Method, device, equipment and medium for predicting risk of elderly dysphagia patient

Abstract

This application relates to a method, device, equipment, and medium for risk prediction in elderly patients with dysphagia. The method includes: acquiring a multidimensional assessment dataset of the subject; the multidimensional assessment dataset includes static risk screening data based on the EAT-10 dietary assessment scale, dynamic swallowing function data based on a modified volumetric viscosity test, swallowing performance data based on the Kubota water swallowing test, and the subject's basic characteristic data; preprocessing the multidimensional assessment dataset to obtain a standardized swallowing assessment dataset; extracting static risk features and dynamic functional features from the standardized swallowing assessment dataset and fusing them to obtain a fused feature vector; inputting the fused feature vector into a preset aspiration risk prediction model to obtain the subject's aspiration risk level; the aspiration risk level includes low risk, intermediate risk, and high risk. This method can achieve efficient prediction of aspiration risk in elderly patients with dysphagia.

Description

Technical Field

[0001] This invention belongs to the field of health management technology, and in particular relates to a method, device, equipment and medium for risk prediction of elderly patients with dysphagia. Background Technology

[0002] In the field of elderly health management and medical monitoring, dysphagia risk prevention and intervention technologies are receiving increasing attention. Around the core needs of risk prediction and reflux monitoring, a multi-dimensional technical system has been formed, encompassing clinical assessment, data modeling, and sensor monitoring. Risk prediction technology, by identifying relevant risk factors and constructing predictive models, provides support for early screening of high-risk individuals. Reflux monitoring technology focuses on the real-time detection of esophageal dysfunction during swallowing, aiming to prevent serious complications such as aspiration and aspiration pneumonia.

[0003] In traditional techniques, risk assessment of dysphagia in the elderly often relies on manual clinical assessment methods such as the Kubota water swallowing test and dietary assessment scales. Results are obtained through on-site operations by medical staff or subjective feedback from patients to determine the condition. Esophageal reflux monitoring often employs single-point sensing, static threshold control, and periodic model updates. Some technologies combine pH and pressure sensors to trigger balloon expansion for reflux protection. Regarding risk prediction modeling, existing research mainly focuses on hospitalized patients. Models are often based on single algorithms or limited feature variables, lacking targeted optimization for elderly populations in the community. Data processing often uses fixed inclusion and exclusion criteria and static threshold determination, failing to fully consider individual differences and dynamic changes in physiological states in elderly patients.

[0004] However, existing methods have significant problems: First, traditional manual assessment is inefficient and highly subjective, making it difficult to conduct large-scale screening of elderly people in the community. Furthermore, early swallowing difficulties are often subtle and easily overlooked as natural signs of aging, leading to missed diagnoses of high-risk individuals. Second, existing risk prediction models are limited to hospital settings, failing to comprehensively identify risk factors in elderly people in the community. They do not fully integrate key features across multiple dimensions, such as age, grip strength, oral health, and underlying diseases. Moreover, the modeling algorithms are simplistic, feature selection is inaccurate, and prediction accuracy and clinical applicability are limited. Third, the single-point sensing and static threshold control of reflux monitoring systems cannot adapt to the characteristics of elderly patients, such as esophageal elasticity deterioration, frequent positional changes, and low swallowing amplitude. This results in insensitive capture of low-amplitude swallowing and early reflux signals, and a disconnect between cuff inflation pressure and actual reflux risk, posing a risk of accidental blockage or excessive pressure causing esophageal mucosal damage. Fourth, traditional model updates employ full retraining and fixed-period updates, which cannot quickly respond to dynamic physiological characteristics such as patient positional changes and esophageal elasticity variations. The long model iteration cycle and high resource consumption make it difficult to achieve individualized, real-time risk warnings and protection. Summary of the Invention

[0005] Therefore, it is necessary to provide a risk prediction method for elderly patients with dysphagia in response to the above-mentioned technical problems.

[0006] Firstly, this application provides a risk prediction method for elderly patients with dysphagia, including:

[0007] Obtain multidimensional assessment datasets of subjects; the multidimensional assessment datasets include static risk screening data based on the dietary assessment scale EAT-10, dynamic swallowing function data based on the modified volumetric viscosity test, swallowing performance data based on the Kubota water swallowing test, and basic characteristic data of subjects;

[0008] The multidimensional assessment dataset is preprocessed to obtain a standardized swallowing assessment dataset;

[0009] Static risk features and dynamic functional features were extracted from a standardized swallowing assessment dataset, and the static risk features and dynamic functional features were fused to obtain a fused feature vector;

[0010] The fused feature vector is input into a pre-defined aspiration risk prediction model to obtain the aspiration risk level of the subject; the aspiration risk level includes low risk, medium risk and high risk.

[0011] In one embodiment, static risk features and dynamic functional features are extracted from a standardized swallowing assessment dataset, including:

[0012] The EAT-10 total score is extracted from the static risk screening data, and the EAT-10 total score is determined as the first static risk characteristic.

[0013] The swallowing performance data were graded to obtain the Kubota drinking water test grade, and the Kubota drinking water test grade was determined as the second static risk characteristic;

[0014] Based on the first static risk characteristic and the second static risk characteristic, the static risk characteristic is obtained;

[0015] Based on dynamic swallowing function data, dynamic functional characteristics are obtained; dynamic functional characteristics include swallowing safety indicators and swallowing efficiency indicators; swallowing safety indicators include at least one of the following under a specific consistency and volume test combination: whether the absolute value of blood oxygen saturation is lower than a preset threshold and whether choking occurs; swallowing efficiency indicators include the swallowing time parameter required to swallow a specific volume of liquid once and the degree of residue determined by auscultation of the larynx and neck after swallowing.

[0016] In one embodiment, dynamic functional characteristics are obtained based on dynamic swallowing function data, including:

[0017] The system analyzes the blood oxygen saturation monitoring signal in dynamic swallowing function data, identifies safety events where blood oxygen saturation drops beyond a preset threshold, and constructs swallowing safety indicators based on these events.

