Functional scoring method and system based on ICF-RS and computer storage medium

Abstract

The invention discloses an ICF-RS-based function scoring method and system and a computer storage medium, and the method comprises the steps: obtaining to-be-scored ICF-RS evaluation data based on an ICF-RS evaluation standard; based on a preset first dichotomy model, obtaining a first classification result of the to-be-scored ICF-RS evaluation data; based on a preset second dichotomy model, obtaining a second classification result of the ICF-RS evaluation data to be scored; based on the first classification result and the second classification result, calculating the distance between the ICF-RS evaluation data to be scored and each function rating; and obtaining the function score of the ICF-RS evaluation data to be scored based on the distance between the ICF-RS evaluation data to be scored and each function score. According to the method, the bottleneck that only 30 item results exist and no overall result exists during ICF-RS application is broken through, ICF-RS overall function grade judgment is achieved, qualitative function level of a patient is achieved, and a user can visually know the overall function condition of the patient.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer-aided medical treatment, and more particularly to an ICF-RS-based function scoring method, system and computer storage medium. BACKGROUND

[0002] ICF (International Classification of Functioning, Disability and Health) is a tool for function and health classification proposed by the World Health Organization (WHO). In rehabilitation clinical and management work, the health status of the subjects needs to be assessed and their functions are graded. Previously, the international disease classification coding ICD-10 was used as the disease classification and diagnosis standard, but ICD-10 does not have relevant content for function evaluation. Therefore, ICD-11 is newly introduced, which adds a function evaluation chapter and provides a standardized function evaluation tool based on ICF.

[0003] The ICF rehabilitation set is a function evaluation rehabilitation set suitable for chronic disease patients in the rehabilitation process, which was introduced by WHO in 2016. It contains 30 items of body function, activity and participation function, all of which are included in the items of function evaluation in ICD-11. In order to localize the ICF rehabilitation set, after more than 10 years of clinical research and verification, the national standard ICF-RS (International Classification of Functioning, Disability and Health Rehabilitation Set) has been formed and the expert consensus has been published. ICF-RS can be applied to the standardized diagnosis, description and coding of diseases and functions of rehabilitation patients in China, and can realize the standardized quality management of health and health-related services such as evaluation, diagnosis, intervention and effect evaluation, data collection, statistics and application, and improve the quality and efficiency of health and health-related services for rehabilitation population, and ensure medical safety. With the help of rehabilitation medical big data in recent years, the payment weight of stroke patient function disorder grade can be obtained and the model can be constructed.

[0004] However, ICF is a classification tool in the initial development, and the evaluation standard of its function level is 5 levels (0-4), and the limited value classification is equidistant; while the frequency of occurrence is a percentage, and the distribution value is not equidistant, which causes that the evaluation results cannot be added and analyzed in clinical application. The unequal or mismatched characteristics of the limited value itself increase the difficulty of the clinical application and promotion of ICF. Therefore, to realize the model construction based on ICF-RS, it is necessary to break through the "bottleneck" of the unequal or mismatched characteristics of ICF itself. Some studies have converted the limited value by different methods. For example, the results of ICF categories are divided into normal or abnormal in 2-level classification, or a counting-based method is used, such as calculating the frequency and percentage of ranked categories. However, the problem has not been solved. This brings great inconvenience to the evaluation of the function level of patients. SUMMARY

[0005] The present application provides an ICF-RS-based function scoring method, system and computer storage medium, which solves the technical problem that the evaluation results of ICF itself cannot be added and analyzed in clinical application in the prior art.

[0006] To solve the above technical problems, the technical solutions of the present application are as follows: The first aspect of the present application provides an ICF-RS-based function scoring method, comprising the following steps: Based on the ICF-RS evaluation standard, the ICF-RS evaluation data to be scored is obtained; Based on a preset first two-classification model, a first classification result of the ICF-RS evaluation data to be scored is obtained, and the first two-classification model is used to classify the ICF-RS evaluation data to be scored into normal / mild / intermediate rating and severe rating; Based on a preset second two-classification model, a second classification result of the ICF-RS evaluation data to be scored is obtained, and the second two-classification model is used to classify the ICF-RS evaluation data to be scored into normal / mild rating and moderate / severe rating; Based on the first classification result and the second classification result, the distance of the ICF-RS evaluation data to be scored from each function rating is calculated; Based on the distance of the ICF-RS evaluation data to be scored from each function rating, the function score of the ICF-RS evaluation data to be scored is obtained.

