Mental health state assessment method based on big data

Abstract

The invention discloses a psychological health state assessment method based on big data, and the method comprises the steps: obtaining social media text information, voice data and physiological signals of a user, and carrying out the preprocessing of the social media text information, the voice data and the physiological signals; constructing an emotion vector by using the preprocessed social media text information, voice data and physiological signals, performing emotion division on the emotion vector through a K-means clustering algorithm, and updating the emotion vector to obtain an emotion stability index; generating personalized psychological health suggestions of the user based on the emotional stability index, dynamically adjusting the emotional state of the user according to the condition that the user accepts the personalized psychological health suggestions, and re-updating the personalized psychological health suggestions; according to the method, the emotion fluctuation of the user can be comprehensively captured, more accurate psychological health assessment is provided, and the timeliness and effectiveness of psychological health suggestions are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of mental health assessment, and in particular to a mental health status assessment method based on big data. Background Art

[0002] In recent years, mental health issues have received increasing attention from society, especially in the rapidly developing digital age. A large number of users share their personal emotions and mental states on social platforms, generating rich data resources. These data resources provide a new perspective for the assessment of personal mental health, making mental health monitoring methods gradually gain attention. Current technologies mainly focus on analyzing the user's emotional state, extracting the user's emotional features through various sentiment analysis algorithms, and using some machine learning techniques for sentiment classification and evaluation. However, there are often some limitations in the methods, such as the complexity of data processing, the lack of accuracy in sentiment classification, and the lag in the generation of personalized recommendations.

[0003] In practical applications, the deficiencies in the analysis of emotional stability are particularly prominent. Many existing systems rely only on static models and fail to fully consider the dynamic changes in the user's emotional state, resulting in the lack of personalization and timeliness of the provided suggestions. For example, traditional emotion classification methods often ignore the instantaneous nature of emotional changes and fail to capture the user's emotional fluctuations in a timely manner. In addition, evaluations based on physiological signals often fail to effectively integrate multiple data related to the user's emotions, resulting in inaccurate judgments on the user's emotional state. These problems have already affected the effectiveness of mental health interventions. Summary of the invention

[0004] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0005] In view of the above existing problems, the present invention is proposed. Therefore, the present invention provides a method for evaluating mental health status based on big data to solve the problems raised in the background technology.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for evaluating mental health status based on big data, comprising:

[0007] Acquire social media text information, voice data, and physiological signals of the user, and pre-process the social media text information, voice data, and physiological signals;

[0008] Using the preprocessed social media text information, voice data and physiological signals, construct the emotion vector, divide the emotion vector into emotions through the K-means clustering algorithm, and update the emotion vector to obtain the emotion stability index;

[0009] Based on the emotional stability index, a personalized mental health suggestion for the user is generated, and according to the user's acceptance of the personalized mental health suggestion, the user's emotional state is dynamically adjusted, and the personalized mental health suggestion is re-updated.

[0010] As a preferred solution of the mental health status assessment method based on big data described in the present invention, the social media text information is preprocessed, including:

[0011] Remove HTML tags, punctuation marks and special characters from social media text messages to obtain plain text, use a word segmentation tool to segment the plain text, and remove stop words from the plain text to obtain valid text;

[0012] The sentiment dictionary is applied to assign a sentiment score to each word in the valid text, and the sentiment strength TS is calculated.

[0013] As a preferred solution of the method for evaluating mental health status based on big data described in the present invention, the speech data is preprocessed, including:

[0014] The speech data is divided into several audio signal segments by frequency division, and the characteristic changes of each audio signal segment are determined by zero crossing rate analysis. The audio signal segment is divided into several sub-segments according to the characteristic changes, and the length of each sub-segment is adjusted according to the natural pauses and emotional changes of the speech, and the emotional change points are identified;

[0015] Determine whether the emotion change point is an emotion turning point. If it is an emotion turning point, mark the current sub-segment as an emotion turning point segment, and extract the pitch P and speech rate R of the audio signal segment to which the sub-segment belongs through the Mel-frequency cepstral coefficient. Otherwise, do not process it and continue to determine the emotion change point in the next sub-segment.

