Natural language based assessment of communication efficacy decline in ad patients

Abstract

The application relates to the technical field of semantic processing, in particular to an AD patient communication efficiency decline evaluation method based on a natural language, which comprises the following steps: acquiring a voice signal of a user, and converting and dividing the voice signal into a plurality of effective vocabulary segments and invalid vocabulary segments; extracting the time length of each vocabulary in each effective vocabulary segment and the time length interval between adjacent vocabularies, obtaining the semantic fluency of the effective vocabulary segments in combination with the difference between adjacent effective vocabulary segments; according to the semantic fluency of the effective vocabulary segments, in combination with the invalid vocabulary segments between adjacent effective vocabulary segments, the overall fluency of the voice signal is obtained, and the communication efficiency decline of the user is visually demonstrated. Through segmented modeling and multi-level quantitative analysis on the voice signal, communication information density, semantic fluency and fluency promotion coefficients are constructed, which are used for evaluating the communication efficiency of the user, so that the accuracy of the communication efficiency evaluation result is improved.

Description

Technical Field

[0001] This invention relates to the field of semantic processing technology, and more specifically to a method for assessing the decline in communication effectiveness in AD patients based on natural language. Background Technology

[0002] As the population ages, the need for early screening and continuous monitoring of cognitive degenerative diseases (such as Alzheimer's disease and mild cognitive impairment) is increasing. Language expression ability is a crucial indicator of changes in cognitive state; as cognitive function declines, patients often experience weakened semantic organization, disordered expression rhythm, increased pauses, and decreased topic coherence. However, current assessments of communication ability largely rely on scale tests, manual interviews, or single acoustic feature analysis based on speech signals, such as speech rate statistics, pause count statistics, or word frequency analysis. These methods are used to assist in judging an individual's language expression status but lack systematic modeling of the relationship between semantic structure evolution and expression progression during continuous expression, making it difficult to accurately assess user communication effectiveness. Summary of the Invention

[0003] To address the technical problem that existing methods for assessing communication skills are insufficient to accurately evaluate users' communication effectiveness, the present invention aims to provide a natural language-based method for assessing the decline in communication effectiveness in AD patients. The specific technical solution adopted is as follows: Acquire the user's voice signal, and convert and divide the voice signal into several valid word segments and invalid word segments; Extract the duration of each word in each effective vocabulary segment and the duration interval between adjacent words to obtain the communication information density of each vocabulary segment; obtain the semantic fluency of the effective vocabulary segment based on the differences between adjacent effective vocabulary segments and the changes in their communication information density. Based on the semantic fluency of effective vocabulary segments, and combined with the invalid vocabulary segments between adjacent effective vocabulary segments, the fluency advancement coefficient of effective vocabulary segments is obtained; based on the fluency advancement coefficient of all effective vocabulary segments, the overall fluency of the speech signal is obtained; based on the fluency advancement coefficient of vocabulary segments, a visual demonstration of the decline in user communication effectiveness is provided.

[0004] Preferably, the step of acquiring the user's voice signal and converting and dividing the voice signal into several valid and invalid word segments includes: The system acquires the user's speech signal sequence using a microphone array; performs time-frequency transformation on the speech signal using a time-frequency analysis algorithm to obtain the time spectrum of the speech signal; calculates the spectral energy value at each moment based on the time spectrum to construct an energy sequence of spectral energy changing over time; converts the speech signal into a word sequence using a speech recognition algorithm and records the time range corresponding to each word in the word sequence; performs minimum value detection on the energy sequence using the AMPD algorithm to obtain several local minimum energy points as candidate pause points, and uses candidate pause points that are not located in the time range corresponding to each word as segmentation points to segment the word sequence, obtaining several word segments. For any word segment, if the word segment does not contain any words, the word segment is recorded as an invalid word segment; if the word segment contains words, the word segment is recorded as a valid word segment.

[0005] Preferably, the step of extracting the duration of each word in each effective vocabulary segment and the duration interval between adjacent words to obtain the communication information density of each vocabulary segment includes: For any valid vocabulary segment, a word segmentation algorithm is used to obtain all words in the valid vocabulary segment. All words in the valid vocabulary segment are matched with a preset meaningless word dictionary, and words belonging to the meaningless word dictionary are selected as meaningless words. Any word in the effective vocabulary segment is recorded as the target word, and words in the effective vocabulary segment that are adjacent to the target word are recorded as the neighboring words of the target word. The interval between the target word and its neighboring words in the speech signal is denoted as the pause duration between the target word and its neighboring words. The pause duration of the target word is the difference between the time span from the end of the pronunciation of the word preceding the target word to the start of the pronunciation of the word following it, minus the average pause duration of the target word and all its neighboring words. The communication information density of the effective vocabulary segment is obtained by combining the difference between the average duration of all words in the speech signal and the pause duration of each word, along with the number of all words in the effective vocabulary segment and the number of meaningless words.