[0007] Further, when obtaining the ICF-RS evaluation data to be scored, missing value processing is also included, and if the number of missing values exceeds a preset proportion threshold, the ICF-RS evaluation data to be scored is removed.

[0008] Furthermore, based on the preset first binary classification model, the first classification result of the ICF-RS evaluation data to be scored is obtained, including: The ICF-RS evaluation data to be scored is processed by the first extreme gradient boosting algorithm and the first support vector machine algorithm respectively to obtain the first output of the first extreme gradient boosting algorithm and the second output of the first support vector machine algorithm; The first and second outputs are processed by the first linear discriminant analysis algorithm to obtain the first classification result of the ICF-RS evaluation data to be scored.

[0009] Furthermore, based on a pre-defined second binary classification model, the second classification results of the ICF-RS evaluation data to be scored are obtained, including: The ICF-RS evaluation data to be scored are processed by the second limit gradient boosting algorithm and the second support vector machine algorithm to obtain the third output of the second limit gradient boosting algorithm and the fourth output of the second support vector machine algorithm. The third and fourth outputs are processed by the second linear discriminant analysis algorithm to obtain the second classification result of the ICF-RS evaluation data to be scored.

[0010] Furthermore, based on the first and second classification results, the distance between the ICF-RS assessment data to be scored and each functional rating is calculated, including: The first classification result is processed by the first Sigmoid function to obtain the first probability value; The second classification result is processed by the second Sigmoid function to obtain the second probability value; Based on the first probability value, the second probability value, and the preset order vector of each functional rating, the distance between the ICF-RS evaluation data to be rated and each functional rating is obtained.

[0011] Furthermore, based on the first probability value, the second probability value, and the preset order vector of each functional rating, the distance between the ICF-RS evaluation data to be rated and each functional rating is obtained, including:

[0012] In the formula, This indicates the distance between the ICF-RS assessment data to be scored and the first... Distance to the functional rating level , These represent the first probability value and the second probability value, respectively. , They represent the first The first and second elements of the order vector of the functional rating.

[0013] Furthermore, based on the distance between the ICF-RS assessment data to be scored and each functional rating, a functional score for the ICF-RS assessment data to be scored is obtained, including: The distance between the ICF-RS assessment data to be scored and each function rating is calculated using a Gaussian probability function, thus obtaining the probability that the ICF-RS assessment data to be scored belongs to each function rating. Based on the probability that the ICF-RS assessment data to be scored belongs to each functional rating, the functional score of the ICF-RS assessment data to be scored is obtained.

[0014] Furthermore, based on the probability that the ICF-RS assessment data to be scored belongs to each functional rating, the functional score of the ICF-RS assessment data to be scored is obtained, including:

[0015] In the formula, This represents the functional score of the ICF-RS assessment data to be scored. This indicates the probability that the ICF-RS assessment data to be scored belongs to the moderate functional rating. This indicates the probability that the ICF-RS assessment data to be scored belongs to the severe functional rating.

[0016] A second aspect of the present invention provides a functional scoring system based on ICF-RS, comprising: The acquisition module acquires the ICF-RS evaluation data to be scored based on the ICF-RS evaluation standard. The first classification module obtains the first classification result of the ICF-RS assessment data to be scored based on a preset first binary classification model. The first binary classification model is used to classify the ICF-RS assessment data to be scored into normal / mild / moderate and severe ratings. The second classification module, based on a preset second binary classification model, obtains the second classification result of the ICF-RS assessment data to be scored. The second binary classification model is used to classify the ICF-RS assessment data to be scored into normal / mild rating and moderate / severe rating. The first calculation module calculates the distance between the ICF-RS assessment data to be scored and each functional rating based on the first classification result and the second classification result. The second calculation module obtains the functional score of the ICF-RS evaluation data to be scored based on the distance between the ICF-RS evaluation data to be scored and each functional rating.

