Alzheimer's disease early warning analysis method and system based on multiple Chinese syllables

Abstract

The embodiment of the invention provides an Alzheimer's disease early warning analysis method and system based on multiple Chinese syllables. The method is applied to the technical field of data processing, and comprises the following steps: collecting voice data of an Alzheimer's patient, performing data perception, and constructing a multi-syllable sample; performing data preprocessing by adopting a multi-level graph embedded sparse feature learning method to obtain multi-syllable voice output features; constructing a classifier, carrying out feature optimization processing on the output features, and outputting to obtain a cost sensitive matrix; a cost-sensitive weighted objective function is constructed on the basis of the cost-sensitive matrix, a multi-syllable fusion analysis diagnosis result is obtained through calculation, and the Alzheimer's disease probability is predicted according to the diagnosis result; and guiding and suggesting the patient according to a probability prediction result. In this way, simple Chinese syllables are adopted for patient data collection, the Alzheimer's disease early warning scheme is obtained in combination with a multi-syllable fusion analysis method, and the accuracy of early warning analysis is improved.

Description

Technical Field

[0001] The present disclosure relates to the field of data processing technology, and in particular to an Alzheimer's disease early warning analysis method and system based on multi-Chinese syllables. Background Art

[0002] Alzheimer's Disease (AD) is a gradually deteriorating neurological disease characterized by progressive cognitive dysfunction. It causes the death of neurons and the breakdown of connections in the brain, which affects memory, thinking, language, and behavior. Patients may lose their memory, get lost, fail to recognize their loved ones, and experience decreased language skills, which ultimately lead to severe cognitive impairment and the loss of the ability to live independently.

[0003] Currently, there are many methods for evaluating and diagnosing Alzheimer's disease, and their comprehensive application can accurately diagnose the patient's cognitive impairment and provide a strong basis for subsequent treatment. However, the above methods are affected by factors such as the subject's cultural level and emotional state, and are costly, complex to operate, and not suitable as routine screening methods. In addition, some methods are somewhat invasive and cannot adapt to today's Alzheimer's disease diagnostic practice.

[0004] Early warning of Alzheimer's disease based on speech analysis has the advantages of non-invasiveness, convenience, low cost, and objectivity. Therefore, it is urgently needed to provide a method for diagnosing and predicting Alzheimer's disease based on speech data and a corresponding system. Summary of the invention

[0005] The present invention provides an Alzheimer's disease early warning analysis method and system based on multiple Chinese syllables. By adopting concise Chinese syllables to collect patient data and combining it with a multi-syllable fusion analysis method, the technical problems that the prior art cannot effectively and accurately diagnose patient cognition and has poor prediction results due to multiple factors affecting patient voice data are solved.

[0006] According to a first aspect of the present disclosure, a method for early warning analysis of Alzheimer's disease based on multiple Chinese syllables is provided, comprising the following steps: Collect speech data of Alzheimer's patients and perform data perception to construct multi-syllable samples; A multi-level graph embedding sparse feature learning method is used to perform data preprocessing on the polysyllabic sample to construct a polysyllabic speech output feature; Constructing a classifier, using a cost-sensitive random configuration network in the classifier to perform feature optimization processing on the polysyllabic speech output features, and outputting a cost-sensitive matrix; A cost-sensitive weighted objective function is constructed based on the cost-sensitive matrix, and the polysyllabic fusion analysis diagnosis results of different classifiers are calculated respectively, and the probability of Alzheimer's disease is predicted according to the diagnosis results; Provide guidance and advice to the patient based on the Alzheimer's disease probability prediction results.

[0007] According to the above aspects and any possible implementation, a further implementation is provided, wherein the process of collecting speech data of Alzheimer's patients and performing data perception to construct a polysyllabic sample is as follows: A recording sample is prepared, and Chinese characters, Chinese syllables, international phonetic symbols and pronunciation prompts are annotated on the recording sample according to Chinese syllables. Alzheimer's patient voice data is collected according to the annotated results to obtain a polysyllabic sample, wherein the polysyllabic sample includes syllable pronunciation data.

[0008] According to the above aspects and any possible implementation, an implementation is further provided, wherein the process of performing data preprocessing on the polysyllabic sample by using the multi-level graph embedding sparse feature learning method to construct the polysyllabic speech output feature is as follows: Extracting features from the syllable pronunciation data to obtain Mel-Cep coefficients, time base errors, and amplitude perturbation speech features, and performing normalization processing to obtain processed speech features; A graph random sparse autoencoder based on random configuration is constructed, and the graph random sparse autoencoder is used to extract the processed speech features to obtain polysyllabic speech output features.