[0006] Preferably, the step of obtaining the communication information density of the effective vocabulary segment based on the difference between the average duration of all words in the speech signal and the pause duration of each word, combined with the number of all words in the effective vocabulary segment and the number of meaningless words, includes: The ratio of the number of words in the effective vocabulary segment to the number of meaningless words is used as the information volume factor of the effective vocabulary segment. The difference between the average duration of all words in the effective vocabulary segment in the speech signal and the stuttering duration of each word is recorded as the fluency factor of each word in the effective vocabulary segment. The information volume factor of the effective vocabulary segment is multiplied by the sum of the fluency factors of all words in the effective vocabulary segment, and the product is taken as the communication information density of the effective vocabulary segment.

[0007] Preferably, obtaining the semantic fluency of an effective vocabulary segment based on the differences between adjacent effective vocabulary segments and the changes in their communicative information density includes: For any valid vocabulary segment, the next valid vocabulary segment is recorded as the continuation vocabulary segment. If the number of words in the valid vocabulary segment is greater than the number of words in the continuation vocabulary segment, then a preset fill word is added at the end of the continuation vocabulary segment until the number of words in the continuation vocabulary segment is equal to the number of words in the valid vocabulary segment. If the number of words in the effective vocabulary segment is less than the number of words in the continuous vocabulary segment, then a preset fill word is added at the end of the effective vocabulary segment until the number of words in the effective vocabulary segment equals the number of words in the continuous vocabulary segment. The word vector of each word in the effective vocabulary segment is obtained by using the word2vec algorithm. Then, the word vectors of the words in the effective vocabulary segment are concatenated according to the order of the words in the effective vocabulary segment to obtain the semantic vector of the effective vocabulary segment. Similarly, obtain the semantic vector of the continuation vocabulary segment; Based on the semantic vectors of the effective vocabulary segment and the semantic vectors of the continuating vocabulary segment, and combined with the communication information density of the effective vocabulary segment and the continuating vocabulary segment, the semantic fluency of the effective vocabulary segment is obtained.

[0008] Preferably, obtaining the semantic fluency of the effective vocabulary segment based on the semantic vector of the effective vocabulary segment and the semantic vector of the continuating vocabulary segment, combined with the communication information density of the effective vocabulary segment and the continuating vocabulary segment, includes: Based on the cosine similarity between the semantic vector of the effective vocabulary segment and the semantic vector of the continuation vocabulary segment, and combined with the difference in communication information density between the effective vocabulary segment and the continuation vocabulary segment, the semantic fluency of the effective vocabulary segment is obtained. The semantic fluency of the effective vocabulary segment is negatively correlated with the cosine similarity between the semantic vector of the effective vocabulary segment and the semantic vector of the continuating vocabulary segment. The semantic fluency of the effective vocabulary segment is positively correlated with the difference in communication information density between the effective vocabulary segment and the continuation vocabulary segment.

[0009] Preferably, obtaining the fluency advancement coefficient of a valid vocabulary segment based on the semantic fluency of the valid vocabulary segment and combining the invalid vocabulary segments between adjacent valid vocabulary segments includes: The sum of the duration corresponding to the effective vocabulary segment and the duration corresponding to the continuation vocabulary segment is recorded as the semantic integration duration of the effective vocabulary segment; the duration corresponding to the invalid vocabulary segment between the effective vocabulary segment and the continuation vocabulary segment is recorded as the communication interruption duration of the effective vocabulary segment. Based on the semantic integration time and communication interruption time of the effective vocabulary segment, and combined with the difference between the effective vocabulary segment and the continuation vocabulary segment, the smooth progress coefficient of the effective vocabulary segment is obtained.

[0010] Preferably, obtaining the fluency progression coefficient of the effective vocabulary segment based on the semantic integration time and communication interruption time of the effective vocabulary segment, combined with the difference between the effective vocabulary segment and the continuing vocabulary segment, includes: The semantic continuity of the effective vocabulary segment is determined by the ratio of the semantic integration time to the communication interruption time. Based on the semantic continuity of the effective vocabulary segment and the difference between the effective vocabulary segment and the continuation vocabulary segment, the smoothness progression coefficient of the effective vocabulary segment is obtained. The fluency progression coefficient of the effective vocabulary segment is positively correlated with the semantic continuity of the effective vocabulary segment; the fluency progression coefficient of the effective vocabulary segment is negatively correlated with the difference between the effective vocabulary segment and the continuating vocabulary segment.

[0011] Preferably, obtaining the overall fluency of the speech signal based on the fluency progression coefficients of all valid word segments includes: The kurtosis of the smoothness progression coefficients of all valid vocabulary segments in the speech signal is obtained. The kurtosis of the smoothness progression coefficients of all valid vocabulary segments in the speech signal is normalized, and the normalized result is used as the overall smoothness of the speech signal.