[0016] As a preferred solution of the method for evaluating mental health status based on big data described in the present invention, the physiological signal is preprocessed, including:

[0017] The physiological signals include heart rate HR and galvanic skin response GSR. Several time windows are created to segment the physiological signals, and the average value of all time windows is used as a dynamic baseline.

[0018] When the dynamic baseline changes, the time point corresponding to the dynamic baseline of the current time window is recorded, and the time points of the dynamic baseline in each time window are connected to form the peak of the time window. The emotional state is defined as "positive fluctuation" and "negative fluctuation" according to the rising and falling trends of the peak of the time window, respectively. The emotional state assessment score S is obtained from the "positive fluctuation" and "negative fluctuation".

[0019] As a preferred solution of the mental health status assessment method based on big data described in the present invention, the emotion vector is constructed by using pre-processed social media text information, voice data and physiological signals, including:

[0020] The sentiment vector is expressed as:

[0021] V=[TS,P,R,S]

[0022] Each time the preprocessed data is constructed into a sentiment vector and saved in the sentiment vector database;

[0023] Among them, the emotional state assessment score S (HR, GSR).

[0024] As a preferred solution of the mental health status assessment method based on big data described in the present invention, wherein: the emotion vector is divided into emotions by a K-means clustering algorithm, and the emotion vector is updated at the same time, including:

[0025] The emotion vector is divided into K emotion categories, and N initial cluster centers are randomly selected from the K emotion categories. The distance between each emotion vector in the emotion vector database and each cluster center is calculated, and the cluster center with the smallest distance is assigned to each emotion vector in descending order to form M clusters.

[0026] For each cluster g, update its cluster center to the mean of all sentiment vectors in the cluster.

[0027] As a preferred solution of the method for evaluating mental health status based on big data described in the present invention, it also includes:

[0028] Check whether the cluster center has changed. If the cluster center has not changed, the algorithm converges and stops iterating; otherwise, recalculate the distance and assign the cluster, and continue iterating.

[0029] As a preferred embodiment of the method for evaluating mental health status based on big data described in the present invention, the emotional stability index is obtained, including:

[0030] When the cluster center of each cluster g is updated, the internal change of each sentiment vector is calculated to obtain the sentiment stability index ESI g .

[0031] As a preferred solution of the mental health status assessment method based on big data described in the present invention, wherein: based on the emotional stability index, generating personalized mental health advice for the user includes:

[0032] If ESI g <0.2, the emotion type is “negative”;

[0033] If 0.2≤ESI g <0.5, the sentiment type is “neutral”;

[0034] If ESI g ≥0.5, the sentiment type is “positive”;

[0035] Generate personalized mental health advice for users based on the determination of emotion type.

[0036] As a preferred solution of the mental health status assessment method based on big data described in the present invention, the user's emotional state is dynamically adjusted according to the user's acceptance of personalized mental health advice, and the personalized mental health advice is updated, including:

[0037] If the user accepts the personalized mental health advice, the user's current emotional state is recorded, and the user's feedback results are regularly collected through questionnaires, and the feedback results are quantified, and the standard deviation and mean of the quantified results are calculated to obtain the user's acceptance value;

[0038] The user's emotional state is adjusted according to the user acceptance value, and the personalized mental health advice is updated.

[0039] Compared with the prior art, the invention has the following beneficial effects:

[0040] 1. By collecting social media text information, voice data and physiological signals to build emotion vectors, a multi-dimensional evaluation method is provided for mental health management systems or models, which not only improves the judgment accuracy of emotional states, but also enhances their personalization and pertinence, effectively solving the problem of the lack of flexibility of traditional methods.