[0017] A third aspect of the present invention provides a computer storage medium storing a computer program, which, when executed by a processor, implements the functional scoring method based on ICF-RS.

[0018] Compared with the prior art, the beneficial effects of the technical solution of the present invention are: This invention, based on an artificial neural network (ANN) operational model, simulates the human information processing process by interconnecting and weighting internal nodes. It establishes a mapping relationship between ICF categories and functional classification results, realizing the conversion from input ICF-RS limit values ​​to output functional levels, and successfully constructing a neural network model for ICF-RS functional classification. This invention overcomes the bottleneck of ICF-RS applications, which only provide 30 item results without an overall result, enabling comprehensive ICF-RS functional level assessment and qualitatively defining the patient's functional level, allowing users to intuitively understand the patient's overall functional status. Attached Figure Description

[0019] Figure 1 A flowchart illustrating a functional scoring method based on ICF-RS provided in an embodiment of the present invention; Figure 2 A schematic diagram illustrating the construction process of the ICF-RS-based functional scoring model provided in this embodiment of the invention; Figure 3 This is a schematic diagram showing the frequency distribution of 30 category limit values ​​in ICF-RS provided in an embodiment of the present invention; Figure 4 Box plots of different functional level scores provided in embodiments of the present invention, wherein, Figure 4 (a) shows the results for the training set. Figure 4 (b) shows the results for the test set; Figure 5 This is a schematic diagram illustrating the score distribution of data points at different functional levels provided in an embodiment of the present invention. Figure 6 This is a schematic diagram of the Spearman correlation analysis results provided in an embodiment of the present invention; Figure 7 This is a schematic diagram of a functional scoring system based on ICF-RS provided in an embodiment of the present invention. Detailed Implementation

[0020] The accompanying drawings are for illustrative purposes only and should not be construed as limiting the scope of this patent. To better illustrate this embodiment, some parts in the accompanying drawings may be omitted, enlarged, or reduced, and do not represent the actual product dimensions; It will be understood by those skilled in the art that certain well-known structures and their descriptions may be omitted in the accompanying drawings.

[0021] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0022] Example 1 This embodiment provides a functional scoring method based on ICF-RS, such as Figure 1 As shown, it includes the following steps: Based on the ICF-RS evaluation criteria, obtain the ICF-RS evaluation data to be scored; Based on the preset first binary classification model, the first classification result of the ICF-RS assessment data to be scored is obtained. The first binary classification model is used to classify the ICF-RS assessment data to be scored into normal / mild / moderate and severe ratings. Based on the preset second binary classification model, the second classification result of the ICF-RS assessment data to be scored is obtained. The second binary classification model is used to classify the ICF-RS assessment data to be scored into normal / mild rating and moderate / severe rating. Based on the first and second classification results, calculate the distance between the ICF-RS assessment data to be scored and each functional rating; Based on the distance between the ICF-RS assessment data to be scored and each functional rating, the functional score of the ICF-RS assessment data to be scored is obtained.

[0023] This embodiment constructs a functional scoring model for stroke, such as Figure 2 As shown, two binary classification models are trained, where the first binary classification model is used to distinguish between normal / mild / moderate and severe stroke, and the second binary classification model is used to distinguish between normal / mild and moderate / severe stroke. Then, based on the outputs of the first and second binary classification models, the relative distances from the data to each functional level of stroke are calculated. Finally, a scoring model is constructed based on the relative distances from the data to each functional level to obtain the functional scores of the ICF-RS assessment data to be scored.

[0024] In a further embodiment, based on the ICF-RS evaluation criteria, the ICF-RS evaluation data to be scored is obtained, specifically as follows: Research subjects and inclusion / exclusion criteria: (1) Study subjects: Stroke patients who were hospitalized in the Rehabilitation Department of Xiangyang Central Hospital for rehabilitation treatment from January 1, 2021 to December 31, 2023 were selected.