[0009] According to the above aspects and any possible implementation, an implementation is further provided, wherein the process of constructing a graph random sparse autoencoder, extracting the processed speech features using the graph random sparse autoencoder, and obtaining the polysyllabic speech output features is as follows: Acquire processed speech features, use the graph random sparse autoencoder to extract the processed speech features, and obtain the output matrix of the hidden layer nodes of the graph random sparse autoencoder; Define a graph random sparse autoencoder weight objective function, and use the hidden layer node output matrix as the input of the weight objective function, and output the graph random sparse autoencoder weight value; The multi-syllabic speech output feature is constructed by combining the hidden layer node output matrix and the weight value.

[0010] According to the above aspects and any possible implementation, an implementation is further provided, wherein the classifier is constructed, and a cost-sensitive random configuration network in the classifier is used to perform feature optimization processing on the polysyllabic speech output features, and the process of outputting a cost-sensitive matrix is ​​as follows: Initializing a cost-sensitive matrix based on the polysyllabic speech output features; A classifier is constructed by using a cost-sensitive random configuration network based on an objective function, and the classifier and an intelligent optimization method are used to update and output the initialized cost-sensitive matrix to obtain a maximized objective function; The cost-sensitive matrix corresponding to the maximized objective function is taken as an output result.

[0011] According to the above aspects and any possible implementation, an implementation is further provided, wherein the maximized objective function is specifically: ; in, represents the objective function of classifier i, is the hidden layer output of classifier i, Represents the classification label, represents the classifier regularization parameter, is the cost-sensitive matrix, is the number of classifiers, is the number of cost-sensitive matrices.

[0012] According to the above aspects and any possible implementation, an implementation is further provided, wherein the process of constructing a cost-sensitive weighted objective function based on the cost-sensitive matrix, respectively calculating the multi-syllable fusion analysis diagnosis results of different classifiers, and predicting the probability of Alzheimer's disease according to the diagnosis results is as follows: Constructing a weighted output matrix according to the cost-sensitive matrix and the hidden layer node output matrix; constructing a cost-sensitive weighted objective function according to the weighted output matrix and the maximization objective function, and calculating a fusion output matrix according to the cost-sensitive weighted objective function; Calculating a diagnosis result based on the weighted output matrix and the fused output matrix; The diagnostic results are analyzed using a voting method to obtain a fusion analysis diagnostic result.

[0013] According to the above aspects and any possible implementation, an implementation is further provided, wherein the objective function based on cost-sensitive weighting is specifically: ; in, represents the original output weight of classifier k, is the weighted output matrix, specifically , is the weighted output matrix of classifier k, is the hidden layer output of classifier k, represents the regularization parameter of the fusion objective function, K is the number of classifiers, is the cost-sensitive matrix, and B is the fusion output matrix, specifically: , is the value of the cost-sensitive weighted objective function.

[0014] According to the above aspects and any possible implementation manner, an implementation manner is further provided, wherein the process of analyzing the diagnosis result by using the voting method to obtain the fusion analysis diagnosis result is: Calculating the probability values ​​of the diagnosis results respectively, wherein the probability values ​​include two categories: Alzheimer's disease probability value and non-Alzheimer's disease probability value; Determine the probability values ​​of Alzheimer's disease and non-Alzheimer's disease in the probability values ​​of each diagnosis result, and vote for the classification with a large probability value; Count the votes and take the category with the most votes as the output value of the fusion analysis diagnosis result.

[0015] According to a second aspect of the present disclosure, there is provided an Alzheimer's disease early warning analysis system based on multiple Chinese syllables, which is used to implement the Alzheimer's disease early warning analysis method based on multiple Chinese syllables as described in the first aspect of the embodiment, the system comprising: a data perception module, a data preprocessing module, a classifier module, a polysyllabic fusion diagnosis module, and an early warning screening and guidance suggestion module; The data perception module is used to collect the speech data of Alzheimer's patients and perform data perception to construct a polysyllabic sample; The data preprocessing module is used to perform data preprocessing on the polysyllabic sample by adopting a multi-level graph embedding sparse feature learning method to construct a polysyllabic speech output feature; The classifier module is used to construct a classifier, and uses a cost-sensitive random configuration network in the classifier to perform feature optimization processing on the polysyllabic speech output features, and outputs a cost-sensitive matrix; The polysyllabic fusion diagnosis module is used to construct a cost-sensitive weighted objective function based on a cost-sensitive matrix, calculate the polysyllabic fusion analysis diagnosis results of different classifiers respectively, and predict the probability of Alzheimer's disease according to the diagnosis results; The early warning screening and guidance suggestion module is used to provide guidance and suggestions to patients based on the diagnosis results of the polysyllabic fusion diagnosis module.