[0041] 3. Using the K-means clustering algorithm to dynamically calculate the emotional stability index, it can reflect the fluctuation of user emotions in real time, solving the problem that traditional methods cannot capture instantaneous emotional changes in time, and providing a more sensitive means of monitoring emotional status;

[0042] 3. Dynamically adjust emotional states through user feedback, achieve real-time updates of emotional states and personalized mental health recommendations, ensure that recommendations can be adjusted as user emotions change, and significantly improve the effectiveness of mental health interventions. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0044] Figure 1 This is an overall flow chart of a method for evaluating mental health status based on big data according to an embodiment of the present invention;

[0045] Figure 2 A comparison chart of user suggestion acceptance rates of a mental health status assessment method based on big data according to an embodiment of the present invention;

[0046] Figure 3 This is a graph showing how a sentiment analysis trend curve of a mental health status assessment method based on big data changes over time according to an embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0048] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0049] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0050] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0051] At the same time, in the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the directions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0052] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0053] Example 1

[0054] Reference Figure 1 , which is the first embodiment of the present invention, and provides a method for evaluating mental health status based on big data, comprising:

[0055] S1, obtaining the user's social media text information, voice data and physiological signals, and preprocessing the social media text information, voice data and physiological signals;

[0056] Specifically, we obtain social media text information, use APIs to collect user posted statuses and comments, and record timestamps;

[0057] Specifically, voice data is obtained, and user voice information is collected through mobile devices;

[0058] Specifically, physiological signals are obtained, using wearable devices (such as smart watches) to collect heart rate (HR) and galvanic skin response (GSR);

[0059] Furthermore, for social media text information, HTML tags, punctuation marks and special characters are removed to obtain plain text, and a word segmentation tool is used to segment the plain text, and stop words in the plain text are removed to obtain valid text;

[0060] It should be noted that the cleaning method uses regular expressions. After cleaning, only plain text without HTML tags, punctuation and special characters remains. However, the plain text contains stop words and the texts are connected to each other. It is necessary to convert the plain text into words through text segmentation to form a vocabulary list.

[0061] Specifically, stop words are removed through a stop word library. The purpose is to remove high-frequency words that contribute nothing to sentiment analysis, such as "de" (的), "shi" (是), etc., to obtain a list of effective words, that is, (effective text);

[0062] Furthermore, apply a sentiment dictionary to assign sentiment scores to each word in the effective text and calculate the sentiment intensity TS;

[0063] Preferably, the sentiment dictionary uses AFINN;

[0064] It should be explained that AFINN is a dictionary based on word sentiment. Each word has a corresponding sentiment score, and the score range is from -5 to 5. A positive number indicates positive sentiment, and a negative number indicates negative sentiment;

[0065] Specifically, the assignment of sentiment scores is expressed as:

[0066]

[0067] Among them, E(w i ) represents the sentiment score corresponding to the word w i in the sentiment dictionary, F(w i ) represents the frequency of the word w i appearing in the word list, V(w i ) represents the total frequency of the word w i appearing in the word list; S(w i ) represents the sentiment score of the word w i ; w i represents the i-th word;

[0068] Specifically, by inputting the sentiment scores, the sentiment scores of each word are accumulated to generate the sentiment intensity TS:

[0069]

[0070] Among them, n represents the number of words;

[0071] Furthermore, for voice data, the voice data is frequency-divided into several audio signal segments. Through zero-crossing rate analysis, the characteristic changes of each audio signal segment are determined. According to the characteristic changes, the audio signal segments are divided into several sub-segments. The length of each sub-segment is adjusted according to the natural pauses and emotional changes of the voice to identify the emotional change points;

[0072] It should be explained that zero-crossing analysis is a commonly used feature in speech signal processing, which refers to the number of times a signal passes through a zero value in the time domain (i.e., the number of times a signal crosses from positive to negative or from negative to positive); in the scheme of the present invention, the change in zero-crossing rate can identify the boundary of emotional fluctuations in speech signals; when the zero-crossing rate changes significantly, it means that the emotion or vocalization method in the speech has changed, and these changes are divided into sub-segments for analysis to capture subtle emotional changes. For example, when a speech signal shows a higher zero-crossing rate, it indicates that the frequency of the speech signal is higher and the emotional fluctuations are more intense; on the one hand, the change trend can obtain potential emotional change points, providing a reference basis for pre-processing for the subsequent extraction of pitch and speech rate by Mel-frequency cepstral coefficients, so as to reduce system computing resources and time; on the other hand, for sub-segments where the zero-crossing rate change remains unchanged, the emotional feature analysis can be directly skipped, thereby improving the overall efficiency of the system;

[0073] Furthermore, it is determined whether the emotion change point is an emotion turning point. If it is an emotion turning point, the current sub-segment is marked as an emotion turning point segment, and the pitch P and speech rate R of the audio signal segment to which the sub-segment belongs are extracted through the Mel frequency cepstrum coefficient. Otherwise, no processing is performed and the emotion change point in the next sub-segment is determined.