[0012] Preferably, the visualization demonstration of the decline in user communication effectiveness based on the fluency progression coefficient of the vocabulary segment includes: The smoothness progression coefficient of invalid word segments is set to 0. The smoothness progression coefficient of all word segments is visualized and mapped according to the green-red false color mapping method. The communication efficiency decay of all speech segments in the speech signal is visualized and demonstrated.

[0013] This invention has the following beneficial effects: By dividing speech signals into effective and invalid lexical segments and constructing quantitative indicators at multiple levels—vocabulary level, lexical segment level, and overall speech level—this application achieves hierarchical modeling and analysis of continuous speech expression structure; by introducing a communication information density indicator, it integrates word pronunciation duration, pause intervals, and the proportion of meaningless words to accurately quantify speech expression efficiency and rhythmic stability; by constructing a semantic fluency indicator, it comprehensively analyzes the semantic differences and information density changes between adjacent lexical segments, quantifying the quality of semantic cohesion; further, by integrating semantic continuity and expression interruption structure through a fluency progression coefficient, it can reflect the stability of the expression process in the time dimension; finally, through overall fluency statistics and visualization mapping, the trend of communication effectiveness changes can be presented in an intuitive form, thereby improving the accuracy of the evaluation results. Attached Figure Description

[0014] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 The flowchart illustrates a method for assessing the decline in communication effectiveness in AD patients based on natural language, as provided in one embodiment of the present invention. Detailed Implementation

[0016] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the natural language-based method for assessing the decline in communication effectiveness in AD patients proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0017] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0018] The specific scheme of the natural language-based AD patient communication efficacy decline assessment method provided by the present invention will be described in detail below with reference to the accompanying drawings.

[0019] Please see Figure 1 The diagram illustrates a flowchart of a natural language-based method for assessing the decline in communication effectiveness in AD patients, according to an embodiment of the present invention. The method includes: Step S101: Acquire the user's voice signal, and convert and divide the voice signal into several valid word segments and invalid word segments.

[0020] It should be noted that in the assessment of the decline in communication effectiveness in AD patients, the decline in language expression ability is usually not reflected in the content of a single word, but rather in dynamic characteristics such as rhythm disorder, decreased semantic cohesion, and increased frequency of pauses and interruptions in the speech expression process. Therefore, this embodiment converts and divides the speech signal into several effective and ineffective word segments. The effective word segments are used to represent the semantic information content actually output by the patient, and the ineffective word segments are used to represent speech segments that do not form effective semantic output. This quantifies the user's semantic output ability and the degree of expression blockage, providing a structured basis for the subsequent construction of communication information density, semantic fluency, and fluency progression coefficient.

[0021] Step S102: Extract the duration of each word in each effective vocabulary segment and the duration interval between adjacent words to obtain the communication information density of each vocabulary segment; obtain the semantic fluency of the effective vocabulary segment based on the differences between adjacent effective vocabulary segments and the changes in their communication information density.

[0022] It should be noted that the decline in communication ability often manifests as a decrease in the amount of effective semantic information output per unit time, an increase in the proportion of meaningless filler words, and a disordered rhythm of vocabulary expression. Therefore, this embodiment extracts the pronunciation duration of each word in an effective vocabulary segment and the pause interval between adjacent words to construct a communication information density index that reflects the efficiency of effective semantic output, which is used to quantify the patient's information organization ability within a single effective vocabulary segment. Furthermore, analyzing only a single expression unit is insufficient to characterize the semantic cohesion ability during continuous expression. Since AD ​​patients often exhibit semantic jumps, repetitions, or topic breaks during continuous expression, this embodiment analyzes the semantic differences between adjacent effective vocabulary segments and combines this with changes in their communication information density to construct a semantic fluency index, thereby comprehensively quantifying the continuity and semantic organization stability of the user's expression.

[0023] Step S103: Based on the semantic fluency of the effective vocabulary segments and the invalid vocabulary segments between adjacent effective vocabulary segments, obtain the fluency advancement coefficient of the effective vocabulary segments; based on the fluency advancement coefficients of all effective vocabulary segments, obtain the overall fluency of the speech signal; based on the fluency advancement coefficients of the vocabulary segments, visualize the decline in user communication effectiveness.

