Multi-dimensional data-based alony risk prediction method, apparatus and device, and medium

Abstract

The invention relates to the technical field of data processing, and discloses a multi-dimensional data-based lonely risk prediction method, device and equipment and a medium, and the scheme provides accurate and comprehensive data support for social risk assessment by collecting multi-dimensional social behavior data of a target object. By calculating the social network features, a quantitative basis is provided for lonely risk assessment. And analyzing the voice data through an emotion analysis model, and capturing voice emotion features of the target object in social interaction. According to the method, social network features and voice emotion features are comprehensively analyzed based on an autism risk assessment model, an autism risk prediction value is generated, the possibility that a target object is in an autism state is visually reflected, an autism intervention strategy is formulated according to an autism risk level, intervention can be effectively performed according to the actual autism risk condition of the target object, and the risk assessment efficiency is improved. And the intervention accuracy and timeliness of the social health of the target object in the home-based care service in the medical field are obviously improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to a loneliness risk prediction method and device based on multi-dimensional data, equipment and medium. BACKGROUND

[0002] With the acceleration of population aging, more and more old people choose to live at home. Although the popularization of intelligent health devices and online service platforms applied in the medical field improves the convenience of the life of the old people to a certain extent, the problem of social loneliness is still prominent. A large number of studies have shown that social isolation and loneliness not only threaten the mental health of the old people, but also are closely related to the occurrence of depression, cognitive impairment and even chronic diseases. Therefore, how to scientifically evaluate the social status of the old people and carry out effective intervention has become an important issue to be solved in the field of smart care for the old.

[0003] At present, there are some products and services on the market that attempt to alleviate the loneliness of the old people, such as intelligent accompanying robots, voice assistants, community activity APPs and regular telephone consolation, etc. These products have a certain positive effect on improving the social participation of the old people, but there are still many deficiencies in practical application.

[0004] Firstly, most of the products are based on a single data source (such as call records, APP use frequency or activity check-in), which is difficult to capture and analyze the real social network structure and interaction quality of the old people in all aspects. Secondly, the identification of loneliness risk mainly relies on regular questionnaires or manual reporting, which lacks active and continuous monitoring of daily behavior and emotional dynamics, and is easy to miss high-risk individuals. Thirdly, the intervention methods are relatively single and passive, mainly relying on regular reminders and activity recommendations, and lack of intelligent and personalized intervention schemes based on individual social characteristics and interest preferences. Finally, the multi-source data such as health, communication and activities are often scattered in different platforms, which cannot be effectively integrated and utilized, affecting the accuracy of risk assessment and the pertinence of intervention measures.

[0005] Overall, the field of social health management of the old people still has obvious shortcomings in data collection dimension, risk identification initiative, intervention intelligence and data fusion, which is difficult to meet the increasingly diversified and personalized needs of pension services. Therefore, how to improve the intervention accuracy of the social health of the old people living at home has become a technical problem to be solved. SUMMARY

[0006] The present application provides a loneliness risk prediction method and device based on multi-dimensional data, equipment and medium, to solve the technical problem of low accuracy of social risk assessment and pertinence of intervention measures of the old people, and improve the intervention accuracy of the social health of the old people living at home.

[0007] In a first aspect, a loneliness risk prediction method based on multi-dimensional data is provided, comprising:

[0008] obtaining multi-dimensional social behavior data of a target object;

[0009] calculating social network features of the target object based on the multi-dimensional social behavior data;

[0010] analyzing sentence sentiment in voice data of the target object based on an emotion analysis model to obtain voice emotion features;

[0011] performing feature analysis on the social network features and the voice emotion features based on a loneliness risk assessment model to generate a loneliness risk prediction value of the target object;

[0012] generating a loneliness intervention strategy for the target object based on the loneliness risk prediction value.