[0018] Endpoint detection is performed on the swallowing audio signal in the dynamic swallowing function data to generate swallowing time parameters;

[0019] The residual level is generated based on the laryngeal auscultation results recorded by clinical assessors during the modified volumetric viscosity test;

[0020] Based on swallowing time parameters and residue level, a swallowing efficiency index is constructed;

[0021] Dynamic functional characteristics were obtained based on swallowing safety and swallowing efficiency indicators.

[0022] In one embodiment, endpoint detection is performed on the swallowing audio signal in the dynamic swallowing function data to generate swallowing time parameters, including:

[0023] Use the following formula to calculate the swallowing time parameter:

[0024]

[0025] in, It is a swallowing time parameter. It is the time when the swallowing action begins. It is the time when the swallowing action ends. It is in time The power spectral density of the swallowing audio signal at the location, It is the integral of the power spectral density of the swallowed audio signal. It is the maximum value of the power spectral density of the swallowed audio signal.

[0026] In one embodiment, the method further includes:

[0027] Based on the risk level of aspiration, a personalized graded intervention plan is generated; the personalized graded intervention plan includes at least one of the following: health education and guidance plan, food texture modification plan, compensatory swallowing strategy training plan, and systemic rehabilitation treatment plan.

[0028] In one embodiment, the multidimensional assessment dataset is preprocessed to obtain a standardized swallowing assessment dataset, including:

[0029] Based on the indicator types of the multidimensional evaluation dataset, the missing data in the multidimensional evaluation dataset is divided into static indicator missing data and dynamic indicator missing data.

[0030] For missing static indicator data, the random forest algorithm is used in conjunction with the feature distribution of the same age group and the same underlying disease group to imputate the missing static indicator data.

[0031] One-hot encoding is performed on the categorical features in the completed static index data to obtain standardized static data.

[0032] To address the missing dynamic indicator data, based on the consistency-volume combination scenario of the improved volumetric viscosity test, the matching degree data of the subjects' age and basic swallowing ability in the same scenario are extracted, and the scenario-based distribution fitting algorithm is used for interpolation to obtain the completed dynamic indicator data.

[0033] Based on the completed dynamic indicator data, standardized dynamic data is obtained;

[0034] The standardized static data and standardized dynamic data are subjected to dimensionality normalization to obtain a standardized swallowing assessment dataset.

[0035] In one embodiment, standardized dynamic data is obtained based on the completed dynamic indicator data, including:

[0036] The swallowing action temporal features in the completed dynamic index data are subjected to hierarchical mapping to obtain temporal hierarchical features.

[0037] The food trait adaptation features in the completed dynamic index data are processed by rule judgment to obtain the adaptation grading features.

[0038] Integrate temporal hierarchical features and adaptability hierarchical features to generate standardized dynamic data.

[0039] Secondly, this application also provides a risk prediction device for elderly patients with dysphagia, comprising:

[0040] The data acquisition module is used to acquire multidimensional assessment datasets of the subjects. The multidimensional assessment datasets include static risk screening data based on the dietary assessment scale EAT-10, dynamic swallowing function data based on the modified volumetric viscosity test, swallowing performance data based on the Kubota water swallowing test, and basic characteristic data of the subjects.

[0041] The preprocessing module is used to preprocess the multi-dimensional assessment dataset to obtain a standardized swallowing assessment dataset;

[0042] The feature extraction module is used to extract static risk features and dynamic functional features from the standardized swallowing assessment dataset, and then fuse the static risk features and dynamic functional features to obtain a fused feature vector;

[0043] The risk level prediction module is used to input the fused feature vector into a preset aspiration risk prediction model to obtain the aspiration risk level of the subject; the aspiration risk level includes low risk, medium risk and high risk.

[0044] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of any of the methods described above.

[0045] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the methods described above.

[0046] The aforementioned method, device, equipment, and medium for predicting the risk of aspiration in elderly patients with dysphagia involves acquiring a multi-dimensional assessment dataset of the subject. This dataset includes static risk screening data based on the EAT-10 dietary assessment scale, dynamic swallowing function data based on a modified volumetric viscosity test, swallowing performance data based on the Kubota water swallowing test, and the subject's basic characteristic data. The multi-dimensional assessment dataset is preprocessed to obtain a standardized swallowing assessment dataset. Static risk features and dynamic functional features are extracted from the standardized swallowing assessment dataset and fused to obtain a fused feature vector. This fused feature vector is input into a pre-defined aspiration risk prediction model to obtain the subject's aspiration risk level, which is categorized as low, intermediate, and high risk. This method enables efficient prediction of aspiration risk in elderly patients with dysphagia. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 A flowchart illustrating a risk prediction method for elderly patients with dysphagia, provided as an exemplary embodiment of this application;

[0049] Figure 2 This is a schematic diagram of a risk prediction device for elderly patients with dysphagia, provided as an exemplary embodiment of this application. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0051] The risk prediction method for elderly patients with dysphagia provided in this application embodiment can be used in medical diagnosis and assessment, daily care assessment, feeding care, nutritional support, health management monitoring, and scientific research and education.

[0052] In one embodiment, such as Figure 1 As shown, a risk prediction method for elderly patients with dysphagia is provided. This embodiment illustrates the application of this method to a prediction terminal. It is understood that this method can also be applied to a server, or to a system including both a terminal and a server, and is implemented through the interaction between the prediction terminal and the server. In this embodiment, the method includes the following steps:

[0053] Step S101: Obtain the multidimensional assessment dataset of the subjects; the multidimensional assessment dataset includes static risk screening data based on the dietary assessment scale EAT-10, dynamic swallowing function data based on the modified volumetric viscosity test, swallowing performance data based on the Kubota water swallowing test, and basic characteristic data of the subjects.

[0054] The multidimensional assessment dataset can be a comprehensive collection of data used to assess the risk of elderly patients with dysphagia, consisting of structured and unstructured multi-source clinical and functional assessment data.

[0055] Static risk screening data based on the EAT-10 dietary assessment scale can be quantitative data obtained through the EAT-10 scale that reflects the patient's subjective feelings and functional self-assessment of swallowing-related symptoms.