[0016] Compared with the prior art, the present invention has the following technical effects: (1) Given that Chinese syllables consist of three parts, namely initial consonants, final consonants and tones, and have a complex structure and multiple combinations, this provides a rich and concise source of information for the diagnosis of Alzheimer's disease based on speech analysis. The present invention analyzes the accuracy and fluency of the initial consonants, final consonants and tones of patients during their pronunciation, and captures such subtle language changes to more carefully assess their cognitive function status. At the same time, in order to reduce the influence of multiple factors such as age, gender, cultural background, and dialect, the system uses Chinese characters, syllables, international phonetic symbols, and pronunciation guidance to ensure the accuracy of the collected data; (2) Due to the influence of pronunciation differences of a single syllable, it is impossible to guarantee a high warning accuracy. Therefore, in order to improve the reliability of early warning diagnosis of Alzheimer's disease, the present invention is based on an evolutionary cost-sensitive random configuration network as a base classifier and a cost-sensitive weighted negative correlation learning model as a multi-syllable fusion model. The evolutionary cost-sensitive base classifier weights the sample weights, which is conducive to overcoming the influence of sample data noise and uneven distribution of sample categories; the cost-sensitive weighted negative correlation learner can fuse the classification results of multiple base classifiers; (3) The present invention can collect data through different terminals such as personal computers, home televisions, mobile phones, tablets, etc., use cloud servers to analyze and process the data, and provide early warning monitoring results in real time, achieving significant improvements and breakthroughs in implementation methods, perception quality, friendly interaction, and health management.

[0017] It should be understood that the contents described in the summary of the invention are not intended to limit the key or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure. In the accompanying drawings, the same or similar reference numerals represent the same or similar elements, among which: Figure 1 A schematic diagram of a process flow of an Alzheimer's disease early warning analysis method based on multiple Chinese syllables according to an embodiment of the present disclosure is shown; Figure 2 A schematic diagram of a data perception process of an Alzheimer's disease early warning analysis method based on multiple Chinese syllables according to an embodiment of the present disclosure is shown; Figure 3 A schematic diagram of a data preprocessing process of an Alzheimer's disease early warning analysis method based on multiple Chinese syllables according to an embodiment of the present disclosure is shown; Figure 4A schematic diagram of a classifier processing process of an Alzheimer's disease early warning analysis method based on multiple Chinese syllables according to an embodiment of the present disclosure is shown; Figure 5 A schematic diagram of a multi-syllable fusion diagnosis process of an Alzheimer's disease early warning analysis method based on multiple Chinese syllables according to an embodiment of the present disclosure is shown; Figure 6 A schematic structural diagram of an Alzheimer's disease early warning analysis system based on multiple Chinese syllables according to an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the embodiments of the present disclosure clearer, the technical solution in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present disclosure.

[0020] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0021] Alzheimer's Disease (AD) is a neurodegenerative disease with a slow progression and worsening over time. According to statistics, the 2024 Alzheimer's Disease Data Report in my country shows that there are nearly 17 million Alzheimer's patients in my country, making it the country with the largest number of Alzheimer's patients in the world. The disease causes gradual damage to the patient's neurons and their neural connections, and eventually dies from the disease or its complications. The early stage of AD is mild cognitive impairment (MCI), in which patients have normal daily living abilities, but there is a progressive decline in cognitive function. From a treatment perspective, AD is irreversible and there are great difficulties in treatment, but if patients can be treated at the MCI stage, the onset of dementia can be effectively delayed.

[0022] Currently, the diagnosis of Alzheimer's disease mainly includes the following aspects: 1) Cognitive assessment: Mini-Mental State Examination, Montreal Cognitive Assessment, Wechsler Adult Intelligence Scale, etc., can assess the patient's cognitive function, including attention, memory, language ability, executive function and other areas.