[0013] In a second aspect, a loneliness risk prediction device based on multi-dimensional data is provided, comprising:

[0014] a multi-dimensional data acquisition module configured to obtain multi-dimensional social behavior data of a target object;

[0015] a social network feature calculation module configured to calculate social network features of the target object based on the multi-dimensional social behavior data;

[0016] a voice emotion feature analysis module configured to analyze sentence sentiment in voice data of the target object based on an emotion analysis model to obtain voice emotion features;

[0017] a loneliness risk assessment module configured to perform feature analysis on the social network features and the voice emotion features based on a loneliness risk assessment model to generate a loneliness risk prediction value of the target object;

[0018] a loneliness intervention policy generation module configured to generate a loneliness intervention strategy for the target object based on the loneliness risk prediction value.

[0019] In a third aspect, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the loneliness risk prediction method based on multi-dimensional data when executing the computer program.

[0020] In a fourth aspect, a computer readable storage medium is provided, which stores a computer program, wherein the computer program is executable by a processor to implement the steps of the loneliness risk prediction method based on multi-dimensional data.

[0021] The scheme realized by the loneliness risk prediction method and device based on multi-dimensional data, the computer device, and the storage medium collects multi-dimensional social behavior data of the target object, enriches the data dimension, more comprehensively reflects the social state of the target object, avoids the wrong evaluation of the social risk of the target object due to one-sided data, and provides accurate and comprehensive data support for social risk evaluation. The multi-dimensional social behavior data of the target object is converted into quantitative indicators with analysis value by calculating the social network characteristics of the target object, which can more accurately identify the position and state of the target object in the social network and provide key quantitative basis for subsequent loneliness risk evaluation. The voice data is analyzed by the sentiment analysis model, which can capture the emotional state of the target object in social interaction and increase the information of the emotional dimension for social risk evaluation. The loneliness risk prediction value can be generated by comprehensively analyzing the social network characteristics and voice emotion characteristics based on the loneliness risk evaluation model, which directly reflects the possibility of the target object falling into a lonely state, provides a key decision basis for formulating a loneliness intervention strategy, and formulates a personalized loneliness intervention strategy according to different loneliness risk levels, which can ensure the pertinence and effectiveness of the intervention measures, realize early warning and timely intervention, and significantly improve the intervention accuracy and timeliness of the social health of the elderly in home-based care. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the description of the embodiments of the present application. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0023] Figure 1 is an application environment schematic diagram of the loneliness risk prediction method based on multi-dimensional data in an embodiment of the present application;

[0024] Figure 2 is a flowchart of the first embodiment of the loneliness risk prediction method based on multi-dimensional data provided by the embodiments of the present application;

[0025] Figure 3 is a social network relationship schematic diagram constructed by taking the elderly at home as the target object provided by the embodiments of the present application;

[0026] Figure 4 is a structure schematic diagram of the loneliness risk prediction device based on multi-dimensional data in an embodiment of the present application;

[0027] Figure 5 is a structure schematic diagram of the computer device in an embodiment of the present application;

[0028] Figure 6Fig. 2 is another structural schematic diagram of a computer device in an embodiment of the present application. DETAILED DESCRIPTION

[0029] The technical solutions in the embodiments of the present application will be clearly and completely described in connection with the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0030] The method for predicting loneliness risk based on multi-dimensional data provided by the embodiments of the present application can be applied in an application environment as shown in Fig. 1. Figure 1 In the application environment shown in Fig. 1, a client communicates with a server through a network. After receiving a request for generating a loneliness intervention strategy of a target object initiated by the client, the server can acquire multi-dimensional social behavior data of the target object; based on the multi-dimensional social behavior data, the server can calculate social network features of the target object; based on an emotion analysis model, the server can analyze sentence emotions in voice data of the target object to obtain voice emotion features; based on a loneliness risk assessment model, the server can analyze the social network features and the voice emotion features to generate a loneliness risk prediction value of the target object; and based on the loneliness risk prediction value, the server can generate a loneliness intervention strategy of the target object.