[0074] Specifically, the Mel frequency cepstral coefficient extracts the pitch P and speech rate R of the audio signal segment to which the subsegment belongs, expressed as:

[0075]

[0076]

[0077] Where A is the number of frames, B is the number of filters, k is the kth cepstral coefficient, that is, the kth component in the Mel frequency cepstral coefficient, I i,j is the jth subsegment of the i-th audio signal segment, l is the index of the frame, I n is the total vocabulary of the audio signal segment, I s is the duration of the audio signal segment;

[0078] Furthermore, the physiological signals include heart rate HR and galvanic skin response GSR. Several time windows are created to segment the physiological signals, and the average value of all time windows is used as the dynamic baseline;

[0079] Specifically, for each time window, the average of heart rate HR and GSR is calculated as the dynamic baseline of the time window:

[0080]

[0081]

[0082] Among them, HRbl,i Expressed as the dynamic baseline bl of the i-th HR, GSR bl,i It is represented as the dynamic baseline bl of the i-th GSR;

[0083] Furthermore, when the dynamic baseline changes, the time point corresponding to the dynamic baseline of the current time window is recorded, and the time points of the dynamic baseline in each time window are connected to form the time window peak. The emotional state is defined as "positive fluctuation" and "negative fluctuation" according to the rising and falling trends of the peak of the time window, and the emotional state evaluation score S is obtained from the "positive fluctuation" and "negative fluctuation";

[0084] Among them, the emotional state assessment score S (HR, GSR);

[0085] Specifically, the peak here refers to the situation where the baseline of the current time window is greater than the baseline of the previous and next adjacent windows;

[0086] Specifically, the peak judgment conditions are as follows:

[0087] HR bl,i >HR bl,i-1 and HR bl,i >HR bl,i+1

[0088] GSR bl,i >GSR bl,i-1 and GSR bl,i >GSR bl,i+1

[0089] Furthermore, whenever a peak is identified, the time point t of the peak is recorded. i ;

[0090] Specifically, a peak rise means that when the dynamic baseline value of the next time window is higher than the dynamic baseline value of the current window, it is defined as an upward trend, that is, "positive fluctuation":

[0091] HR bl,i+1 >HR bl,i or GSR bl,i+1 >GSR bl,i

[0092] Specifically, the peak decline means that when the dynamic baseline value of the next time window is lower than the dynamic baseline value of the current window, it is defined as a downward trend, that is, "negative fluctuation":

[0093] HR bl,i+1 <HR bl,i or GSR bl,i+1 <GSR bl,i

[0094] Specifically, the emotional state assessment score is divided into a positive fluctuation score and a negative fluctuation score, expressed as:

[0095]

[0096]

[0097] Among them, S positive Expressed as the positive volatility score, S negative Expressed as negative volatility score, N positive Indicates the number of positive fluctuations, that is, the number of times the identified peaks show an upward trend within a certain time window, N negative Indicates the number of negative fluctuations, that is, the number of times the identified peaks show a downward trend within a certain time window;

[0098] Specifically, the emotional state assessment score S(HR,GSR) is expressed as:

[0099] S(HR,GSR)=S positive -S negative

[0100] It should be noted that higher scores on the affective state assessment indicate more positive affective states, while lower scores indicate more negative affective states;

[0101] S2, using the pre-processed social media text information, voice data and physiological signals to construct a sentiment vector, divide the sentiment vector into emotions through the K-means clustering algorithm, and update the sentiment vector to obtain the sentiment stability index;

[0102] Furthermore, the sentiment vector is constructed as:

[0103] V=[TS,P,R,S]

[0104] Furthermore, each preprocessed data is constructed into a sentiment vector and stored in a sentiment vector database;

[0105] Furthermore, the emotion vector is divided into K emotion categories, N initial cluster centers are randomly selected from the K emotion categories, and the distance between each emotion vector in the emotion vector database and each cluster center is calculated. In descending order, the cluster center with the smallest distance is assigned to each emotion vector to form M clusters.