[0056] Dynamic swallowing function data based on modified volumetric viscosity testing can be physiological and behavioral signal data that are synchronously monitored and recorded by instruments during the modified volumetric viscosity testing process, and are used to reflect the real-time safety and efficiency of swallowing actions.

[0057] Swallowing performance data based on the Kubota water swallowing test can be used to quickly grade a patient's swallowing function by obtaining standard Kubota water swallowing test procedures.

[0058] The basic characteristic data of the subjects can be basic static data that reflects the overall health status and individual differences of the elderly subjects.

[0059] Specifically, the predictive terminal can extract structured data such as the total score of the EAT-10 scale, the Kubota water swallowing test grade results, the age of the subject, and underlying diseases from standardized electronic medical records or dedicated swallowing function assessment records. At the same time, the predictive terminal can connect to or import time-series signal files generated by blood oxygen saturation monitors and audio acquisition devices during modified volumetric viscosity tests, thereby generating a complete assessment dataset including static risk screening, dynamic swallowing function, swallowing performance, and basic characteristics.

[0060] Step S102: Preprocess the multidimensional assessment dataset to obtain a standardized swallowing assessment dataset.

[0061] The standardized swallowing assessment dataset can be a structured swallowing risk assessment dataset with a uniform format, consistent dimensions, and direct applicability for feature extraction and model input.

[0062] Specifically, the prediction terminal can classify missing items in the multi-dimensional assessment dataset according to indicator type, then standardize the classification features, and perform normalization operations on data of different dimensions and scales to finally obtain a standardized swallowing assessment dataset.

[0063] Step S103: Extract static risk features and dynamic functional features from the standardized swallowing assessment dataset, and fuse the static risk features and dynamic functional features to obtain a fused feature vector.

[0064] Among them, static risk characteristics can be swallowing disorder risk-related indicators extracted from the assessment dataset that reflect the subject's relatively stable or non-changing over time.

[0065] Dynamic functional characteristics can be indicators extracted from assessment datasets that reflect the real-time functional and safety performance of subjects during actual swallowing.

[0066] A fused feature vector can be a high-dimensional feature data structure generated by integrating extracted static risk features and dynamic functional features through a feature fusion algorithm.

[0067] Specifically, the predictive terminal can screen and extract swallowing disorder risk-related indicators that reflect the subject's relatively stable or non-changing swallowing risk from a standardized swallowing assessment dataset, generating static risk features. Then, it extracts indicators that characterize the subject's real-time function and safety during actual swallowing, obtaining dynamic functional features. Subsequently, the predictive terminal integrates the static risk features and dynamic functional features using a pre-set feature fusion algorithm, ultimately generating a fused feature vector. Illustratively, the pre-set feature fusion algorithm can be a weighted feature concatenation fusion algorithm. This algorithm first assigns independent weights to the static risk features and dynamic functional features, with weight coefficients determined by the contribution of the two types of features in historical prediction data. The basic weight of the static risk features is based on the clinical diagnostic correlation of the EAT-10 scale and the Kubota water swallowing test, while the weight of the dynamic functional features is dynamically adjusted based on the aspiration risk correlation of swallowing safety and efficiency indicators. Then, the weighted features are sequentially concatenated according to dimensions to form an initial fused feature matrix. Finally, principal component analysis is used to optimize the matrix by dimensionality reduction, eliminating redundant feature dimensions and generating a fused feature vector with simplified dimensions and complete information.

[0068] Step S104: Input the fused feature vector into the preset aspiration risk prediction model to obtain the aspiration risk level of the subject; the aspiration risk level includes low risk, medium risk and high risk.

[0069] The pre-defined aspiration risk prediction model can be a lightweight classification model based on multi-feature fusion. This lightweight classification model structure includes an input layer, a feature adaptation layer, a classification layer, and an output layer. The input layer receives the fused feature vector. The feature adaptation layer uses a fully connected network to perform dimensionality mapping and information enhancement on the static and dynamic fused features. The classification layer uses a hybrid architecture combining logistic regression and a lightweight neural network to achieve risk grading. The output layer outputs the risk category. The core function of this model is to integrate multi-dimensional swallowing risk features. The input is a high-dimensional vector resulting from the fusion of static and dynamic features, and the output is three categories of aspiration risk: low, medium, and high. This determines the subject's aspiration risk level, achieving both predictive accuracy and clinical suitability.

[0070] The aspiration risk level can be a classification of the likelihood of aspiration (food or liquid entering the airway) in elderly patients with dysphagia, based on the output of an aspiration risk prediction model.

[0071] Specifically, the prediction terminal can input the generated fusion feature vector into the preset aspiration risk prediction model. The model will eventually output the risk level corresponding to the subject through the output layer. Based on this, the prediction terminal can obtain the subject's aspiration risk level. The aspiration risk level is a classification of the subject's likelihood of aspiration after eating, specifically covering low risk, medium risk, and high risk.

[0072] In the aforementioned risk prediction method for elderly patients with dysphagia, the prediction terminal acquires a multi-dimensional assessment dataset of the subject. This dataset includes static risk screening data based on the EAT-10 dietary assessment scale, dynamic swallowing function data based on a modified volumetric viscosity test, swallowing performance data based on the Kubota water swallowing test, and the subject's basic characteristic data. The multi-dimensional assessment dataset is preprocessed to obtain a standardized swallowing assessment dataset. Static risk features and dynamic functional features are extracted from the standardized swallowing assessment dataset and fused to obtain a fused feature vector. This fused feature vector is input into a pre-defined aspiration risk prediction model to obtain the subject's aspiration risk level, which includes low, intermediate, and high risk. This method enables efficient prediction of aspiration risk in elderly patients with dysphagia.

[0073] In one embodiment, extracting static risk features and dynamic functional features from a standardized swallowing assessment dataset may include the following steps:

[0074] Step S201: Extract the EAT-10 total score from the static risk screening data and determine the EAT-10 total score as the first static risk feature.

[0075] The total EAT-10 score is a quantitative value obtained by summing the scores of the 10 items on the EAT-10 diet assessment scale after screening elderly subjects for subjective symptoms of dysphagia.