[0031] In the present application, for the technical problem of low accuracy of social risk assessment of the elderly and low pertinence of intervention measures in the medical field, the multi-dimensional social behavior data of the target object is collected, the data dimension is enriched, and the social state of the target object is more comprehensively reflected, so as to avoid incorrect assessment of the social risk of the target object due to one-sided data and to provide accurate and comprehensive data support for social risk assessment. By calculating the social network features of the target object, the multi-dimensional social behavior data of the target object is converted into quantitative indicators with analysis value, so as to more accurately identify the position and state of the target object in the social network and to provide a key quantitative basis for subsequent loneliness risk assessment. By analyzing the voice data through the emotion analysis model, the emotional state of the target object in social interaction can be captured, and the information of the emotional dimension is added to the social risk assessment. Based on the loneliness risk assessment model, the social network features and the voice emotion features are comprehensively analyzed, a reliable loneliness risk prediction value is generated, the possibility of the target object falling into a lonely state is intuitively reflected, a key decision basis for formulating a loneliness intervention strategy is provided, personalized loneliness intervention strategies are formulated according to different loneliness risk levels, the pertinence and effectiveness of the intervention measures are ensured, early warning and timely intervention are realized, and the intervention accuracy and timeliness of the social health of the elderly in home-based care are significantly improved.

[0032] The client can be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers and portable wearable devices. The server can be implemented by an independent server or a server cluster composed of multiple servers.

[0033] The application will be described in detail below through specific embodiments.

[0034] Please refer to Figure 2 as shown, Figure 2 The flowchart of the first embodiment of the method for predicting the lonely risk based on multi-dimensional data provided by the embodiments of the application includes the following steps:

[0035] S101: Obtain multi-dimensional social behavior data of a target object;

[0036] In an embodiment, the multi-dimensional social behavior data refers to various data about the social behavior of the target object (such as an old person) collected from multiple dimensions (such as different social platforms, different social relationships, different interaction modes, etc.), including interaction frequency, duration, content, object and other information, reflecting the multi-aspect performance and characteristics of the target object in social activities.

[0037] Specifically, the multi-dimensional social behavior data can include but is not limited to detailed information about the call records, meeting times, message exchanges, joint activity participation and other aspects between different social objects (such as family members, friends, service personnel, community activities, etc.).

[0038] To obtain the multi-dimensional social behavior data of the target object, at least one social data management platform can be connected, and the multi-dimensional social behavior data of the target object can be collected through the social data management platform.

[0039] Further, at least one social data management platform is connected; based on the identity of the target object, the social behavior data of at least one data type corresponding to the target object in each social data management platform is collected, and the multi-dimensional social behavior data of the target object is obtained.

[0040] In an embodiment, the social data management platform refers to those platforms or systems that can store, manage and provide social data, such as the service platform of communication software, the activity management platform, the data management system of intelligent voice terminal, etc., which provides data sources and interfaces for collecting and integrating social data.

[0041] In an embodiment, the social communication software (such as WeChat, QQ, various mobile communication applications, etc.), activity management platform (community activity registration platform, interest group organization platform, etc.), intelligent voice terminal (smart speaker, voice assistant device, etc.) and the like can be automatically collected and standardized various types of social data. By formulating a unified data format and specification, data from different platforms and channels are converted and integrated to ensure data consistency and comparability, laying a foundation for subsequent analysis.

[0042] Specifically, the social behavior data of the target object can be collected in various social data management platforms based on the identity of the target object, such as name, account number, phone number, etc.

[0043] Among them, the social object refers to various individuals or groups that have social interactions with the target object, including family members (such as children, grandchildren, etc.), friends, neighbors, community service personnel, etc., as well as various social activities that the target object participates in, etc.

[0044] First, the collected various types of social behavior data can be data cleaned. All types of social behavior data are comprehensively screened to filter out invalid data, such as duplicate records, incorrect information, and obviously abnormal data points (such as negative call duration, etc.), to improve data quality and accuracy and ensure the reliability of subsequent analysis.