[0106] Specifically, the distance between each sentiment vector and the cluster center is calculated, and the sentiment vector is assigned to the cluster center with the smallest distance, resulting in:

[0107]

[0108] Furthermore, for each cluster g, its cluster center is updated to the mean of all sentiment vectors in the cluster;

[0109] Specifically, the updated clustering formula is expressed as:

[0110]

[0111] Among them, c g represents the center of the g-th cluster, x i represents the i-th sentiment vector, N g All sentiment vectors in the g-th cluster;

[0112] Furthermore, check whether the cluster center has changed. If the cluster center has not changed, the algorithm converges and stops iterating. Otherwise, recalculate the distance and assign the clusters, and continue iterating.

[0113] Furthermore, when the cluster center of each cluster g is updated, the internal change of each sentiment vector is calculated to obtain the sentiment stability index ESI g ;

[0114] Specifically, calculate the internal change Δx of each sentiment vector i as follows:

[0115]

[0116] Specifically, the Emotional Stability Index (ESI) g It is expressed as:

[0117]

[0118] Among them, u g represents the mean of the g-th cluster;

[0119] It should be noted that considering the internal changes between cluster centers and sentiment vectors can flexibly capture small fluctuations in sentiment. Smoothing is not limited to capturing extreme emotional fluctuations, but can also analyze the continuous changes in emotions, providing a basis for long-term emotional monitoring;

[0120] S3. Generate personalized mental health advice for users based on the emotional stability index, dynamically adjust the user's emotional state based on the user's acceptance of personalized mental health advice, and re-update the personalized mental health advice;

[0121] Furthermore, personalized mental health advice requires first determining the emotional stability index, and the rules are as follows:

[0122] If ESI g <0.2, the emotion type is “negative”;

[0123] If 0.2≤ESI g <0.5, the sentiment type is “neutral”;

[0124] If ESI g ≥0.5, the sentiment type is “positive”;

[0125] Furthermore, based on the determination of the emotion type, personalized mental health advice is generated for the user;

[0126] Specifically, "negative" is manifested as negative emotions such as anxiety and depression, "neutral" is manifested as calmness, emotional stability and other emotional expressions are not obvious, and "positive" is manifested as positive emotions such as happiness, excitement, enthusiasm, etc., and corresponding personalized mental health suggestions are provided to users according to the emotional type. For example, in the case of "negative", breathing rhythm guidance (such as 4 seconds to inhale and 4 seconds to exhale) and listening companionship are provided; in the case of "positive", users are encouraged to engage in self-improvement or activities with a strong sense of achievement (completing work tasks or learning new skills, etc.), and use the current positive emotional state to promote self-improvement; in the case of "neutral", users are advised to engage in relaxing social activities with friends or recommend users to participate in their favorite activities or hobbies;

[0127] Furthermore, if the user accepts personalized mental health advice, the user's current emotional state is recorded, and the user's feedback results are regularly collected through questionnaires, and the feedback results are quantified, and the standard deviation and mean of the quantified results are calculated to obtain the user's acceptance value;

[0128] Specifically, the regular schedule is once a day, once every three days, once every five days, or once a week;

[0129] Specifically, suppose the user answered N questionnaire questions, and the feedback result of each question is expressed as f i The feedback result F is:

[0130] F=[f1,f2,f3,…,f i ]

[0131] Specifically, the feedback results are quantified to obtain the quantified mean and standard deviation:

[0132]

[0133]

[0134] It should be noted that the standard deviation measures the degree of dispersion of user feedback. If the standard deviation is small, it means that the user's feedback on each question is relatively consistent. If the standard deviation is large, it means that there are large differences in user feedback on different aspects.

[0135] Specifically, the user acceptance value is obtained based on the mean and standard deviation of the user feedback results:

[0136]

[0137] in, It is an adjustment parameter used to balance the influence of mean and standard deviation.