[0076] Specifically, the prediction terminal can retrieve static risk screening data from the standardized swallowing assessment dataset, locate the relevant item score information of the diet assessment scale EAT-10, accumulate the scores of the 10 items according to the scale scoring rules, calculate the total score of EAT-10, and then the prediction terminal will identify the features and solidify the dimensions of the total score of EAT-10, and determine it as the first static risk feature.

[0077] Step S202: The swallowing performance data is graded to obtain the Kubota drinking test grade, and the Kubota drinking test grade is determined as the second static risk characteristic.

[0078] The Kubota water swallowing test grade can be the result of a graded assessment of the subject's swallowing function using the Kubota water swallowing test, a standardized clinical assessment method.

[0079] Specifically, the prediction terminal can retrieve swallowing performance data from the standardized swallowing assessment dataset, then extract the subject's drinking process record information related to the Kubota drinking test from the swallowing performance data, and then classify the subject's swallowing performance according to the standardized grading rules of the Kubota drinking test to generate the corresponding Kubota drinking test level.

[0080] Furthermore, the prediction terminal performs feature classification and data formatting on the drinking water test level of the depression, and identifies it as the second static risk feature.

[0081] Step S203: Based on the first static risk feature and the second static risk feature, obtain the static risk feature.

[0082] Specifically, the prediction terminal can retrieve the determined first static risk feature and second static risk feature, and then combine the first static risk feature and second static risk feature in a structured way and unify the dimensions to eliminate the differences in the scale and format barriers between different features. Then, the integrated feature data is validated to remove invalid or abnormal data, and finally static risk features are generated.

[0083] Step S204: Based on dynamic swallowing function data, obtain dynamic functional characteristics; dynamic functional characteristics include swallowing safety indicators and swallowing efficiency indicators; swallowing safety indicators include at least one of the following under a specific consistency and volume test combination: whether the absolute value of blood oxygen saturation is lower than a preset threshold and whether choking occurs; swallowing efficiency indicators include the swallowing time parameter required to swallow a specific volume of liquid once and the degree of residue determined by auscultation of the larynx and neck after swallowing.

[0084] Specifically, the predictive terminal can generate dynamic functional features that include dimensions related to swallowing safety and efficiency by using dynamic swallowing function data from a standardized swallowing assessment dataset through signal analysis, index extraction, and feature integration.

[0085] In this embodiment, the prediction terminal achieves refined feature mining of the standardized swallowing assessment dataset by extracting and integrating static risk features and generating dynamic functional features, thereby improving the scientificity and reliability of the overall risk prediction process.

[0086] In one embodiment, obtaining dynamic functional characteristics based on dynamic swallowing function data may include the following steps:

[0087] Step S301: Analyze the blood oxygen saturation monitoring signal in the dynamic swallowing function data, identify safety events where blood oxygen saturation drops beyond a preset threshold, and construct swallowing safety indicators based on these safety events.

[0088] Among them, a safety event can be an abnormal physiological event identified from the blood oxygen saturation monitoring signal of dynamic swallowing function data during the modified volumetric viscosity test, which indicates that there is a risk of airway aspiration during the swallowing process of the subject, specifically referring to a situation where the decrease in blood oxygen saturation exceeds a preset threshold.

[0089] Swallowing safety indicators can be used to quantitatively assess the airway safety of subjects during real-time swallowing.

[0090] Specifically, the predictive terminal can separate the blood oxygen saturation monitoring signal from the dynamic swallowing function data and perform signal analysis and noise reduction. Then, it compares the processed blood oxygen saturation data with a preset threshold to identify safety events in which the blood oxygen saturation drops beyond the threshold. Subsequently, the predictive terminal integrates information such as the frequency of occurrence of safety events and the corresponding test consistency volume combination at the time of occurrence to construct a swallowing safety index.

[0091] Step S302: Endpoint detection is performed on the swallowing audio signal in the dynamic swallowing function data to generate swallowing time parameters.

[0092] Among them, the swallowing time parameter can be used to characterize the duration of a subject's completion of a full swallowing action and the rhythmic characteristics of the swallowing process.

[0093] Specifically, the predictive terminal can separate the swallowing audio signal from the dynamic swallowing function data and perform noise reduction preprocessing. Then, it performs endpoint detection on the processed audio signal to identify the start and end time points of the swallowing action. Subsequently, the predictive terminal combines the power spectral density correlation features of the audio signal to complete the quantization calculation and finally generate the swallowing time parameters.

[0094] Step S303: Based on the laryngeal auscultation results recorded by the clinical assessor during the modified volumetric viscosity test, a residual level is generated.

[0095] The residue level can be a graded assessment based on the results of auscultation of the throat by clinical assessors during the modified volumetric viscosity test, determining whether there is food or liquid residue in the throat after the subject has swallowed and the amount of residue.

[0096] Specifically, the predictive terminal can extract the original results of larynx auscultation recorded by clinical assessors from dynamic swallowing function data. Then, based on the preset residue grading rules, it can classify the pharyngeal residue situation reflected in the auscultation results, determine whether there is food or liquid residue and the range of residue amount, and finally generate a residue level that reflects the state of pharyngeal residue after the subject swallows.

[0097] Step S304: Construct a swallowing efficiency index based on swallowing time parameters and residual level.

[0098] Specifically, the predictive terminal can assign weights and integrate dimensions of swallowing time parameters and residual degree levels according to preset swallowing efficiency assessment rules. Then, the integrated data is quantitatively calibrated and standardized to finally obtain a swallowing efficiency index that characterizes the efficiency of the subject's swallowing action.

[0099] Step S305: Based on swallowing safety indicators and swallowing efficiency indicators, dynamic functional characteristics are obtained.

[0100] Specifically, the predictive terminal can align and unify the swallowing safety indicators and swallowing efficiency indicators according to preset dynamic feature integration rules, eliminating data type differences between different indicators; then, it can assign weights to the integrated indicator data, with the swallowing safety indicators focusing on airway protection risks and the swallowing efficiency indicators focusing on the quality of swallowing action completion; finally, it can fuse the preprocessed swallowing safety indicators and swallowing efficiency indicators to generate dynamic functional features.