[0025] (2) Patient inclusion criteria: ① Meets the diagnostic criteria for stroke and is confirmed by CT or MRI examination; ② Stable vital signs; ③ Age ≥ 18 years old; ④ Duration of illness ≥ 1 day; ⑤ Conscious and capable of cognitive ability (Simplified Intelligence Test ≥ 6 points) and able to cooperate in completing the assessment; ⑥ Capable of verbal communication and able to answer questions, or although unable to communicate verbally, possesses certain reading and writing abilities and can read the text of the questions and answer them; ⑦ The patient or family member is informed and has signed a consent form.

[0026] (3) Exclusion criteria for patients: those who cannot cooperate to complete the entire assessment process.

[0027] Evaluation indicators and methods: (1) National Standard ICF-RS Assessment: This part of the data is used for model construction and internal model validation. Three therapists with 3 or more years of clinical work experience, who have undergone unified and strict face-to-face training and passed the assessment, and who have obtained training qualification certificates, were selected. They were able to accurately and skillfully conduct professional assessments of each patient. The ICF-RS assessment of the research subjects was completed within 24 hours of admission: ① Assessment of the original limit values ​​of 30 categories (0: normal, 1: mild abnormality, 2: moderate abnormality, 3: severe abnormality, 4: completely problematic, 8: "not specifically mentioned" means that there is insufficient information to describe the severity of the problem, 9: "not applicable" means that the category is not applicable to the assessed subject), including three dimensions: physical function (9 categories), daily living (14 categories), and social participation (7 categories); ② Expert assessment of the three dimensions of physical function, activity, and participation, as well as the overall functional level. Based on the assessment results of the original limit values, the experts gave the patient's functional level (0: normal function, 1: mild functional impairment, 2: moderate functional impairment, 3: severe functional impairment).

[0028] Data preprocessing: (1) Limit values ​​8 and 9 are treated as missing values; (2) If more than 10% of the 30 categories in a sample are missing, the sample is removed.

[0029] In a further embodiment, based on a preset first binary classification model, the first classification result of the ICF-RS evaluation data to be scored is obtained, including: The ICF-RS evaluation data to be scored is processed by the first extreme gradient boosting algorithm and the first support vector machine algorithm respectively to obtain the first output of the first extreme gradient boosting algorithm and the second output of the first support vector machine algorithm; The first and second outputs are processed by the first linear discriminant analysis algorithm to obtain the first classification result of the ICF-RS evaluation data to be scored.

[0030] Based on a pre-defined second binary classification model, the second classification results of the ICF-RS assessment data to be scored are obtained, including: The ICF-RS evaluation data to be scored are processed by the second limit gradient boosting algorithm and the second support vector machine algorithm to obtain the third output of the second limit gradient boosting algorithm and the fourth output of the second support vector machine algorithm. The third and fourth outputs are processed by the second linear discriminant analysis algorithm to obtain the second classification result of the ICF-RS evaluation data to be scored.

[0031] In this embodiment, the ICF-RS category constraint values ​​belong to ordinal data for classification. The expected model output is a score of the functional level of stroke patients. Functional grading is ordinal categorical data, while scores are numerical data. Therefore, the classification model uses methods that can handle ordinal categorical data, such as Support Vector Machine (SVM), eXtremegradient boosting (XGBoost), Logistic regression (LR), and Linear discriminant analysis (LDA), to construct the stroke functional classification model; the scoring model is constructed based on the output of the classification model.

[0032] Two binary classification models are used to integrate the results of XGB and SVM using LDA. XGB and SVM each output real numbers representing the relative distance between each data point and a predefined class (normal / mild / moderate, moderate, normal / mild, moderate / severe). These output values ​​are used as input to LDA. After integrating the results of XGB and SVM, the LDA output better describes the relative distance between each data point and the predefined class. The real numbers output by the first model describe the relative position of each data point to the predefined class boundaries of the normal / mild / moderate and severe classes. The real numbers output by the second model describe the relative position of each data point to the predefined class boundaries of the normal / mild and moderate / severe classes. The parameters of the two binary classification models are as follows: First binary classification model: SVM:C=0.5, gamma=0.05,kernel='rbf' XGB:gamma=0,max_depth=3,min_child_weight=10,n_estimators=100,learning_rate=0.1.