[0138] Furthermore, the user's emotional state is adjusted through the user's acceptance value, and personalized mental health suggestions are updated;

[0139] Specifically, if the Q value is high, it means that the user's acceptance of personalized suggestions is good, that is, the suggestion effect is significant, the user feedback is positive and consistent, and the user's current emotional state is consistent with the previously recorded state; if the Q value is low, it means that the user's acceptance of suggestions is not high, the feedback effect is not ideal, or the feedback results are discrete, and the user's current emotional state does not match the previously recorded state, indicating that the personalized suggestions need to be readjusted.

[0140] Example 2

[0141] Reference Figure 2 and Figure 3 , which is the second embodiment of the present invention, and this embodiment provides a method for evaluating mental health status based on big data, including: in order to verify the effectiveness of the solution of the present invention, a simulation experiment is designed for testing, the simulation experiment uses social media text information, voice data, and physiological signals (including heart rate and skin electrical response) as input data sources, and the data are used to construct an emotion vector, and finally the emotion stability index is calculated through the K-means clustering algorithm to generate personalized mental health advice;

[0142] Ten participants were randomly selected, covering different ages and genders. The participants filled out a mental health questionnaire before the experiment to obtain their basic emotional state. Social media data: The text data of the participants were collected through the social platform API, including the status and comments posted by individuals, with a time range of the last week. Voice data: The voice records of the participants were collected through the microphone of the smartphone, and each voice recording lasted no less than 1 minute. Physiological signals: The heart rate (HR) and galvanic skin response (GSR) of the participants were collected through the smart watch, with a collection frequency of once per second.

[0143] The simulation experiment is divided into two groups: the experimental group uses the method of the present invention, and the control group uses the existing technology (text sentiment analysis technology). Ten random data sets are obtained as shown in Table 1.

[0144] Table 1

[0145]

[0146] As can be seen from Table 1, in the ten random experimental data sets, the emotional stability index (ESI) of the experimental group ranged from 0.18 to 0.62. Users with higher ESI values ​​(A, D, G) showed more stable emotional fluctuations, and these users also had a relatively high acceptance rate for personalized suggestions, up to 90%; while the emotion classification of the control group relied on text data, lacked the support of multimodal data, and the emotion fluctuation detection was not accurate enough; and the suggestion acceptance rate of the control group was significantly lower than that of the experimental group, with the highest being only 65%, indicating that the existing methods could not flexibly capture and timely adjust the emotional suggestions when the emotions fluctuated violently; in addition, the suggestion acceptance rate of the experimental group ranged from 50% to 90%, while the suggestion acceptance rate of the control group was relatively low, ranging from 40% to 65%; the experimental group was able to provide appropriate suggestions in a timely manner when emotions fluctuated by dynamically adjusting the emotional state and personalized suggestions, which significantly improved the user's acceptance of the suggestions; among them, the acceptance rate is the acceptance value in the scheme;

[0147] In addition, through Figure 2 It can be seen that as the number of user samples gradually increases, compared with the prior art, the solution of the present invention can still maintain a relatively high acceptance rate of about 3-5%; and Figure 3 ,Through the sentiment analysis trend curve, we can see that the sentiment ,trend curve remains stable over time, which fully demonstrates the scientific ,nature of the scheme of the present invention and plays an important ,role in regulating the user’s emotions.

[0148] Those skilled in the art will appreciate that the embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the application can adopt the form of complete hardware embodiments, complete software embodiments, or embodiments in combination with software and hardware. Moreover, the application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program codes. The scheme in the embodiments of the present application can be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.

[0149] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0150] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0151] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0152] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0153] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A method for evaluating mental health status based on big data, characterized in that: include: Acquire social media text information, voice data, and physiological signals of the user, and pre-process the social media text information, voice data, and physiological signals; Using the preprocessed social media text information, voice data and physiological signals, construct the emotion vector, divide the emotion vector into emotions through the K-means clustering algorithm, and update the emotion vector to obtain the emotion stability index; Based on the emotional stability index, a personalized mental health suggestion for the user is generated, and according to the user's acceptance of the personalized mental health suggestion, the user's emotional state is dynamically adjusted, and the personalized mental health suggestion is re-updated.