[0101] In this embodiment, the prediction terminal extracts and integrates swallowing safety and efficiency-related indicators step by step to generate dynamic functional features that can comprehensively characterize the real-time swallowing function status of the subject, providing dynamic dimensional data support for subsequent aspiration risk prediction.

[0102] In one embodiment, endpoint detection is performed on the swallowing audio signal in the dynamic swallowing function data to generate swallowing time parameters, including:

[0103] Use the following formula to calculate the swallowing time parameter:

[0104]

[0105] in, It is a swallowing time parameter. It is the time when the swallowing action begins. It is the time when the swallowing action ends. It is in time The power spectral density of the swallowing audio signal at the location, It is the integral of the power spectral density of the swallowed audio signal. It is the maximum value of the power spectral density of the swallowed audio signal.

[0106] Specifically, the swallowing action is accompanied by physiological activities such as the contraction of the throat muscles and the closure of the airway. These physiological activities generate audio signals of specific frequencies, and the power spectral density of the swallowing audio signal... It can reflect the energy changes during the swallowing process. To quantify the total energy characteristics throughout the entire swallowing period, integral calculations are required. to Power spectral density within the interval Accumulate.

[0107] In this embodiment, the predictive terminal generates swallowing time parameters of quantified swallowing duration and rhythm by performing endpoint detection on the swallowing audio signal and calculating the ratio of the power spectral density integral to the maximum value. This provides a dynamic quantitative basis for constructing swallowing efficiency indicators and improves the scientific nature of swallowing function assessment.

[0108] In one embodiment, the method may further include:

[0109] Based on the risk level of aspiration, a personalized graded intervention plan is generated; the personalized graded intervention plan includes at least one of the following: health education and guidance plan, food texture modification plan, compensatory swallowing strategy training plan, and systemic rehabilitation treatment plan.

[0110] Among them, the personalized graded intervention program can be a set of differentiated intervention strategies customized based on the subject's aspiration risk level (low risk, medium risk, high risk) and combined with the subject's individual physiological condition, underlying diseases, swallowing function characteristics and other factors. It is a program that connects risk prediction and clinical intervention.

[0111] Specifically, the predictive terminal can determine the aspiration risk level of a subject and simultaneously acquire supporting data such as the subject's individual physiological condition, underlying diseases, and swallowing function characteristics. Then, based on preset tiered intervention rules, it matches corresponding intervention programs for different risk levels: low risk is matched with health education and guidance programs, medium risk is matched with food texture modification and compensatory swallowing strategy training programs, and high risk is matched with systemic rehabilitation treatment programs. Schematic, the preset tiered intervention rules can be standardized decision-making criteria that combine the subject's aspiration risk level, individual physiological condition, and swallowing function characteristics to match differentiated intervention methods for different risk levels. Finally, the program content is customized in detail based on the individual characteristics of the subject to generate a highly adaptable personalized tiered intervention program.

[0112] In this embodiment, the prediction terminal combines the subject's aspiration risk level with individual physiological, disease, and swallowing function characteristics, and matches and customizes intervention plans according to the graded intervention rules, thus achieving an effective connection between risk prediction and clinical intervention, and providing suitable personalized swallowing disorder intervention strategies for subjects at different risk levels.

[0113] In one embodiment, preprocessing the multidimensional assessment dataset to obtain a standardized swallowing assessment dataset may include the following steps:

[0114] Step S401: Based on the indicator types of the multidimensional evaluation dataset, the missing data in the multidimensional evaluation dataset is divided into static indicator missing data and dynamic indicator missing data.

[0115] Among them, missing static indicator data can be missing data in a multi-dimensional assessment dataset where static assessment indicators such as age and underlying diseases have not been collected or entered.

[0116] Missing data for dynamic metrics can be the missing data corresponding to dynamic evaluation metrics in a multi-dimensional evaluation dataset.

[0117] Specifically, the prediction terminal can identify and classify the various indicators in the multi-dimensional assessment dataset. Indicators that reflect the relatively stable state of the subject and do not change dynamically with the swallowing process are classified as static assessment indicators. Indicators that are strongly correlated with real-time swallowing actions and fluctuate dynamically with the swallowing process are classified as dynamic assessment indicators.

[0118] Furthermore, the prediction terminal will filter out the indicator items in the dataset that have missing values, and classify them as static indicator missing data and dynamic indicator missing data respectively.

[0119] Step S402: For the missing static indicator data, the random forest algorithm is used to imputate the data by combining the feature distribution of the same age group and the same underlying disease group, so as to obtain the complete static indicator data.

[0120] Specifically, the prediction terminal can retrieve the missing data of the classified static indicators, and at the same time extract the complete static feature data of subjects of the same age group and the same underlying disease group from the database and analyze the feature distribution pattern; then, based on the distribution pattern, a random forest imputation model is built, and the existing static features of the subjects corresponding to the missing data are used as the model input. The model's feature association learning ability is used to infer the reasonable value of the missing indicator; finally, the inferred value is backfilled into the original dataset to complete the missing data of the static indicators and obtain the missing static indicator data.

[0121] Step S403: Perform one-hot encoding on the classification features in the completed static index data to obtain standardized static data.

[0122] Standardized static data can be structured static evaluation data that can be directly used for subsequent feature extraction and model input, obtained by one-hot encoding of the classification features in the completed static index data and processing them with a unified format and dimensions.

[0123] Specifically, the predictive terminal can filter out classification features such as the Kubota drinking water test level, underlying disease type, and oral function grade from the supplemented static indicator data; then, it performs one-hot encoding on the classification features to transform non-numerical category information into discrete numerical vectors that can be recognized by the model, thereby eliminating the hierarchical bias of the classification features.

[0124] Furthermore, the prediction terminal performs format standardization and dimension unification on the encoded classification features and numerical static features such as EAT-10 total score and age, ultimately generating standardized static data with a standardized structure and data type.