[0024] It should be noted that communication impairments in AD patients are not only reflected in a decline in semantic organization ability, but also in dynamic blockage characteristics such as increased pauses and prolonged speech interruptions during expression. Therefore, this embodiment, based on semantic fluency, further combines the interruption duration corresponding to invalid lexical segments between adjacent effective lexical segments to comprehensively quantify the ability to continuously advance expression, obtaining a fluency advancement coefficient. This coefficient is used to represent the mutual constraint relationship between semantic cohesion ability and the duration of expression interruptions, thereby reflecting the continuity level of the expression advancement process. When semantic fluency is high and the interruption duration is short, the fluency advancement coefficient is relatively high, indicating that the expression process is continuous and stable. When semantic differences increase or the interruption duration increases, the fluency advancement coefficient decreases accordingly, thereby reflecting the degree of obstruction in expression advancement. At the same time, in order to evaluate the communication effectiveness of the speech signal from an overall perspective, this embodiment performs statistical analysis on the fluency advancement coefficients of all effective lexical segments. By characterizing the concentration and fluctuation characteristics of the advancement coefficient distribution pattern, an overall fluency index of the speech signal is obtained to quantify the stability and continuity level of the user's expression during the complete dialogue process.

[0025] It should be further explained that, after obtaining the fluency progression coefficient of each vocabulary segment, in order to better analyze the fluctuation pattern of expression progression ability and locate potential communication barriers, this embodiment performs a visualization mapping based on the fluency progression coefficient of each vocabulary segment, transforming the continuous expression process into a graphical form with time distribution characteristics, so as to intuitively present the changes in communication effectiveness at different expression stages, thereby improving the accuracy of the evaluation results.

[0026] Step S201: Acquire the user's voice signal, and convert and divide the voice signal into several valid word segments and invalid word segments.

[0027] Specifically, the user's speech signal sequence is acquired through a microphone array; the speech signal is transformed into a time-frequency spectrum using a time-frequency analysis algorithm; the spectral energy value at each moment is calculated based on the time-frequency spectrum to construct an energy sequence in which the spectral energy changes over time; the speech signal is converted into a word sequence using a speech recognition algorithm, and the time range corresponding to each word in the word sequence is recorded; the energy sequence is subjected to minimum value detection using the AMPD algorithm to obtain several local minimum energy points as candidate pause points, and the candidate pause points that are not located in the time range corresponding to each word are used as segmentation points to segment the word sequence, resulting in several word segments. Since the time-frequency analysis algorithm and the AMPD algorithm are well-known existing technologies, they will not be described in detail in this embodiment. For any word segment, if the word segment does not contain any words, the word segment is recorded as an invalid word segment; if the word segment contains words, the word segment is recorded as a valid word segment.

[0028] It should be noted that pauses, interruptions, or speech intervals in continuous speech signals typically manifest as a significant decrease in spectral energy. Therefore, by performing minimum value detection on the time-varying sequence of the spectral energy of the speech signal, we can accurately locate natural dividing points in the expressive rhythm, making the segmentation of speech segments more consistent with the actual expressive rhythm characteristics. Furthermore, using the presence or absence of identifiable words as the criterion for dividing effective and ineffective lexical segments helps to distinguish between actual semantic output content and content that has not formed effective semantic content. This allows for the separation of the semantic generation process from the expressive interruption behavior while preserving the speech temporal structure information, thereby accurately assessing the decline in user communication effectiveness.

[0029] Step S202: Extract the duration of each word in each effective vocabulary segment and the duration interval between adjacent words to obtain the communication information density of each vocabulary segment; obtain the semantic fluency of the effective vocabulary segment based on the differences between adjacent effective vocabulary segments and the changes in their communication information density.

[0030] Specifically, for any valid vocabulary segment, a word segmentation algorithm is used to obtain all words in the valid vocabulary segment. All words in the valid vocabulary segment are matched with a preset meaningless word dictionary, and words belonging to the meaningless word dictionary are selected as meaningless words. Any word in the effective vocabulary segment is recorded as the target word, and words in the effective vocabulary segment that are adjacent to the target word are recorded as the neighboring words of the target word. The interval between the target word and its neighboring words in the speech signal is denoted as the pause duration between the target word and its neighboring words. The pause duration of the target word is calculated by subtracting the average pause duration of the target word and all its neighboring words from the time span between the end of pronunciation of the word preceding the target word and the start of pronunciation of the word following the target word. (If the target word has no preceding word, the pause duration of the target word is calculated by subtracting the average pause duration of the target word and all its neighboring words from the time span between the start of the effective word segment and the start of pronunciation of the word following the target word; if the target word has no following word, the pause duration of the target word is calculated by subtracting the average pause duration of the target word and all its neighboring words from the time span between the end of pronunciation of the word preceding the target word and the end of the effective word segment.) The ratio of the number of words in the effective vocabulary segment to the number of meaningless words is used as the information volume factor of the effective vocabulary segment. The difference between the average duration of all words in the effective vocabulary segment in the speech signal and the stuttering duration of each word is recorded as the fluency factor of each word in the effective vocabulary segment. The information volume factor of the effective vocabulary segment is multiplied by the sum of the fluency factors of all words in the effective vocabulary segment, and the product is taken as the communication information density of the effective vocabulary segment.