2. The method for evaluating mental health status based on big data according to claim 1, characterized in that: Preprocessing the social media text information includes: Remove HTML tags, punctuation marks and special characters from social media text messages to obtain plain text, use a word segmentation tool to segment the plain text, and remove stop words from the plain text to obtain valid text; The sentiment dictionary is applied to assign a sentiment score to each word in the valid text, and the sentiment strength TS is calculated.

3. The method for evaluating mental health status based on big data according to claim 1, characterized in that: Preprocessing the voice data includes: The speech data is divided into several audio signal segments by frequency division, and the characteristic changes of each audio signal segment are determined by zero crossing rate analysis. The audio signal segment is divided into several sub-segments according to the characteristic changes, and the length of each sub-segment is adjusted according to the natural pauses and emotional changes of the speech, and the emotional change points are identified; Determine whether the emotion change point is an emotion turning point. If it is an emotion turning point, mark the current sub-segment as an emotion turning point segment, and extract the pitch P and speech rate R of the audio signal segment to which the sub-segment belongs through the Mel-frequency cepstral coefficient. Otherwise, do not process it and continue to determine the emotion change point in the next sub-segment.

4. The method for evaluating mental health status based on big data according to claim 1, characterized in that: Preprocessing the physiological signal includes: The physiological signals include heart rate HR and galvanic skin response GSR. Several time windows are created to segment the physiological signals, and the average value of all time windows is used as a dynamic baseline. When the dynamic baseline changes, the time point corresponding to the dynamic baseline of the current time window is recorded, and the time points of the dynamic baseline in each time window are connected to form the time window peak. The emotional state is defined as "positive fluctuation" and "negative fluctuation" according to the rising and falling trends of the peak of the time window, respectively. The emotional state assessment score S is obtained from the "positive fluctuation" and "negative fluctuation".

5. The method for evaluating mental health status based on big data as claimed in claim 2, 3 or 4, characterized in that: Using pre-processed social media text information, voice data, and physiological signals, we construct sentiment vectors, including: The sentiment vector is expressed as: V=[TS,P,R,S] Each time the preprocessed data is constructed into a sentiment vector and saved in the sentiment vector database; Among them, the emotional state assessment score S (HR, GSR).

6. The method for evaluating mental health status based on big data according to claim 5, characterized in that: The emotion vector is divided into emotion classes by using a K-means clustering algorithm, and the emotion vector is updated at the same time, including: The emotion vector is divided into K emotion categories, and N initial cluster centers are randomly selected from the K emotion categories. The distance between each emotion vector in the emotion vector database and each cluster center is calculated, and the cluster center with the smallest distance is assigned to each emotion vector in descending order to form M clusters. For each cluster g, update its cluster center to the mean of all sentiment vectors in the cluster.

7. The method for evaluating mental health status based on big data according to claim 6, characterized in that: Also includes: Check whether the cluster center changes. If the cluster center does not change, the algorithm converges and stops iterating. Otherwise, recalculate the distances and assign clusters, and continue iterating.

8. The method for evaluating mental health status based on big data according to claim 6, characterized in that: Get an emotional stability index, including: When the cluster center of each cluster g is updated, the internal change of each sentiment vector is calculated to obtain the sentiment stability index ESI g .

9. The method for evaluating mental health status based on big data according to claim 8, characterized in that: Based on the emotional stability index, a personalized mental health recommendation for the user is generated, including: If ESI g <0.2, the sentiment type is "negative"; If 0.2≤ESI g <0.5, the sentiment type is "neutral"; If ESI g ≥0.5, the sentiment type is "positive"; Generate personalized mental health advice for users based on the determination of emotion type.

10. The method for evaluating mental health status based on big data according to claim 9, characterized in that: Dynamically adjusting the user's emotional state based on the user's acceptance of personalized mental health advice, and re-updating the personalized mental health advice, including: If the user accepts the personalized mental health advice, the user's current emotional state is recorded, and the user's feedback results are regularly collected through questionnaires, and the feedback results are quantified, and the standard deviation and mean of the quantified results are calculated to obtain the user's acceptance value; The user's emotional state is adjusted according to the user acceptance value, and the personalized mental health advice is updated.

Images