[0033] Second binary classification model: SVM:C=0.9, gamma=0.01,kernel='rbf' XGB:reg_alpha=1,reg_lambda=0,gamma=3,max_depth=2,min_child_weight=6,n_estimators=100,learning_rate=0.1.

[0034] In a further embodiment, based on the first classification result and the second classification result, the distance between the ICF-RS assessment data to be scored and each functional rating is calculated, including: After processing the first classification result using the first Sigmoid function, a first probability value is obtained. The first probability value represents the probability that the ICF-RS assessment data to be scored is close to the severe level. ); After processing the second classification result using the second Sigmoid function, a second probability value is obtained. The second probability value represents the probability that the ICF-RS assessment data to be scored is close to the moderate / severe level. ); Based on the first probability value, the second probability value, and the preset order vector of each functional rating, the distance between the ICF-RS evaluation data to be rated and each functional rating is obtained, including:

[0035] In the formula, This indicates the distance between the ICF-RS assessment data to be scored and the first... Distance to the functional rating level , These represent the first probability value and the second probability value, respectively. , They represent the first The first and second elements of the order vector of the functional rating.

[0036] In a specific embodiment, the preset order vector of each function rating can be: This indicates Category 1 (normal / mild). This indicates Category 2 (moderate). This indicates Category 3 (severe).

[0037] In a further embodiment, based on the distance between the ICF-RS assessment data to be scored and each functional rating, a functional score for the ICF-RS assessment data to be scored is obtained, including: The distance between the ICF-RS assessment data to be scored and each function rating is calculated using a Gaussian probability function, thus obtaining the probability that the ICF-RS assessment data to be scored belongs to each function rating. Based on the probability that the ICF-RS assessment data to be scored belongs to each functional rating, the functional score of the ICF-RS assessment data to be scored is obtained, including:

[0038] In the formula, This represents the functional score of the ICF-RS assessment data to be scored. This indicates the probability that the ICF-RS assessment data to be scored belongs to the moderate functional rating. This indicates the probability that the ICF-RS assessment data to be scored belongs to the severe functional rating.

[0039] In this embodiment, the scoring model aims to provide more information for patients in a uniform classification. The model output score should satisfy the following: for data of different classifications, the score for a higher classification should be higher than the score for a lower classification. Therefore, we extended the previous classification model to calculate patient scores. First, based on calculating the distance from the ICF-RS assessment data to be scored to the three categories, we used a Gaussian probability function to calculate the weight of the ICF-RS assessment data belonging to each category. The weight values ​​were then normalized to ensure that the sum of the three weight values ​​was 1. The normalized weight values ​​are the probabilities that the ICF-RS assessment data belongs to the three categories. Next, the obtained probabilities were weighted and summed to calculate the score, and finally, the score was mapped to the range of 0-100.

[0040] This embodiment also verifies the constructed scoring model, specifically as follows: A 5-fold cross-validation method was used for internal validation of the model's performance. The original dataset was divided into five equal-sized subsets, four of which were used to train the model, and the remaining subset was used to test the model. This process was repeated five times; each time, a different subset was selected as the test set, and the rest were used as the training set. Finally, the average of the five performance evaluation results was taken to obtain the final evaluation result.

[0041] The performance evaluation metric for the scoring model is Grade Consistency (GC). Since the overall function and functional levels of the ICF-RS across its three dimensions correspond to the patient's original limit values, a higher grade (0-3) corresponds to a larger original limit value (0-4). Therefore, patients with higher grades should score higher than those with lower grades. The following calculations were performed to determine the proportion of patients with higher grades who scored lower than those with lower grades: make Represents the ICF-RS dataset. express The i-th sample in the series, express The j-th term in Experts indicated that... The assessment of the classification, Representation model pairs The output score, .

[0042] make The model represents the Consistency of levels in:

[0043] in Represents a set of evaluation pairs with different labels:

[0044] =

[0045] To further verify the clinical application value of the scoring model, this embodiment also included 120 stroke patients for correlation analysis between the scoring model output results and the results of commonly used behavioral scales for stroke.