[0023] 2) Imaging examination: Through medical examinations such as cranial CT scan and MRI, check for intracranial lesions, movement disorders, etc., especially atrophy of the hippocampus and medial temporal lobe, which is more common in Alzheimer's disease.

[0024] 3) Laboratory tests: also include thyroid function tests, blood tests, etc. to rule out other possible causes of cognitive impairment.

[0025] 4) Cerebrospinal fluid examination: Detection of certain specific biomarkers in the cerebrospinal fluid, such as amyloid β-42 (Aβ42), tau protein, etc. The abnormal levels of these biomarkers are helpful in diagnosing Alzheimer's disease.

[0026] 5) Genetic testing: For patients with a family history or early-onset Alzheimer's disease, genetic testing may help identify the cause, such as testing the APOE gene.

[0027] 6) Positron emission tomography: Uses specific tracers, such as fluorodeoxyglucose or amyloid tracers, to observe the metabolic activity and amyloid deposition in the brain, which is also an important means of diagnosing Alzheimer's disease.

[0028] The combined use of these methods can more accurately diagnose the patient's cognitive impairment and provide a strong basis for subsequent treatment. However, the above methods: 1) are affected by factors such as the subject's cultural level and emotional state; 2) to 6) are high in cost and complex in operation, and are not suitable as routine screening methods; 3) some of the methods are traumatic.

[0029] At present, with the development of artificial intelligence technology, early warning of Alzheimer's disease based on speech analysis has been widely studied at home and abroad. In order to reduce the influence of various factors such as age, gender, cultural background, dialect, etc. on speech diagnosis, the present invention uses concise Chinese syllables to collect patient data, and combines the multi-syllable fusion analysis method to provide a new early warning scheme for Alzheimer's disease for Chinese people whose mother tongue is Chinese, which is conducive to improving the accuracy of early warning analysis of Alzheimer's disease.

[0030] Reference Figure 1 As shown, this embodiment provides an Alzheimer's disease early warning analysis method based on multiple Chinese syllables, comprising the following steps: S101. Collect speech data of Alzheimer's patients and perform data perception to construct a polysyllabic sample.

[0031] like Figure 2As shown, in this embodiment, by providing the Chinese characters, Chinese syllables, international phonetic symbols and pronunciation prompts of the corresponding syllables before recording the Chinese syllable samples on any medium with an operating system, a microphone and a recording medium such as a mobile phone, a tablet, a computer, a television, etc., the accuracy of the recording is ensured.

[0032] Specifically, the cognitive impairment caused by AD affects the ability to express language, which is then reflected in the process and content of language expression. Therefore, by collecting the voice of the user performing a preset descriptive task (such as collecting voice through the microphone of an electronic device) to obtain voice information, the algorithm based on voice analysis recognizes the user's language expression ability and then evaluates the degree of cognitive impairment of the user, thereby detecting AD and MCI based on voice information.

[0033] Among them, the preset description task may include describing the content of the preset image and describing the names of different target objects of the target type within a preset time. It can be understood that impaired cognitive function will affect the accuracy of the description of the preset image. For the content in the image, the more severe the cognitive function is impaired, the worse the accuracy of the description is. When performing the description task, the display screen of the electronic device can display the preset image. The accuracy of the description of the preset image by a normal user and a user with impaired cognitive function is obviously different; similarly, impaired cognitive function will also affect the fluency of the description. For example, the names of different target objects of the target type described within a preset time can be to say as many animal names as possible within a preset time (such as 30 seconds, 1 minute, 2 minutes, etc.). Therefore, by collecting the voice information of the user performing the preset description task, it can be used to detect the degree of cognitive impairment, thereby detecting the probability of the user suffering from AD and MCI.

[0034] S102. Use a multi-level graph embedding sparse feature learning method to perform data preprocessing on the polysyllabic sample to construct a polysyllabic speech output feature.

[0035] like Figure 3 As shown, in this embodiment, for audio data of different syllables, speech features such as Mel-frequency cepstral coefficients (MFCC), time base error (Jitter), amplitude perturbation (Shimmer), etc. are extracted, and feature normalization is performed to obtain normalized preprocessing features, which are then input into multiple graph random sparse autoencoders for sparse feature learning and generate combined features. ,in, , Voice features.