[0125] Step S404: For the missing dynamic index data, based on the consistency-volume combination scenario of the improved volumetric viscosity test, extract the matching degree data of the subjects' age and basic swallowing ability in the same scenario, and perform interpolation through the scenario-based distribution fitting algorithm to obtain the completed dynamic index data.

[0126] Among them, the scenario-based distribution fitting algorithm is a data interpolation algorithm designed for specific consistency-volume combination scenarios in improved volumetric viscosity testing. The core is to first divide the data scenarios according to the test consistency (such as thin liquid, thick liquid, paste, etc.) and volume combination, and then extract the correlation features such as the age and basic swallowing ability matching degree of the subjects in the same scenario. By fitting the distribution pattern of complete dynamic indicator data in the scenario, a special algorithm is used to deduce the reasonable value of missing indicators.

[0127] The completed dynamic indicator data can be a dynamic swallowing function dataset that is free of gaps and has consistent data logic, obtained by interpolating missing values ​​for missing dynamic indicator data through a scenario-based distribution fitting algorithm.

[0128] Specifically, the prediction terminal can retrieve the missing dynamic indicator data that has been classified, and match the missing data to the corresponding test scenario based on the consistency-volume combination of the improved volumetric viscosity test. Then, it extracts the related feature data such as age and swallowing ability matching degree of other subjects in the same scenario, analyzes the distribution pattern of the complete dynamic indicators in the scenario, and builds a scenario-based distribution fitting algorithm model. Then, it inputs the existing related features of the subjects to be imputed into the model to calculate the reasonable values ​​of the missing dynamic indicators. Finally, it fills the original dataset with the calculated values ​​to obtain the completed dynamic indicator data.

[0129] Step S405: Based on the completed dynamic indicator data, standardized dynamic data is obtained.

[0130] Standardized dynamic data can be structured dynamic swallowing assessment data used to characterize and directly extract dynamic functional features.

[0131] Specifically, the predictive terminal can first standardize and classify the time-related features of swallowing actions based on the completed dynamic indicator data and refer to clinical standards. Then, it can verify the food property adaptation-related features according to the scenario rules of the modified volumetric viscosity test and complete the standardization process. Finally, it can integrate the processed features, unify the data format and dimensions, and obtain standardized dynamic data.

[0132] Step S406: Perform dimensionality normalization on the standardized static data and standardized dynamic data to obtain a standardized swallowing assessment dataset.

[0133] Specifically, the predictive terminal can analyze the dimensional distribution and dimensional differences between standardized static data and standardized dynamic data, and select the min-max normalization method as a unified processing method. Then, it performs scaling processing on different dimensional indicators such as numerical features in standardized static data and time-series hierarchical features in standardized dynamic data, and maps the value range of various types of data to the [0,1] interval to eliminate the dimensional and magnitude deviations between different indicators. Finally, the two types of normalized data are integrated and spliced ​​to obtain a standardized swallowing assessment dataset.

[0134] In this embodiment, the prediction terminal performs a series of preprocessing operations on the multi-dimensional assessment dataset, such as missing data classification, targeted imputation, feature encoding, and dimension normalization, to generate a standardized swallowing assessment dataset with uniform format, consistent dimensions, and complete data. This provides a high-quality and highly adaptable data foundation for subsequent feature extraction and aspiration risk prediction, thereby improving the accuracy and standardization of the overall risk prediction process.

[0135] In one embodiment, obtaining standardized dynamic data based on the completed dynamic indicator data may include the following steps:

[0136] Step S501: Perform hierarchical mapping processing on the swallowing action temporal features in the completed dynamic indicator data to obtain temporal hierarchical features.

[0137] Among them, the temporal characteristics of swallowing actions can be quantitative features related to the time dimension of swallowing actions in the completed dynamic indicator data.

[0138] Temporal grading features can be standardized features obtained by performing grading mapping on the temporal features of swallowing actions according to the clinical swallowing function grading standards.

[0139] Specifically, the predictive terminal can filter out the timing characteristics of swallowing actions, such as the duration of swallowing actions, the interval between action stages, and the frequency of swallowing per unit time, from the completed dynamic indicator data; then it can retrieve authoritative clinical swallowing function grading standards and establish a mapping relationship between timing characteristics and clinical grading, such as dividing swallowing duration into normal, mildly prolonged, moderately prolonged, and severely prolonged intervals according to the standards.

[0140] Furthermore, the prediction terminal can perform hierarchical transformation by comparing the specific numerical values ​​of each time series feature with the mapping relationship, thereby eliminating the dimensional differences of the original data; then, the format of the transformed hierarchical results is standardized to generate time series hierarchical features.

[0141] Step S502: Perform rule-based judgment processing on the food trait adaptation features in the completed dynamic index data to obtain the adaptation grading features.

[0142] Among them, food texture adaptation characteristics can be found in the completed dynamic index data, which reflect the swallowing adaptation of subjects to different consistency-volume combinations of food in the modified volume viscosity test.

[0143] Adaptability grading features can be standardized features obtained by performing rule-based judgment processing on the adaptability features of food characteristics according to the scenario-based judgment rules of the modified volumetric viscosity test.

[0144] Specifically, the predictive terminal can filter food property suitability characteristics such as coughing occurrence, blood oxygen saturation fluctuation, and throat residue in different consistency-volume combination food testing scenarios from the supplemented dynamic indicator data. Then, it retrieves the scenario-based judgment rules corresponding to the modified volumetric viscosity test to establish the correspondence between suitability characteristics and clinical suitability levels. For example, the absence of coughing and no significant blood oxygen fluctuation is judged as a high suitability level, while frequent coughing and severe residue are judged as a low suitability level. Subsequently, the specific data of each suitability characteristic are compared with the rules to complete the level judgment, eliminating the differences in the original data of different testing scenarios. Finally, the judgment results are formatted and generated suitability grading characteristics.

[0145] Step S503: Integrate temporal hierarchical features and adaptability hierarchical features to generate standardized dynamic data.