[0046] Based on the principles of sample size calculation in medical research and the results of previous ICF-RS reliability and validity studies, PASS software was used to calculate the sample size. Assuming moderate and strong correlation for criterion validity, |r| was set to 0.3, the Power value to 0.80, and the α value to 0.05. The sample size estimated using PASS software should be no less than 82 cases. 120 stroke patients with concurrent hand, walking, and balance dysfunction who were hospitalized at Xiangyang Central Hospital between June 2023 and June 2024, and who underwent relevant behavioral scale assessments, were selected for analysis.

[0047] Inclusion criteria: 1. Meeting the relevant diagnostic criteria for stroke and confirmed by CT or MRI examination; 2. Stable vital signs; 3. First-time onset; 4. Duration of illness 1-6 months; 5. Age ≥18 years; 6. Presence of limb motor impairment; 7. Conscious and able to cooperate with assessment (Simplified Intelligence Test). 7 points), able to conduct daily verbal communication; 8. The patient or family member is informed and signs a consent form.

[0048] Exclusion criteria: those with critical illness and unstable vital signs; those with poor compliance who are unable to complete the trial or refuse to participate in the trial.

[0049] Research Process: ① Three therapists with 3 or more years of clinical experience were selected. They underwent standardized and rigorous face-to-face training, passed the assessment, and obtained training certificates. They were all able to accurately and skillfully apply the ICF-RS quantitative standards to conduct professional assessments of each patient. The study subjects were assessed using the ICF-RS, Fulg-Meyer Assessment Scale (FMA), Hemiplegic Hand Function Grading, Holden Walking Function Grading, Level 3 Balance Function Assessment, and Modified Barthel Index (MBI) within 24 hours of admission.

[0050] ICF-RS Results: Referring to the assignment method of ICF category limit values ​​1-4 corresponding to different levels of functional impairment in previous studies, the three dimensions and overall functional classification results given by the APP—mild functional impairment, moderate functional impairment, and severe functional impairment—were assigned values ​​of 1 to 3 respectively. Functional impairment behavioral assessment results: ① FMA score: Total score is 100 points (66 points for upper limbs, 34 points for lower limbs). The higher the score, the better the patient's motor function. Correlation analysis was performed based on the score results. ② Hemiplegic hand function classification: Based on the assessment classification results, the active hand A, active hand B, assistive hand A, assistive hand B, assistive hand C, and disuse hand were assigned values ​​of 1 to 6. ③ Holden gait function classification: Based on the assessment classification results, the following were assigned values: can walk independently anywhere, can walk independently on flat ground but requires assistance on stairs or slopes, requires one person to supervise or provide verbal guidance, requires one person to provide intermittent assistance, requires one person to provide continuous assistance, cannot walk or requires assistance from two or more people. The corresponding values ​​were assigned values ​​of 1 to 6. ④ Simple balance three-level assessment: Based on the assessment classification results, the following were assigned values: maintain balance under light external force, maintain dynamic balance for more than 10 seconds, maintain static balance for more than 10 seconds, and cannot maintain balance at all. The corresponding values ​​were assigned values ​​of 1 to 4. ⑤ MBI: This scale has 10 items, with a full score of 100 points. The higher the score, the better the patient's ability to live independently. 100 points indicates normal dependence, 71-99 points indicates mild dependence, 46-70 points indicates moderate dependence, 21-45 points indicates severe dependence, and 0-20 points indicates complete dependence, with corresponding values ​​of 1-5.

[0051] Spearman correlation analysis was used to analyze the correlation between the ICF-RS score and the scores of the functional impairment behavioral scales (FMA, hemiplegic hand function, Holden gait function, simple balance function, etc.) and the MBI scale. A |r| > 0.5 was considered a strong correlation, 0.1–0.3 a moderate correlation, and < 0.1 a weak correlation / no correlation. All values ​​were considered statistically significant (P < 0.05).

[0052] A total of 2,812 stroke patients were included in the model construction and validation. General information of the patients is detailed in Table 1.

[0053] Table 1 General Information of Stroke Patients

[0054] Note: N = 2812; IS, ischemic stroke; ICH, hemorrhagic stroke.

[0055] The frequency distribution of limit values ​​in 2812 stroke patients is shown in Table 2 and Figure 3 .