[0031] As an example, the specific formula for calculating the communication information density of the effective vocabulary segment is as follows: In the formula, This indicates the communication information density of the effective vocabulary segment; This indicates the number of words in the valid vocabulary segment; This indicates the number of meaningless words in the valid vocabulary segment; This represents the average duration of all words in the valid vocabulary segment in the speech signal; Indicates the first word in the valid vocabulary segment The duration of pause for each word; This represents a pre-defined, extremely small positive number. The specific value can be set according to the actual situation. This embodiment does not make a hard requirement. In this embodiment, it is used as... The purpose of this example is to avoid the situation where the denominator is 0 when performing fractional operations, which would cause the division by zero to crash.

[0032] It should be noted that the communication information density is a dimensionless engineering index in the data feature processing process. The lower the proportion of meaningless words used by users during communication, the higher the semantic output efficiency. The larger the value, the higher the semantic output efficiency and the greater the information density; This represents the difference between the average duration of all words in the effective vocabulary segment and the pause duration of each word in the speech signal. The pause duration represents the effective time share of a word used for semantic output during expression. When the pause duration increases, this difference decreases, indicating that the degree of interference from pauses during word expression increases. Furthermore, the difference between the net expression duration of each word and the overall average phonation duration of the vocabulary segment is calculated to quantify the consistency between the expression rhythm of each word and the overall expression benchmark. When the net expression duration is close to the mean, it indicates that the expression rhythm is stable and the structure of communication information output is balanced. When the net expression duration is significantly lower than the mean, it indicates that there is expression delay or pause phenomenon, which affects the communication information density.

[0033] Furthermore, for any valid vocabulary segment, the next valid vocabulary segment is denoted as the continuation vocabulary segment. If the number of words in the valid vocabulary segment is greater than the number of words in the continuation vocabulary segment, then a preset fill word is added at the end of the continuation vocabulary segment until the number of words in the continuation vocabulary segment is equal to the number of words in the valid vocabulary segment. The word vector of the fill word is a zero vector. If the number of words in the effective vocabulary segment is less than the number of words in the continuous vocabulary segment, then a preset fill word is added at the end of the effective vocabulary segment until the number of words in the effective vocabulary segment equals the number of words in the continuous vocabulary segment. The word vector of each word in the effective vocabulary segment is obtained by using the word2vec algorithm. Then, the word vectors of the words in the effective vocabulary segment are concatenated according to the order of the words in the effective vocabulary segment to obtain the semantic vector of the effective vocabulary segment. Similarly, the semantic vectors of the continuation word segments are obtained. Since the word2vec algorithm is a well-known existing technology, it will not be described in detail in this embodiment.

[0034] Based on the semantic vectors of the effective vocabulary segment and the semantic vectors of the continuating vocabulary segment, and combined with the communication information density of the effective vocabulary segment and the continuating vocabulary segment, the semantic fluency of the effective vocabulary segment is obtained. The semantic fluency of the effective vocabulary segment is negatively correlated with the cosine similarity between the semantic vector of the effective vocabulary segment and the semantic vector of the continuating vocabulary segment. The semantic fluency of the effective vocabulary segment is positively correlated with the difference in communication information density between the effective vocabulary segment and the continuating vocabulary segment (for the last effective vocabulary segment, let the semantic fluency of the last effective vocabulary segment be equal to the semantic fluency of the second-to-last effective vocabulary segment).

[0035] As an example, the specific formula for calculating the semantic fluency of effective vocabulary segments is as follows: In the formula, This indicates the semantic fluency of the effective vocabulary segment; The word vector representing the set of feature words of the effective vocabulary segment; Word vectors representing the set of characteristic words in a continuation segment; This indicates the communication information density of the effective vocabulary segment; Indicates the density of communication information in a continuous vocabulary segment; Take the cosine function; This represents the function that takes the absolute value. This represents a pre-defined, extremely small positive number. The specific value can be set according to the actual situation. This embodiment does not make a hard requirement. In this embodiment, it is used as... The example is used to illustrate this point, in order to avoid the situation where the denominator is zero during fraction operations.

[0036] It should be noted that the semantic fluency is a dimensionless engineering index in the data feature processing process. Since the vocabulary size of different effective vocabulary segments varies, this embodiment constructs a benchmark vocabulary set for the current effective vocabulary segment and its continuating vocabulary segments, and performs unified imputation processing on missing words, so that the two vocabulary segments are vectorized in a unified semantic space, thereby accurately quantifying the semantic similarity between them. Furthermore, cosine similarity is used to characterize the semantic direction consistency between adjacent vocabulary segments. When the cosine similarity decreases, it indicates that the semantic topic has jumped or broken. At the same time, the relative difference between communication information densities is used to characterize the degree of change in expression intensity. When the difference in information density increases, it indicates that the expression rhythm or information output efficiency has fluctuated. In this way, the semantic fluency of the effective vocabulary segment is obtained, thereby obtaining the user's semantic organization ability and topic continuity ability in the continuous expression process.

