal value corrected for the k-th anomaly point. For the first k The force signal value of the previous normal sampling point before the anomaly point. For the first k Force signal value of a normal sampling point after an anomaly point For the first k The timestamps corresponding to the anomalies For the first k The timestamp corresponding to the previous normal sampling point before each anomaly point. For the first k The timestamp corresponding to the next normal sampling point after each anomaly point.

[0032] S3. Extract multi-dimensional feature vectors from standardized time-series data, including time-domain dynamic features, frequency-domain rhythmic features, time-frequency domain coupling features, and technique synergy features, as detailed below: Multi-dimensional feature vectors are extracted from standardized segmented time-series data, including four dimensions: time-domain dynamic features, frequency-domain rhythm features, time-frequency domain coupling features, and manipulation synergy features, totaling 32 sub-features. The complete 32 sub-features include: 13 time-domain dynamic features, 3 frequency-domain rhythm features, 8 time-frequency domain coupling features (GFI, energy proportion of different time windows, frequency center offset, etc.), and 8 manipulation synergy features (CSI, phase synchronization index, synergy of different frequency bands, etc.).

[0033] ① Temporal dynamic feature extraction: Temporal dynamic features reflect the basic mechanical characteristics of acupuncture operation, including: needle insertion stage: average needle insertion speed, peak needle insertion force, force rise time, and needle insertion stability (standard deviation of force signal). Needle manipulation stage: lifting and thrusting amplitude, lifting and thrusting frequency, twisting angle, twisting frequency, standard deviation of force fluctuation, maximum force value, minimum force value; Needle withdrawal phase: average needle withdrawal speed, needle withdrawal time, residual force value.

[0034] Needle insertion smoothness is defined as the average of the standard deviations of the sliding window of the force signal during the needle insertion stage, with a window size of 50 sampling points.

[0035] ② Frequency Domain Rhythmic Feature Extraction. Frequency domain rhythmic features reflect the rhythmicity and stability of acupuncture procedures and are extracted using Fast Fourier Transform (FFT). Specifically, this includes: Main frequency: The frequency at which the power spectral density of the force signal is highest during the needle-walking phase; Main frequency band energy percentage: The proportion of energy within the main frequency ±0.5Hz range to the total energy; Harmonic distortion: The ratio of the sum of the energies of all harmonics to the fundamental frequency energy.

[0036] Harmonic distortion is calculated as the ratio of the sum of the energies of the first three harmonics (2f0, 3f0, 4f0) to the fundamental frequency energy.

[0037] ③ Time-frequency domain coupling feature extraction: Time-frequency domain coupling features reflect the frequency variation characteristics of the force signal at different time scales, especially the features related to "qi acquisition". This is achieved through continuous wavelet transform (…). CWT Extraction was performed using Morlet wavelet as the mother wavelet.

[0038] This embodiment proposes an air-gathering characteristic index ( GFI (This is used to quantify and evaluate the operator's ability to induce qi). GFI The calculation formula is: in, These are continuous wavelet transform coefficients. and This refers to the start and end times of the needle-carrying phase. and This is the characteristic frequency band of the gas extraction process. The sampling frequency.

[0039] The determination of the 2-8Hz characteristic frequency band of Qi was based on the analysis of electromyographic signals from 100 clinical patients with Qi. This frequency band was significantly correlated with the occurrence of Qi sensations such as soreness, numbness, distension, and heaviness (P<0.01). GFI The higher the value, the higher the proportion of energy of the force signal within the characteristic frequency band of Qi attainment, and the stronger the operator's ability to induce Qi attainment. Clinical studies have shown that when... GFI When the value is greater than 0.6, the patient's sensation of obtaining qi is significantly enhanced.

[0040] ④ Extraction of Technique Coordination Features: Technique coordination features reflect the degree of coordination between the two basic techniques of lifting and thrusting and twisting, and are an important indicator for evaluating advanced acupuncture skills. The cross-correlation between the force signals in the lifting / thrusting direction and the twisting direction is calculated using cross-wavelet transform.

[0041] This invention innovatively proposes the lifting-twisting synergy index (…). CSI The calculation formula is as follows: in, These are the continuous wavelet transform coefficients of the thrust-direction force signal. These are the continuous wavelet transform coefficients of the twisting direction force signal. These are the amplitude coefficients of the cross wavelet transform.