[0056] Table 2. Frequency distribution results of 30 category limits [n(%)]

[0057] The training and testing results of various performance indicators of the stroke functional score model are shown in Figure 3 and... Figure 4 .

[0058] like Figure 4 The box plot shown demonstrates that the main areas of different levels of functional impairment do not overlap, indicating that the ICF-RS scoring model can effectively distinguish different levels of functional impairment in most stroke patients.

[0059] Table 3. Accuracy Training and Testing Results of the Overall Functional Score Model for Stroke

[0060] Figure 5 The score distribution of data points corresponding to different functional levels is displayed. Data points include all training and test data from five tests, as well as fuzzy recognition data. Blue "x" marks indicate data assessed as mild or moderate by different doctors. Red "x" marks indicate data assessed as moderate or severe by different doctors. This illustrates that overlapping scores may be due to the model being affected by data quality; different evaluators or even the same evaluator at different times may use different evaluation criteria. The scoring model in this embodiment can provide evaluators with a stable reference, allowing them to make more consistent judgments. In the future, based on more stable and higher-quality data, the scoring quality of the scoring system will be even higher.

[0061] This embodiment further conducted a correlation analysis between the ICF-RS score and the results of the Stroke Functional Behavioral Scale, and found that the overall ICF-RS functional score was strongly correlated with the FMA total score, hemiplegic hand function grade, Holden gait function grade, sitting and standing balance grade, and MBI grade (P<0.05). See Table 4 for details. Figure 6 .

[0062] Table 4. Correlation analysis between ICF-RS scores and stroke functional behavioral scale results.

[0063] Note: HHFS: Hemiplegic Hand Functional Classification, HWFC: Holden Gait Functional Classification, MBI: Modified Barthel Index. The second embodiment of the present invention also provides a functional scoring system based on ICF-RS, such as Figure 7 As shown, it includes: The acquisition module acquires the ICF-RS evaluation data to be scored based on the ICF-RS evaluation standard. The first classification module obtains the first classification result of the ICF-RS assessment data to be scored based on a preset first binary classification model. The first binary classification model is used to classify the ICF-RS assessment data to be scored into normal / mild / moderate and severe ratings. The second classification module, based on a preset second binary classification model, obtains the second classification result of the ICF-RS assessment data to be scored. The second binary classification model is used to classify the ICF-RS assessment data to be scored into normal / mild rating and moderate / severe rating. The first calculation module calculates the distance between the ICF-RS assessment data to be scored and each functional rating based on the first classification result and the second classification result. The second calculation module obtains the functional score of the ICF-RS evaluation data to be scored based on the distance between the ICF-RS evaluation data to be scored and each functional rating.

[0064] A third embodiment of the present invention also provides a computer storage medium storing a computer program, which, when executed by a processor, implements the functional scoring method based on ICF-RS.

[0065] The same or similar labels correspond to the same or similar parts; The terms used to describe positional relationships in the accompanying drawings are for illustrative purposes only and should not be construed as limiting this patent. Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art can make other variations or modifications based on the above description. It is neither necessary nor possible to exhaustively describe all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A functional scoring method based on ICF-RS, characterized in that, Includes the following steps: Based on the ICF-RS evaluation criteria, obtain the ICF-RS evaluation data to be scored; Based on the preset first binary classification model, the first classification result of the ICF-RS assessment data to be scored is obtained. The first binary classification model is used to classify the ICF-RS assessment data to be scored into normal / mild / moderate and severe ratings. Based on the preset second binary classification model, the second classification result of the ICF-RS assessment data to be scored is obtained. The second binary classification model is used to classify the ICF-RS assessment data to be scored into normal / mild rating and moderate / severe rating. Based on the first and second classification results, calculate the distance between the ICF-RS assessment data to be scored and each functional rating; Based on the distance between the ICF-RS assessment data to be scored and each functional rating, the functional score of the ICF-RS assessment data to be scored is obtained.

2. The functional scoring method based on ICF-RS according to claim 1, characterized in that, When acquiring ICF-RS assessment data to be scored, the process also includes handling missing values. If the number of missing values ​​exceeds a preset proportion threshold, the ICF-RS assessment data to be scored is removed.