[0037] Step S203: Based on the semantic fluency of the effective vocabulary segments and the invalid vocabulary segments between adjacent effective vocabulary segments, obtain the fluency advancement coefficient of the effective vocabulary segments; based on the fluency advancement coefficients of all effective vocabulary segments, obtain the overall fluency of the speech signal.

[0038] Specifically, the sum of the duration corresponding to the effective vocabulary segment and the duration corresponding to the continuation vocabulary segment is recorded as the semantic integration duration of the effective vocabulary segment; the duration corresponding to the invalid vocabulary segment between the effective vocabulary segment and the continuation vocabulary segment is recorded as the communication interruption duration of the effective vocabulary segment. The semantic continuity of the effective vocabulary segment is determined by the ratio of the semantic integration time to the communication interruption time. Based on the semantic continuity of the effective vocabulary segment and the difference between the effective vocabulary segment and the continuation vocabulary segment, the smoothness progression coefficient of the effective vocabulary segment is obtained. The smoothness progression coefficient of the effective vocabulary segment is positively correlated with the semantic continuity of the effective vocabulary segment; the smoothness progression coefficient of the effective vocabulary segment is negatively correlated with the difference between the effective vocabulary segment and the continuating vocabulary segment (for the last effective vocabulary segment, let the smoothness progression coefficient of the last effective vocabulary segment be equal to the smoothness progression coefficient of the second-to-last effective vocabulary segment).

[0039] As an example, the formula for calculating the fluency advancement coefficient of the effective vocabulary segment is: In the formula, This represents the smoothness progression coefficient of the effective vocabulary segment; This indicates the semantic integration time of the effective vocabulary segment; This indicates the duration of communication interruption for the aforementioned valid vocabulary segment; This indicates the semantic fluency of the effective vocabulary segment; This represents the average semantic fluency of all valid word segments; The function represents a linear normalization function, specifically normalized using the maximum and minimum value normalization method. Its normalized value range is [0,1]. The maximum and minimum values ​​can be obtained based on historical experiments or prior experience. Adjusting, calibrating, or optimizing the maximum and minimum values ​​does not constitute a limitation of this invention. This represents the function that takes the maximum value. This represents a pre-defined, extremely small positive number. The specific value can be set according to the actual situation. This embodiment does not make a hard requirement. In this embodiment, it is used as... The example is used to illustrate this point, in order to avoid the situation where the denominator is zero during fraction operations.

[0040] It should be noted that the smoothness advancement coefficient is a dimensionless engineering index in the data feature processing process, and the semantic integration time and communication interruption time represent the continuous semantic output time and the expression interruption time, respectively. This represents the ratio between the time during which meaning can be recognized and the time during which meaning cannot be recognized during the transition from one effective lexical segment to the next. A larger value indicates a tighter semantic connection and stronger semantic continuity during the transition. Furthermore, combined with... The reciprocal term makes the fluency progression coefficient not only affected by the temporal structure but also by the degree of semantic fluctuation. When the semantic fluency of a certain effective lexical segment... Close to the overall average level When this condition stabilizes, it indicates that the semantic change is within the normal fluctuation range. When the deviation from the mean is significant, it indicates that the semantic jump or expression imbalance is aggravated. The reciprocal term has an inhibitory effect on the propagation coefficient, thereby reducing the smooth propagation coefficient of the effective vocabulary segment, thus quantifying the semantic expression propagation ability.

[0041] Furthermore, the kurtosis of the smoothness progression coefficients of all effective vocabulary segments in the speech signal is obtained, and the kurtosis of the smoothness progression coefficients of all effective vocabulary segments in the speech signal is normalized. The normalized result is used as the overall smoothness of the speech signal. Since kurtosis is a well-known prior art, it will not be described in detail in this embodiment.

[0042] It should be noted that the fluency advancement coefficient is used to characterize the local advancement ability of each effective lexical segment under the combined effect of semantic continuity and expression stability. By calculating the kurtosis of the fluency advancement coefficient of all effective lexical segments, the concentration and stability of the expression advancement ability in the time dimension can be reflected from the perspective of the overall distribution pattern. When the distribution of the fluency advancement coefficient is more concentrated and the kurtosis is higher, it indicates that most lexical segments maintain a relatively consistent advancement level and the overall expression structure is relatively stable. When the distribution tends to be dispersed and the kurtosis decreases, it indicates that the expression advancement ability fluctuates significantly and the communication structure is loose.

[0043] Furthermore, the smoothness advancement coefficient of invalid word segments is set to 0. The smoothness advancement coefficient of all word segments is visualized and mapped according to the green-red false color mapping method. The communication efficiency degradation of all speech segments in the speech signal is visualized and demonstrated. Since the green-red false color mapping method is a well-known existing technology, it will not be described in detail in this embodiment. If the variance of the smoothness advancement coefficient of all valid word segments is 0, the mean of the smoothness advancement coefficient is directly mapped to the overall smoothness. If the variance is not 0, the kurtosis of the smoothness advancement coefficient is calculated to avoid the collapse when obtaining the peak value.