[0042] CSI The index reflects both the amplitude correlation and phase synchronization of the lifting and twisting techniques. When the phase difference of the cross wavelet transform coefficients is within ±π / 6, the two techniques are considered to be in a synchronized state. CSI The value ranges from 0 to 1; a higher value indicates better synergy between the lifting and twisting techniques. Clinical studies have shown that experienced acupuncturists... CSI The value is usually above 0.8, while for beginners... CSI The value is generally below 0.5.

[0043] S4. Construct a feature weight allocation model based on the improved analytic hierarchy process (AHP) to determine the weight coefficients of each feature dimension and sub-feature, as follows: S41. Construct a hierarchical model. Construct a three-layer hierarchical model comprising a target layer, a criterion layer, and an indicator layer: Target level: Comprehensive score of acupuncture skills; Criterion layer: time-domain dynamic characteristics, frequency-domain rhythmic characteristics, time-frequency domain coupling characteristics, and technique synergy characteristics; Indicator layer: Each criterion layer contains 32 sub-features.

[0044] S42. Constructing a Fuzzy Consistency Matrix: Ten acupuncture experts with over 20 years of clinical experience were invited to conduct pairwise comparisons of the relative importance of elements at each level, constructing a fuzzy judgment matrix using a 0.1-0.9 scaling method. Then, the fuzzy judgment matrix was transformed into a fuzzy consistency matrix using the following formula: in, The fuzzy consistency matrix is ​​the first... i Line 1 j The element values ​​of the column, For the fuzzy judgment matrix, the first... i The sum of all elements in the row, i.e. , For the fuzzy judgment matrix, the first... j The sum of all elements in the row, i.e. , is the order of the matrix.

[0045] S43. Calculate the initial weights: Calculate the relative weights of each level element based on the fuzzy consistency matrix. The calculation formula is as follows: in, Let be the initial subjective weight of the i-th element.

[0046] S44. Entropy Weight Method for Dynamic Weight Adjustment: Assessment data and expert scores from 100 acupuncturists of different levels are collected. The entropy weight method is used to calculate the objective weight of each feature. Then, the subjective weight and objective weight are merged to obtain the basic weight coefficient. The fusion formula is as follows: in, These are the base weighting coefficients after merging. This is the fusion coefficient between subjective and objective weights, with a value ranging from 0.4 to 0.6. For the first i The objective weights of each feature (calculated using the entropy weight method).

[0047] S45. Dynamic Weight Adaptive Adjustment: Based on the type of acupoint being assessed (e.g., limb acupoints, trunk acupoints, head and face acupoints) and the manipulation technique (e.g., lifting and thrusting method, twisting method, even tonification and sedation method), the basic weight coefficient is adaptively adjusted. The adjustment formula is as follows: in, This is the final overall weighting coefficient. For characteristic adjustment coefficients, This represents the coefficient for acupoint type. This represents the coefficient for the operating technique.

[0048] Establish a complete table of dynamic adjustment coefficients, including acupoint type coefficients. : Head and face 0.8, limbs 1.0, trunk 1.2; manipulation technique coefficient : Lifting and inserting method 1.3, twisting method 1.2, even tonification and sedation method 1.0; characteristic adjustment coefficient The correlation between features and therapeutic efficacy is determined, ranging from 0.8 to 1.3. For example, for acupoints on the head and face, due to the thinner skin and richer nerve supply, the speed and stability of needle insertion are more important; therefore, the corresponding feature adjustment coefficient is [not specified]. Take 1.2; for operations primarily using the thrusting and lifting method, the thrusting and lifting amplitude and frequency are more important, and the corresponding operation technique coefficients are... Take 1.3.

[0049] S5. Determine the expert standard value and standard deviation of each sub-feature, establish a single-feature scoring model based on fuzzy membership function, and calculate the individual score of each sub-feature, as follows: Based on the feature type, a single-feature scoring model is established using different forms of fuzzy membership functions. For each sub-feature, a standard value is determined based on standard operating data from 10 senior experts. and standard deviation .

[0050] Fuzzy membership functions are classified into three types based on feature type: The optimal form feature is achieved using a symmetric Gaussian function: The larger the better characteristic uses a raised half-Gaussian function: The smaller the better feature, the more likely it is to use a reduced half-Gaussian function: In the formula, For the examinee i In the feature dimension, the th j Individual scores of sub-features For the examinee i In the feature dimension, the th j The actual extracted feature values ​​of each sub-feature For the first i In the feature dimension, the th j Expert standard values ​​for individual features For the first i In the feature dimension, the th j Standard deviation of expert ratings for individual characteristics.