[0036] Specifically, taking single-shot feature learning as an example, the learning process of the i-th graph random sparse autoencoder is as follows: The graph random sparse autoencoder consists of Lmax nodes, assuming Nodes have been generated, and the hidden layer nodes are generated using the random configuration incremental learning method. Parameters, hidden layer nodes The output matrix is ​​as follows: (1) in, To extract the speech features of each training sample, N represents the number of samples; represents the output of node L; and Represent the weight and bias of node L respectively; represents the activation function and T is the transposed matrix.

[0037] The weight objective function of the i-th graph random sparse autoencoder is defined as shown in formula (2): represents the weight of the i-th graph random sparse autoencoder, then the output of the i-th graph random sparse autoencoder is , specifically: (2) (3) in, ; Tr represents the trace or number of traces of the matrix; and represents the reconstructed sparse features of samples i and j, express and The similarity of and Represent different pairs of data samples respectively. If and If they are neighbors, we get the weight coefficient , represents the Gaussian kernel parameter, otherwise ; represents the graph Laplacian matrix, Depend on The composition of each element, represents a diagonal matrix; , represents the sparse weight regularization parameter; represents the graph regularization parameter.

[0038] S103, constructing a classifier, using the cost-sensitive random configuration network in the classifier to perform feature optimization processing on the polysyllabic speech output features, and outputting a cost-sensitive matrix.

[0039] like Figure 4As shown, in this embodiment, in order to reduce the impact of medical data sample noise and category imbalance on the classification results, taking classifier i as an example, the graph of input pronunciation i is embedded with sparse features, that is, the multi-syllable speech output features, and the cost-sensitive random configuration network based on the objective function of formula (4) is used as the classifier: (4) in, represents the objective function of classifier i; is the hidden layer output of classifier i; Indicates the classification label; represents the classifier regularization parameter, is the number of classifiers.

[0040] Introducing the cost-sensitive matrix , where i=1,2,3,…,k, The classifier is weighted, and the cost-sensitive matrix is ​​updated by iterative optimization combined with the intelligent optimization algorithm until the maximum number of iterations of the intelligent optimization algorithm is met. The cost-sensitive matrix that can meet the maximum diagnostic accuracy is selected as the final matrix to improve the Alzheimer's disease warning accuracy of a single pronunciation.

[0041] S104. A cost-sensitive weighted objective function is constructed based on the cost-sensitive matrix, and the multi-syllable fusion analysis diagnosis results of different classifiers are calculated respectively, and the probability of Alzheimer's disease is warned according to the diagnosis results.

[0042] like Figure 5 As shown, in order to achieve multi-syllable fusion diagnosis, the evolutionary cost sensitivity matrix of each syllable classifier is And its hidden layer output matrix , construct the weighted output matrix ,in , taking formula (5) as the objective function, solve the fusion output matrix , specifically: (5) in, represents the original output weight of classifier k; ; represents the regularization parameter of the fusion objective function, is the cost-sensitive weighted objective function value, is the weighted output matrix of classifier k.

[0043] Finally, the diagnostic results are calculated respectively , using voting method to obtain fusion diagnosis results .

[0044] In this embodiment, a soft voting method is used to determine the fusion diagnosis result. Soft voting is an integrated strategy in multi-model voting, which is particularly suitable for classification problems. It is a probability-based voting mechanism, in which the prediction result of each model is regarded as an estimate of a certain probability of the category, and the final prediction result is determined based on the weighted average of the prediction probabilities of all models. The final voting result can often utilize the complementary functions between single-classification models to reduce the error of a single classifier and improve the prediction performance and classification accuracy.

[0045] Specifically, in this embodiment, the diagnosis results are calculated respectively The probability value includes two categories: Alzheimer's disease probability value and non-Alzheimer's disease probability value; Determine the probability values ​​of Alzheimer's disease and non-Alzheimer's disease in the probability values ​​of each diagnosis result, and vote for the classification with a large probability value; Calculate the number of votes and take the classification with the largest number of votes as the output fusion diagnosis output result P.