[0044] It should be noted that by setting the fluency progression coefficient of invalid vocabulary segments to 0, a clear low-value benchmark is formed on the time axis for the expression interruption interval. The fluency progression coefficient of all vocabulary segments is visualized through a green-red false color mapping method, transforming the abstract numerical distribution into a continuous color gradient change. This provides an overall fluency statistical result at the macro level, while also showing the specific location and trend of communication effectiveness decline at the micro level, thus improving the intuitiveness of the evaluation results.

[0045] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0046] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for assessing the decline in communication efficacy in AD patients based on natural language processing, characterized in that... The method includes: Acquire the user's voice signal, and convert and divide the voice signal into several valid word segments and invalid word segments; Extract the duration of each word in each effective vocabulary segment and the duration interval between adjacent words to obtain the communication information density of each vocabulary segment; obtain the semantic fluency of the effective vocabulary segment based on the differences between adjacent effective vocabulary segments and the changes in their communication information density. Based on the semantic fluency of effective vocabulary segments, and combined with the invalid vocabulary segments between adjacent effective vocabulary segments, the fluency advancement coefficient of effective vocabulary segments is obtained; based on the fluency advancement coefficient of all effective vocabulary segments, the overall fluency of the speech signal is obtained; based on the fluency advancement coefficient of vocabulary segments, a visual demonstration of the decline in user communication effectiveness is provided.

2. The method for assessing the decline in communication efficacy in AD patients based on natural language, as described in claim 1, is characterized in that... The process of acquiring the user's voice signal and converting and dividing the voice signal into several valid and invalid word segments includes: The system acquires the user's speech signal sequence using a microphone array; performs time-frequency transformation on the speech signal using a time-frequency analysis algorithm to obtain the time spectrum of the speech signal; calculates the spectral energy value at each moment based on the time spectrum to construct an energy sequence of spectral energy changing over time; converts the speech signal into a word sequence using a speech recognition algorithm and records the time range corresponding to each word in the word sequence; performs minimum value detection on the energy sequence using the AMPD algorithm to obtain several local minimum energy points as candidate pause points, and uses candidate pause points that are not located in the time range corresponding to each word as segmentation points to segment the word sequence, obtaining several word segments. For any word segment, if the word segment does not contain any words, the word segment is recorded as an invalid word segment; if the word segment contains words, the word segment is recorded as a valid word segment.

3. The method for assessing the decline in communication efficacy in AD patients based on natural language, as described in claim 1, is characterized in that... The step of extracting the duration of each word in each effective vocabulary segment and the duration interval between adjacent words to obtain the communication information density of each vocabulary segment includes: For any valid vocabulary segment, a word segmentation algorithm is used to obtain all words in the valid vocabulary segment. All words in the valid vocabulary segment are matched with a preset meaningless word dictionary, and words belonging to the meaningless word dictionary are selected as meaningless words. Any word in the effective vocabulary segment is recorded as the target word, and words in the effective vocabulary segment that are adjacent to the target word are recorded as the neighboring words of the target word. The interval between the target word and its neighboring words in the speech signal is denoted as the pause duration between the target word and its neighboring words. The pause duration of the target word is the difference between the time span from the end of the pronunciation of the word preceding the target word to the start of the pronunciation of the word following it, minus the average pause duration of the target word and all its neighboring words. The communication information density of the effective vocabulary segment is obtained by combining the difference between the average duration of all words in the speech signal and the pause duration of each word, along with the number of all words in the effective vocabulary segment and the number of meaningless words.

4. The method for assessing the decline in communication effectiveness in AD patients based on natural language, as described in claim 3, is characterized in that... The step of obtaining the communication information density of the effective vocabulary segment by combining the difference between the average duration of all words in the speech signal and the pause duration of each word, and by combining the number of all words in the effective vocabulary segment and the number of meaningless words, includes: The ratio of the number of words in the effective vocabulary segment to the number of meaningless words is used as the information volume factor of the effective vocabulary segment. The difference between the average duration of all words in the effective vocabulary segment in the speech signal and the stuttering duration of each word is recorded as the fluency factor of each word in the effective vocabulary segment. The information volume factor of the effective vocabulary segment is multiplied by the sum of the fluency factors of all words in the effective vocabulary segment, and the product is taken as the communication information density of the effective vocabulary segment.