[0146] Specifically, the prediction terminal can perform consistency verification on the dimensional structure and data format of the temporal grading features and the adaptability grading features. Then, it can orderly splice the grading information reflecting swallowing rhythm in the temporal grading features and the grading information reflecting food tolerance in the adaptability grading features to generate an initial dynamic feature matrix. Finally, it can perform data format regularization and redundant information removal on the initial dynamic feature matrix to unify the feature expression form and magnitude range, and generate standardized dynamic data.

[0147] In this embodiment, the prediction terminal generates standardized dynamic data by hierarchically mapping the time-series features of dynamic indicator data, determining food compatibility feature rules, and integrating features, thus achieving standardized processing of dynamic indicators.

[0148] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0149] Based on the same inventive concept, this application also provides a device for predicting the risk of elderly patients with dysphagia as described above. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the risk prediction device for elderly patients with dysphagia provided below can be found in the limitations of the risk prediction method for elderly patients with dysphagia described above, and will not be repeated here.

[0150] In one exemplary embodiment, such as Figure 2 As shown, a risk prediction device 600 for elderly patients with dysphagia is provided, comprising:

[0151] The data acquisition module 601 is used to acquire the multidimensional assessment dataset of the subjects. The multidimensional assessment dataset includes static risk screening data based on the dietary assessment scale EAT-10, dynamic swallowing function data based on the modified volumetric viscosity test, swallowing performance data based on the Kubota water swallowing test, and basic characteristic data of the subjects.

[0152] Preprocessing module 602 is used to preprocess the multidimensional assessment dataset to obtain a standardized swallowing assessment dataset;

[0153] The feature extraction module 603 is used to extract static risk features and dynamic functional features from the standardized swallowing assessment dataset, and to fuse the static risk features and dynamic functional features to obtain a fused feature vector;

[0154] The grade prediction module 604 is used to input the fused feature vector into the preset aspiration risk prediction model to obtain the aspiration risk level of the subject; the aspiration risk level includes low risk, medium risk and high risk.

[0155] In one embodiment, the feature extraction module extracts static risk features and dynamic functional features from a standardized swallowing assessment dataset, including:

[0156] The score extraction unit is used to extract the EAT-10 total score from the static risk screening data and determine the EAT-10 total score as the first static risk feature.

[0157] The Kubota rating unit is used to rate the swallowing performance data, obtain the Kubota drinking test rating, and determine the Kubota drinking test rating as the second static risk characteristic.

[0158] The static risk feature integration unit is used to obtain static risk features based on the first static risk feature and the second static risk feature;

[0159] The dynamic functional feature generation unit is used to obtain dynamic functional features based on dynamic swallowing function data. The dynamic functional features include swallowing safety indicators and swallowing efficiency indicators. The swallowing safety indicators include at least one of the following: whether the absolute value of blood oxygen saturation is lower than a preset threshold and whether choking occurs under a specific consistency and volume test combination. The swallowing efficiency indicators include the swallowing time parameter required to swallow a specific volume of liquid once and the degree of residue determined by auscultation of the larynx and neck after swallowing.

[0160] In one embodiment, the dynamic functional feature generation unit obtains dynamic functional features based on dynamic swallowing function data, including:

[0161] The indicator construction subunit is used to analyze the blood oxygen saturation monitoring signal in the dynamic swallowing function data, identify safety events where blood oxygen saturation drops beyond a preset threshold, and construct swallowing safety indicators based on these safety events.

[0162] The parameter generation subunit is used to perform endpoint detection on the swallowing audio signal in the dynamic swallowing function data and generate swallowing time parameters.

[0163] The grading subunit is used to generate a residual level based on the laryngeal auscultation results recorded by clinical assessors during the modified volumetric viscosity test;

[0164] The efficiency index construction sub-unit is used to construct swallowing efficiency indexes based on swallowing time parameters and residual level.

[0165] The functional feature integration subunit is used to obtain dynamic functional features based on swallowing safety indicators and swallowing efficiency indicators.

[0166] In one embodiment, the parameter generation subunit performs endpoint detection on the swallowing audio signal in the dynamic swallowing function data to generate swallowing time parameters, including:

[0167] Use the following formula to calculate the swallowing time parameter:

[0168]

[0169] in, It is a swallowing time parameter. It is the time when the swallowing action begins. It is the time when the swallowing action ends. It is in time The power spectral density of the swallowing audio signal at the location, It is the integral of the power spectral density of the swallowed audio signal. It is the maximum value of the power spectral density of the swallowed audio signal.

[0170] In one embodiment, the device further includes:

[0171] The intervention program module is used to generate personalized, graded intervention programs based on the risk level of aspiration. The personalized, graded intervention programs include at least one of the following: health education and guidance program, food texture modification program, compensatory swallowing strategy training program, and systemic rehabilitation treatment program.

[0172] In one embodiment, the preprocessing module preprocesses the multidimensional assessment dataset to obtain a standardized swallowing assessment dataset, including:

[0173] The missing data classification unit is used to classify missing data in a multidimensional evaluation dataset into static indicator missing data and dynamic indicator missing data based on the indicator type of the multidimensional evaluation dataset.

[0174] The static indicator missing imputation unit is used to imput missing static indicator data by using the random forest algorithm combined with the feature distribution of the same age group and the same underlying disease group to obtain the complete static indicator data.

[0175] The static feature encoding unit is used to perform one-hot encoding on the classification features in the completed static index data to obtain standardized static data.

[0176] The dynamic indicator missing imputation unit is used to extract the age and swallowing basic ability matching degree data of the subjects in the same scenario based on the consistency-volume combination scenario of the improved volume viscosity test for the missing dynamic indicator data. The imputation is performed by the scenario-based distribution fitting algorithm to obtain the completed dynamic indicator data.

[0177] The dynamic data standardization unit is used to obtain standardized dynamic data based on the completed dynamic indicator data;

[0178] The dimensionality normalization unit is used to perform dimensionality normalization on standardized static data and standardized dynamic data to obtain a standardized swallowing assessment dataset.

[0179] In one embodiment, the dynamic data standardization unit obtains standardized dynamic data based on the completed dynamic indicator data, including:

[0180] The hierarchical mapping subunit is used to perform hierarchical mapping on the swallowing action temporal features in the completed dynamic index data to obtain temporal hierarchical features.