3. The functional scoring method based on ICF-RS according to claim 1, characterized in that, Based on the preset first binary classification model, the first classification result of the ICF-RS evaluation data to be scored is obtained, including: The ICF-RS evaluation data to be scored is processed by the first extreme gradient boosting algorithm and the first support vector machine algorithm respectively to obtain the first output of the first extreme gradient boosting algorithm and the second output of the first support vector machine algorithm; The first and second outputs are processed by the first linear discriminant analysis algorithm to obtain the first classification result of the ICF-RS evaluation data to be scored.

4. The functional scoring method based on ICF-RS according to claim 1, characterized in that, Based on a pre-defined second binary classification model, the second classification results of the ICF-RS assessment data to be scored are obtained, including: The ICF-RS evaluation data to be scored are processed by the second limit gradient boosting algorithm and the second support vector machine algorithm to obtain the third output of the second limit gradient boosting algorithm and the fourth output of the second support vector machine algorithm. The third and fourth outputs are processed by the second linear discriminant analysis algorithm to obtain the second classification result of the ICF-RS evaluation data to be scored.

5. The functional scoring method based on ICF-RS according to any one of claims 1 to 4, characterized in that, Based on the first and second classification results, the distance between the ICF-RS assessment data to be scored and each functional rating is calculated, including: The first classification result is processed by the first Sigmoid function to obtain the first probability value; The second classification result is processed by the second Sigmoid function to obtain the second probability value; Based on the first probability value, the second probability value, and the preset order vector of each functional rating, the distance between the ICF-RS evaluation data to be rated and each functional rating is obtained.

6. The functional scoring method based on ICF-RS according to claim 5, characterized in that, Based on the first probability value, the second probability value, and the preset order vector of each functional rating, the distance between the ICF-RS evaluation data to be rated and each functional rating is obtained, including: In the formula, This indicates the distance between the ICF-RS assessment data to be scored and the first... Distance to the functional rating level , These represent the first probability value and the second probability value, respectively. , They represent the first The first and second elements of the order vector of the functional rating.

7. The functional scoring method based on ICF-RS according to claim 6, characterized in that, Based on the distance between the ICF-RS assessment data to be scored and each functional rating, the functional score of the ICF-RS assessment data to be scored is obtained, including: The distance between the ICF-RS assessment data to be scored and each function rating is calculated using a Gaussian probability function, thus obtaining the probability that the ICF-RS assessment data to be scored belongs to each function rating. Based on the probability that the ICF-RS assessment data to be scored belongs to each functional rating, the functional score of the ICF-RS assessment data to be scored is obtained.

8. The functional scoring method based on ICF-RS according to claim 7, characterized in that, Based on the probability that the ICF-RS assessment data to be scored belongs to each functional rating, the functional score of the ICF-RS assessment data to be scored is obtained, including: In the formula, This represents the functional score of the ICF-RS assessment data to be scored. This indicates the probability that the ICF-RS assessment data to be scored belongs to the moderate functional rating. This indicates the probability that the ICF-RS assessment data to be scored belongs to the severe functional rating.

9. A functional scoring system based on ICF-RS, characterized in that, include: The acquisition module acquires the ICF-RS evaluation data to be scored based on the ICF-RS evaluation standard. The first classification module obtains the first classification result of the ICF-RS assessment data to be scored based on a preset first binary classification model. The first binary classification model is used to classify the ICF-RS assessment data to be scored into normal / mild / moderate and severe ratings. The second classification module, based on a preset second binary classification model, obtains the second classification result of the ICF-RS assessment data to be scored. The second binary classification model is used to classify the ICF-RS assessment data to be scored into normal / mild rating and moderate / severe rating. The first calculation module calculates the distance between the ICF-RS assessment data to be scored and each functional rating based on the first classification result and the second classification result. The second calculation module obtains the functional score of the ICF-RS evaluation data to be scored based on the distance between the ICF-RS evaluation data to be scored and each functional rating.

10. A computer storage medium, characterized in that, The computer storage medium stores a computer program, which, when executed by a processor, implements the ICF-RS-based functional scoring method according to any one of claims 1 to 8.

Images