5. The method for assessing the decline in communication efficacy in AD patients based on natural language, as described in claim 1, is characterized in that... The step of obtaining the semantic fluency of effective vocabulary segments based on the differences between adjacent effective vocabulary segments and the changes in their communicative information density includes: For any valid vocabulary segment, the next valid vocabulary segment is recorded as the continuation vocabulary segment. If the number of words in the valid vocabulary segment is greater than the number of words in the continuation vocabulary segment, then a preset fill word is added at the end of the continuation vocabulary segment until the number of words in the continuation vocabulary segment is equal to the number of words in the valid vocabulary segment. If the number of words in the effective vocabulary segment is less than the number of words in the continuous vocabulary segment, then a preset fill word is added at the end of the effective vocabulary segment until the number of words in the effective vocabulary segment equals the number of words in the continuous vocabulary segment. The word vector of each word in the effective vocabulary segment is obtained by using the word2vec algorithm. Then, the word vectors of the words in the effective vocabulary segment are concatenated according to the order of the words in the effective vocabulary segment to obtain the semantic vector of the effective vocabulary segment. Similarly, obtain the semantic vector of the continuation word segment; Based on the semantic vectors of the effective vocabulary segment and the semantic vectors of the continuating vocabulary segment, and combined with the communication information density of the effective vocabulary segment and the continuating vocabulary segment, the semantic fluency of the effective vocabulary segment is obtained.

6. The method for assessing the decline in communication efficacy in AD patients based on natural language, as described in claim 5, is characterized in that... The step of obtaining the semantic fluency of the effective vocabulary segment based on the semantic vector of the effective vocabulary segment and the semantic vector of the continuating vocabulary segment, combined with the communication information density of the effective vocabulary segment and the continuating vocabulary segment, includes: Based on the cosine similarity between the semantic vector of the effective vocabulary segment and the semantic vector of the continuation vocabulary segment, and combined with the difference in communication information density between the effective vocabulary segment and the continuation vocabulary segment, the semantic fluency of the effective vocabulary segment is obtained. The semantic fluency of the effective vocabulary segment is negatively correlated with the cosine similarity between the semantic vector of the effective vocabulary segment and the semantic vector of the continuating vocabulary segment. The semantic fluency of the effective vocabulary segment is positively correlated with the difference in communication information density between the effective vocabulary segment and the continuation vocabulary segment.

7. The method for assessing the decline in communication effectiveness in AD patients based on natural language, as described in claim 5, is characterized in that... The step of obtaining the fluency advancement coefficient of a valid vocabulary segment based on the semantic fluency of the valid vocabulary segment, combined with the invalid vocabulary segments between adjacent valid vocabulary segments, includes: The sum of the duration corresponding to the effective vocabulary segment and the duration corresponding to the continuation vocabulary segment is recorded as the semantic integration duration of the effective vocabulary segment; the duration corresponding to the invalid vocabulary segment between the effective vocabulary segment and the continuation vocabulary segment is recorded as the communication interruption duration of the effective vocabulary segment. Based on the semantic integration time and communication interruption time of the effective vocabulary segment, and combined with the difference between the effective vocabulary segment and the continuation vocabulary segment, the smooth progress coefficient of the effective vocabulary segment is obtained.

8. The method for assessing the decline in communication effectiveness in AD patients based on natural language, as described in claim 7, is characterized in that... The step of obtaining the fluency progression coefficient of the effective vocabulary segment based on the semantic integration time and communication interruption time of the effective vocabulary segment, combined with the difference between the effective vocabulary segment and the continuing vocabulary segment, includes: The semantic continuity of the effective vocabulary segment is determined by the ratio of the semantic integration time to the communication interruption time. Based on the semantic continuity of the effective vocabulary segment and the difference between the effective vocabulary segment and the continuation vocabulary segment, the smoothness progression coefficient of the effective vocabulary segment is obtained. The fluency progression coefficient of the effective vocabulary segment is positively correlated with the semantic continuity of the effective vocabulary segment; the fluency progression coefficient of the effective vocabulary segment is negatively correlated with the difference between the effective vocabulary segment and the continuating vocabulary segment.

9. The method for assessing the decline in communication efficacy in AD patients based on natural language, as described in claim 1, is characterized in that... The process of obtaining the overall fluency of the speech signal based on the fluency progression coefficients of all valid word segments includes: The kurtosis of the smoothness progression coefficients of all valid vocabulary segments in the speech signal is obtained. The kurtosis of the smoothness progression coefficients of all valid vocabulary segments in the speech signal is normalized, and the normalized result is used as the overall smoothness of the speech signal.

10. The method for assessing the decline in communication efficacy in AD patients based on natural language according to claim 1, characterized in that, The visualization demonstration of the decline in user communication effectiveness based on the fluency progression coefficient of vocabulary segments includes: The smoothness progression coefficient of invalid word segments is set to 0. The smoothness progression coefficient of all word segments is visualized and mapped according to the green-red false color mapping method. The communication efficiency decay of all speech segments in the speech signal is visualized and demonstrated.

Images