[0051] S6. Calculate the comprehensive skill score based on the feature weight coefficients and individual item scores, and generate an assessment analysis report, as follows: The overall skill score is calculated based on the feature weight coefficients and individual item scores. The formula for calculating the overall skill score is: in, The overall skills score is 100 points. For the first i The combined weight of each feature dimension, For the first i The number of sub-features contained in each dimension. For the first i In the dimension of the first j The overall weight of individual features.

[0052] An abnormal operation deduction mechanism has been added. For serious operational errors detected (such as needle insertion force exceeding 30N for more than 1 second, or needle withdrawal residual force exceeding 5N), 5-10 points will be deducted from the comprehensive score each time, until all points are deducted.

[0053] Based on the comprehensive scoring results, acupuncture skill levels are divided into four levels: Excellent: 90 points and above; Good: 80-89 points; Pass: 60-79 points; Fail: below 60 points.

[0054] At the same time, a detailed assessment and analysis report is generated, including an analysis of the advantages and disadvantages of each stage of operation, a comparative analysis with expert standards, and suggestions for improvement of weak links.

[0055] Based on the above technical solution, this embodiment has been verified, as follows: In this embodiment, 30 acupuncture practitioners of varying levels were invited, including 5 senior experts (with over 20 years of experience), 10 attending physicians (with 10-20 years of experience), 10 resident physicians (with 3-10 years of experience), and 5 interns (with less than 3 years of experience). The assessment was conducted at Hegu (LI4) (acupoint on the limbs), Zusanli (ST36) (acupoint on the lower limbs), and Baihui (GV20) (acupoint on the head and face). The manipulation techniques included lifting and thrusting, twisting, and balanced tonification and sedation.

[0056] The above method was used to assess and score each operator, and three senior experts were invited to conduct independent subjective assessments. The average score was taken as the expert score. Pearson correlation coefficient and intraclass correlation coefficient (ICC) were used to evaluate the consistency between the assessment results of this invention and the expert scores.

[0057] The results showed that in all combinations of acupoints and manipulation techniques, the Pearson correlation coefficient was above 0.92 and the ICC was above 0.90, indicating that the scoring results of the method of the present invention were highly consistent with the subjective scores of experts. Furthermore, the method of the present invention could accurately distinguish operators of different levels; the analysis of variance showed that the differences in scores between different level groups were statistically significant (P<0.001).

[0058] The comparison results of core feature values ​​show that the average value of senior experts is... GFI The value is 0.78±0.06, with an average of CSI The value was 0.85±0.04; the average value for interns was... GFI The value is 0.42±0.11, with an average of CSI The value was 0.48±0.12, and the difference between the two groups was statistically significant (P<0.001). After dynamic weight adjustment, the consistency between the scoring results and expert scores improved from 0.89 to 0.94, which is significantly better than the traditional fixed-weight method. Compared with existing assessment systems that only extract time-domain features, the scoring accuracy of this invention is improved by 18%.

[0059] Therefore, the present invention adopts the above-mentioned objective assessment and scoring method for acupuncture skills based on force feedback time sequence data, which solves the core pain points of traditional acupuncture skill assessment such as strong subjectivity, inconsistent standards, and inability to quantify "deqi". It realizes the full-process, high-precision, and objective assessment of acupuncture skills and can be widely applied to acupuncture teaching in TCM colleges, physician qualification examinations, and international acupuncture training and certification.

[0060] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. An objective assessment and scoring method for acupuncture skills based on force feedback time-series data, characterized by: Includes the following steps: S1. Construct an acupuncture force feedback data acquisition system to collect raw three-dimensional force feedback time sequence data under standard assessment movements; S2. Preprocess the original force feedback time series data to obtain standardized segmented time series data; S3. Extract multi-dimensional feature vectors from standardized time-series data, including time-domain dynamic features, frequency-domain rhythm features, time-frequency domain coupling features, and technique synergy features; S4. Construct a feature weight allocation model based on the improved analytic hierarchy process (AHP) and determine the weight coefficients of each feature dimension and sub-feature. S5. Determine the expert standard value and standard deviation of each sub-feature, establish a single-feature scoring model based on fuzzy membership function, and calculate the individual score of each sub-feature. S6. Calculate the comprehensive skill score based on the feature weight coefficients and individual scores, and generate an assessment analysis report.

2. The objective assessment and scoring method for acupuncture skills based on force feedback time-series data according to claim 1, characterized in that: In S1, the acupuncture force feedback data acquisition system includes: a simulated human acupuncture model, a three-dimensional force sensor installed at the acupoints of the model, a data acquisition card, and a computer processing unit; the sampling frequency of the three-dimensional force sensor is not less than 1000Hz, the force measurement range is 0-50N, and the measurement accuracy is not less than 0.01N.