[0046] like Figure 6 As shown, this embodiment also provides an Alzheimer's disease early warning analysis system based on multiple Chinese syllables, which is used to implement the Alzheimer's disease early warning analysis method based on multiple Chinese syllables described in the above embodiment, and the system includes: a data perception module 1, a data preprocessing module 2, a classifier module 3, a multi-syllable fusion diagnosis module 4 and an early warning screening and guidance suggestion module 5; The data perception module 1 is used to collect the speech data of Alzheimer's patients and perform data perception to construct a polysyllabic sample; The data preprocessing module 2 is used to perform data preprocessing on the polysyllabic sample by adopting a multi-level graph embedding sparse feature learning method to construct a polysyllabic speech output feature; The classifier module 3 is used to construct a classifier, and uses the cost-sensitive random configuration network in the classifier to perform feature optimization processing on the multi-syllable speech output features, and outputs a cost-sensitive matrix; The polysyllabic fusion diagnosis module 4 is used to construct a cost-sensitive weighted objective function based on the cost-sensitive matrix, calculate the polysyllabic fusion analysis diagnosis results of different classifiers respectively, and warn the probability of Alzheimer's disease according to the diagnosis results; The early warning screening and guidance suggestion module 5 is used to provide guidance and suggestions to patients based on the diagnosis results of the polysyllabic fusion diagnosis module 4.

[0047] In this embodiment, based on the patient data of a certain tertiary-level A hospital, 124 healthy elderly samples and 75 patients with mild cognitive impairment were collected. The pronunciation combinations of the above samples were collected separately. ,one ,Room The pronunciation data of the invention can be used to assist in the diagnosis of mild cognitive impairment in the early stages of Alzheimer's disease with an accuracy rate of over 70%.

[0048] During the implementation of the case, users use mobile phones and other platforms to input syllables according to the guidance. After the input syllables are preprocessed, the classifier module and the multi-syllable fusion diagnosis module can obtain the probability of Alzheimer's disease under single syllables and the probability of Alzheimer's disease under polysyllables, and can provide guidance such as precautions based on the probabilities.

[0049] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should be aware that the present disclosure is not limited by the order of the actions described, because according to the present disclosure, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required by the present disclosure.

[0050] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of this disclosure can be achieved, and this document is not limited here.

[0051] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A method for early warning analysis of Alzheimer's disease based on multi-Chinese syllables, characterized in that: The following steps are involved: Collect speech data of Alzheimer's patients and perform data perception to construct multi-syllable samples; A multi-level graph embedding sparse feature learning method is used to perform data preprocessing on the polysyllabic sample to construct a polysyllabic speech output feature; Constructing a classifier, using a cost-sensitive random configuration network in the classifier to perform feature optimization processing on the polysyllabic speech output features, and outputting a cost-sensitive matrix; A cost-sensitive weighted objective function is constructed based on the cost-sensitive matrix, and the polysyllabic fusion analysis diagnosis results of different classifiers are calculated respectively, and the probability of Alzheimer's disease is predicted according to the diagnosis results; Provide guidance and advice to the patient based on the Alzheimer's disease probability prediction results.

2. The method for early warning analysis of Alzheimer's disease based on multi-Chinese syllables according to claim 1 is characterized in that: The process of collecting the speech data of Alzheimer's patients and performing data perception to construct a polysyllabic sample is as follows: A recording sample is prepared, and Chinese characters, Chinese syllables, international phonetic symbols and pronunciation prompts are annotated on the recording sample according to Chinese syllables. Alzheimer's patient voice data is collected according to the annotated results to obtain a polysyllabic sample, wherein the polysyllabic sample includes syllable pronunciation data.

3. The method for early warning analysis of Alzheimer's disease based on multi-Chinese syllables according to claim 2 is characterized in that: The process of using the multi-level graph embedding sparse feature learning method to perform data preprocessing on the polysyllabic sample and constructing the polysyllabic speech output feature is as follows: Extracting features from the syllable pronunciation data to obtain Mel-Cep coefficients, time base errors, and amplitude perturbation speech features, and performing normalization processing to obtain processed speech features; A graph random sparse autoencoder based on random configuration is constructed, and the graph random sparse autoencoder is used to extract the processed speech features to obtain polysyllabic speech output features.

4. The method for early warning analysis of Alzheimer's disease based on multi-Chinese syllables according to claim 3 is characterized in that: The process of constructing a graph random sparse autoencoder, extracting the processed speech features using the graph random sparse autoencoder, and obtaining the polysyllabic speech output features is as follows: Acquire processed speech features, use the graph random sparse autoencoder to extract the processed speech features, and obtain a hidden layer node output matrix of the graph random sparse autoencoder; Define a graph random sparse autoencoder weight objective function, and use the hidden layer node output matrix as the input of the weight objective function, and output the graph random sparse autoencoder weight value; The multi-syllabic speech output feature is constructed by combining the hidden layer node output matrix and the weight value.