[0181] The rule determination subunit is used to perform rule determination processing on the food trait adaptation features in the completed dynamic index data to obtain the adaptation grading features.

[0182] The feature integration subunit is used to integrate temporal hierarchical features and adaptive hierarchical features to generate standardized dynamic data.

[0183] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of a risk prediction method for elderly patients with dysphagia as described above.

[0184] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0185] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0186] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A method for predicting the risk of dysphagia in elderly patients, characterized in that, The method includes: Obtain a multidimensional assessment dataset of the subjects; the multidimensional assessment dataset includes static risk screening data based on the dietary assessment scale EAT-10, dynamic swallowing function data based on the modified volumetric viscosity test, swallowing performance data based on the Kubota water swallowing test, and basic characteristic data of the subjects. The multidimensional assessment dataset is preprocessed to obtain a standardized swallowing assessment dataset; Static risk features and dynamic functional features are extracted from the standardized swallowing assessment dataset, and the static risk features and dynamic functional features are fused to obtain a fused feature vector; The fused feature vector is input into a preset aspiration risk prediction model to obtain the aspiration risk level of the subject; the aspiration risk level includes low risk, medium risk and high risk.

2. The method according to claim 1, characterized in that, The extraction of static risk features and dynamic functional features from the standardized swallowing assessment dataset includes: The EAT-10 total score is extracted from the static risk screening data, and the EAT-10 total score is determined as the first static risk feature; The swallowing performance data is graded to obtain the Kubota drinking water test grade, and the Kubota drinking water test grade is determined as the second static risk characteristic; The static risk characteristics are obtained based on the first static risk characteristics and the second static risk characteristics; Based on the dynamic swallowing function data, the dynamic functional characteristics are obtained; the dynamic functional characteristics include swallowing safety indicators and swallowing efficiency indicators; the swallowing safety indicators include at least one of the following under a specific consistency and volume test combination: whether the absolute value of blood oxygen saturation is lower than a preset threshold and whether choking occurs; the swallowing efficiency indicators include the swallowing time parameter required to swallow a specific volume of liquid once and the degree of residue determined by auscultation of the larynx and neck after swallowing.

3. The method according to claim 2, characterized in that, The dynamic functional characteristics obtained based on the dynamic swallowing function data include: The blood oxygen saturation monitoring signal in the dynamic swallowing function data is analyzed to identify safety events where the blood oxygen saturation drops beyond a preset threshold, and the swallowing safety index is constructed based on the safety events. Endpoint detection is performed on the swallowing audio signal in the dynamic swallowing function data to generate the swallowing time parameter; The residual level is generated based on the laryngeal auscultation results recorded by the clinical assessor during the modified volumetric viscosity test; Based on the swallowing time parameter and the residue level, the swallowing efficiency index is constructed; The dynamic functional characteristics are obtained based on the swallowing safety index and the swallowing efficiency index.

4. The method according to claim 3, characterized in that, The step of performing endpoint detection on the swallowing audio signal in the dynamic swallowing function data to generate swallowing time parameters includes: The swallowing time parameter is calculated using the following formula: in, It is a swallowing time parameter. It is the time when the swallowing action begins. It is the time when the swallowing action ends. It is in time The power spectral density of the swallowing audio signal at the location, It is the integral of the power spectral density of the swallowed audio signal. It is the maximum value of the power spectral density of the swallowed audio signal.

5. The method according to claim 1, characterized in that, The method further includes: Based on the aspiration risk level, a personalized graded intervention plan is generated; the personalized graded intervention plan includes at least one of the following: a health education and guidance plan, a food texture modification plan, a compensatory swallowing strategy training plan, and a systemic rehabilitation treatment plan.

6. The method according to any one of claims 1 to 5, characterized in that, The preprocessing of the multi-dimensional assessment dataset to obtain a standardized swallowing assessment dataset includes: Based on the indicator types of the multidimensional evaluation dataset, the missing data in the multidimensional evaluation dataset is divided into static indicator missing data and dynamic indicator missing data. For the missing static indicator data, the random forest algorithm is used in combination with the feature distribution of the same age group and the same underlying disease group to imputate the missing static indicator data. The classification features in the completed static index data are subjected to one-hot encoding to obtain standardized static data; For the missing dynamic index data, based on the consistency-volume combination scenario of the improved volumetric viscosity test, the age and swallowing basic ability matching data of the subjects in the same scenario are extracted, and the scenario-based distribution fitting algorithm is used for interpolation to obtain the completed dynamic index data. Based on the completed dynamic indicator data, standardized dynamic data is obtained; The standardized static data and the standardized dynamic data are subjected to dimensionality normalization to obtain the standardized swallowing assessment dataset.

7. The method according to claim 6, characterized in that, The standardized dynamic data obtained based on the completed dynamic indicator data includes: The swallowing action temporal features in the completed dynamic index data are subjected to hierarchical mapping processing to obtain temporal hierarchical features; The food trait adaptation features in the completed dynamic index data are subjected to rule-based judgment processing to obtain adaptation grading features. The standardized dynamic data is generated by integrating the temporal classification features and the adaptability classification features.

8. A risk prediction device for elderly patients with dysphagia, characterized in that, The device includes: The data acquisition module is used to acquire the multidimensional assessment dataset of the subjects; the multidimensional assessment dataset includes static risk screening data based on the dietary assessment scale EAT-10, dynamic swallowing function data based on the modified volumetric viscosity test, swallowing performance data based on the Kubota water swallowing test, and the subjects' basic characteristic data. The preprocessing module is used to preprocess the multi-dimensional assessment dataset to obtain a standardized swallowing assessment dataset; The feature extraction module is used to extract static risk features and dynamic functional features from the standardized swallowing assessment dataset, and to fuse the static risk features and the dynamic functional features to obtain a fused feature vector; The grade prediction module is used to input the fused feature vector into a preset aspiration risk prediction model to obtain the aspiration risk level of the subject; the aspiration risk level includes low risk, medium risk and high risk.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images