3. The objective assessment and scoring method for acupuncture skills based on force feedback time-series data according to claim 1, characterized in that: In S2, the preprocessing of the original force feedback time series data specifically includes: S21. Wavelet thresholding denoising method is used to remove high-frequency noise and baseline drift from the original data; S22. Automatically identify and segment the three operation stages of needle insertion, needle movement, and needle withdrawal based on the force change rate threshold method; S23. Perform length normalization and amplitude standardization on the data of each stage; S24. An improved isolated forest algorithm is used to detect and correct operational anomalies.

4. The objective assessment and scoring method for acupuncture skills based on force feedback time-series data according to claim 1, characterized in that: In S3, the time-domain dynamic characteristics include: average needle insertion speed, peak needle insertion force, force rise time, and needle insertion stability during the needle insertion phase; lifting and thrusting amplitude, lifting and thrusting frequency, twisting angle, twisting frequency, force fluctuation standard deviation, maximum force value, and minimum force value during the needle movement phase; and average needle withdrawal speed, needle withdrawal time, and residual force value during the needle withdrawal phase. Among these, needle insertion stability is defined as the average value of the sliding window standard deviation of the force signal during the needle insertion phase, with a window size of 50 sampling points.

5. The objective assessment and scoring method for acupuncture skills based on force feedback time-series data according to claim 1, characterized in that: In S3, the frequency domain rhythm features are extracted through fast Fourier transform, including the main frequency, the main frequency band energy ratio, and the harmonic distortion during the needle-walking stage; where the harmonic distortion is the ratio of the sum of the first three harmonic energies to the fundamental frequency energy, and the main frequency band energy ratio is the proportion of energy within the range of the main frequency ± 0.5 Hz to the total energy.

6. The objective assessment and scoring method for acupuncture skills based on force feedback time-series data according to claim 1, characterized in that: In S3, the time-frequency domain coupling features are extracted using Morlet wavelet continuous wavelet transform, including the gas-gathering feature index. GFI The calculation formula is as follows: ； in, These are continuous wavelet transform coefficients. and This refers to the start and end times of the needle-carrying phase. and This is the characteristic frequency band of the gas extraction process. The sampling frequency.

7. The objective assessment and scoring method for acupuncture skills based on force feedback time-series data according to claim 6, characterized in that: In S3, the synergistic features of the manipulations are extracted through cross-wavelet transform, including the thrust-twist synergistic index. CSI The calculation formula is as follows: ； in, These are the continuous wavelet transform coefficients of the thrust-direction force signal. These are the continuous wavelet transform coefficients of the twisting direction force signal. These are the amplitude coefficients of the cross wavelet transform.

8. The objective assessment and scoring method for acupuncture skills based on force feedback time-series data according to claim 1, characterized in that: In S4, the improved analytic hierarchy process specifically includes: S41. Construct a hierarchical model that includes a target layer, a criterion layer, and an indicator layer; S42. Construct a fuzzy judgment matrix using the 0.1-0.9 scaling method and transform it into a fuzzy consistency matrix; S43. Calculate the initial subjective weights of each level of elements based on the fuzzy consistency matrix; S44. Combining expert scoring data, the entropy weight method is used to dynamically adjust the weights to obtain the basic weight coefficients. S45. Based on the type of acupoints to be assessed and the operating techniques, the basic weight coefficients are adaptively adjusted to obtain the final comprehensive weight coefficients.

9. The objective assessment and scoring method for acupuncture skills based on force feedback time-series data according to claim 1, characterized in that: In S5, fuzzy membership functions are divided into three types based on feature type: The optimal form feature is achieved using a symmetric Gaussian function: ； The larger the better characteristic uses a raised half-Gaussian function: ； The smaller the better feature, the more likely it is to use a reduced half-Gaussian function: ； In the formula, For the examinee i In the feature dimension, the th j Individual scores of sub-features For the examinee i In the feature dimension, the th j The actual extracted feature values ​​of each sub-feature For the first i In the feature dimension, the th j Expert standard values ​​for individual features For the first i In the feature dimension, the th j Standard deviation of expert ratings for individual characteristics.

10. The objective assessment and scoring method for acupuncture skills based on force feedback time-series data according to claim 9, characterized in that: In S6, the formula for calculating the comprehensive skills score is: ； in, For comprehensive skills assessment, For the first i The combined weight of each feature dimension, For the first i The number of sub-features contained in each dimension. For the first i In the dimension of the first j The overall weight of individual features.

Images