5. The Alzheimer's disease early warning analysis method based on multi-Chinese syllables according to claim 1 is characterized in that: The process of constructing a classifier, using a cost-sensitive random configuration network in the classifier to perform feature optimization processing on the polysyllabic speech output features, and outputting a cost-sensitive matrix is ​​as follows: Initializing a cost-sensitive matrix based on the polysyllabic speech output features; A classifier is constructed by using a cost-sensitive random configuration network based on an objective function, and the classifier and an intelligent optimization method are used to update and output the initialized cost-sensitive matrix to obtain a maximized objective function; The cost-sensitive matrix corresponding to the maximized objective function is taken as an output result.

6. The method for early warning analysis of Alzheimer's disease based on multi-Chinese syllables according to claim 5, characterized in that: The maximization objective function is specifically: ； in, represents the objective function of classifier i, is the hidden layer output of classifier i, Represents the classification label, represents the classifier regularization parameter, is the cost-sensitive matrix, is the number of classifiers, is the number of cost-sensitive matrices.

7. The method for early warning analysis of Alzheimer's disease based on multi-Chinese syllables according to claim 6, characterized in that: The process of constructing a cost-sensitive weighted objective function based on the cost-sensitive matrix, respectively calculating the multi-syllable fusion analysis diagnosis results of different classifiers, and predicting the probability of Alzheimer's disease according to the diagnosis results is as follows: Constructing a weighted output matrix according to the cost-sensitive matrix and the hidden layer node output matrix; constructing a cost-sensitive weighted objective function according to the weighted output matrix and the maximization objective function, and calculating a fusion output matrix according to the cost-sensitive weighted objective function; Calculating a diagnosis result based on the weighted output matrix and the fused output matrix; The diagnostic results are analyzed using a voting method to obtain a fusion analysis diagnostic result.

8. The method for early warning analysis of Alzheimer's disease based on multi-Chinese syllables according to claim 7, characterized in that: The objective function based on cost-sensitive weighting is specifically: ； in, represents the original output weight of classifier k, is the weighted output matrix, specifically , is the weighted output matrix of classifier k, is the hidden layer output of classifier k, represents the regularization parameter of the fusion objective function, K is the number of classifiers, is the cost-sensitive matrix, and B is the fusion output matrix, specifically: , is the value of the cost-sensitive weighted objective function.

9. The method for early warning analysis of Alzheimer's disease based on multi-Chinese syllables according to claim 7, characterized in that: The process of analyzing the diagnosis results by the voting method to obtain the fusion analysis diagnosis results is as follows: Calculating the probability values ​​of the diagnosis results respectively, wherein the probability values ​​include two categories: Alzheimer's disease probability value and non-Alzheimer's disease probability value; Determine the probability values ​​of Alzheimer's disease and non-Alzheimer's disease in the probability values ​​of each diagnosis result, and vote for the classification with a large probability value; Count the votes and take the category with the most votes as the output value of the fusion analysis diagnosis result.

10. An Alzheimer's disease early warning analysis system based on multiple Chinese syllables, used to implement the Alzheimer's disease early warning analysis method based on multiple Chinese syllables as claimed in any one of claims 1 to 9, characterized in that: The system comprises: a data perception module (1), a data preprocessing module (2), a classifier module (3), a multi-syllable fusion diagnosis module (4) and an early warning screening and guidance suggestion module (5); The data perception module (1) is used to collect speech data of Alzheimer's patients and perform data perception to construct a polysyllabic sample; The data preprocessing module (2) is used to perform data preprocessing on the polysyllabic sample by adopting a multi-level graph embedding sparse feature learning method to construct a polysyllabic speech output feature; The classifier module (3) is used to construct a classifier, and uses a cost-sensitive random configuration network in the classifier to perform feature optimization processing on the multi-syllable speech output features, and outputs a cost-sensitive matrix; The polysyllabic fusion diagnosis module (4) is used to construct a cost-sensitive weighted objective function based on a cost-sensitive matrix, calculate the polysyllabic fusion analysis diagnosis results of different classifiers respectively, and predict the probability of Alzheimer's disease according to the diagnosis results; The early warning screening and guidance suggestion module (5) is used to provide guidance and suggestions to the patient based on the diagnosis result of the polysyllabic fusion diagnosis module (4).

Images