[0045] Second, the types of interactive objects in the social behavior data are labeled. The interactive objects in the data are labeled in detail to determine their types, including family members (children, grandchildren, etc.), friends (old friends, newly acquainted friends, etc.), service personnel (community service personnel, nursing personnel, etc.), activities (various community-organized activities, interest group activities, etc.), to facilitate understanding of the impact of different types of social interactions on the psychological state of the elderly.

[0046] Finally, the social behavior data is feature extracted from the cleaned data to extract multiple key features, including frequency, which reflects the number of social interactions within a certain time, such as the number of calls per week; duration, which represents the duration of each interaction, such as the duration of a meeting; distribution, which represents the dispersion of social activities in different times, different scenarios, etc., such as more social activities on weekdays or weekends, at home or in community public places, etc.; and emotional score, which is obtained by evaluating the text and voice content in the interaction through emotion analysis technology, used to measure the emotional tendency contained in the social interaction, such as positive, negative or neutral.

[0047] S102: Based on the multi-dimensional social behavior data, the social network features of the target object are calculated;

[0048] In an embodiment, the social network features can include, but are not limited to, sociality, interaction intensity, social diversity, interaction duration distribution, sentiment tendency, etc.

[0049] The sociality is used to measure the connection breadth of the target object in the social network, i.e., the number of associations with other objects. Specifically, the total number of connections between the target object and all other objects with interaction records is counted. For example, if the target object has interactions with family members, friends, service personnel, community activities, etc., each category of interaction object corresponds to a connection, and the sociality is equal to the sum of the number of these connections. For example, if the target object has interactions with 3 family members, 5 friends, 2 service personnel, and participates in 4 community activities, then the sociality is 3+5+2+4=14.

[0050] The interaction intensity is used to reflect the overall activity level of the target object's social interaction, which is a comprehensive reflection of interaction frequency and interaction duration, etc. Specifically, the interaction frequency of the target object with different categories of objects is multiplied by the corresponding interaction duration, then all category results are added, and finally divided by the total number of interaction categories to obtain the average interaction intensity.

[0051] For example, assuming that the interaction frequency with family members is f f , the interaction duration is d f , the interaction frequency with friends is f r , the interaction duration is d r , the interaction frequency with service personnel is f s , the interaction duration is d s , the interaction frequency with community activities is f a , and the interaction duration is d a , then the interaction intensity I is calculated as follows:

[0052]

[0053] The social diversity is used to reflect the distribution of the target object's social activities among different types of objects, i.e., whether its social life is rich and diverse. Specifically, the proportion of the target object's interaction records with each category of object is calculated, and then the Shannon entropy formula is used to quantify the diversity:

[0054]

[0055] where n represents the number of categories of social objects (such as family members, friends, etc.), p i is the proportion of the target object's interaction records with the i-th category of object, i.e., the ratio of the interaction frequency of this category to the total interaction frequency.

[0056] The interaction duration distribution is used to show the proportion of time spent by the target object in different categories of social interaction, which helps to understand the time allocation tendency of the target object. Specifically, the proportion of interaction duration of the target object with each category of object to the total interaction duration is calculated respectively.

[0057] For example, assuming that the total interaction duration is D, the interaction duration with the i-th category of object is D i , then the proportion of interaction duration of this category P i is:

[0058]

[0059] The sentiment tendency is used to evaluate the overall state of the emotions conveyed by the target object in social interaction, whether it is positive, negative or neutral. Specifically, the sentiment scores of the target object interacting with each category of object are summed up respectively, and then divided by the number of interactions of each category to obtain the average sentiment score of each category of interaction. Finally, the mean value of the average sentiment scores of all categories is calculated.

[0060] For example, assuming that the total sentiment score of interaction with family members is E f , the number of interactions is C f , the total sentiment score of interaction with friends is E r , the number of interactions is C r , the total sentiment score of interaction with service personnel is E s , the number of interactions is C s , the total sentiment score of interaction with community activities is E a , the number of interactions is C a , then the sentiment tendency S is calculated as:

[0061]

[0062] Further, based on the multi-dimensional social behavior data, at least one social object of the target object is extracted, and a social network between the target object and each social object is constructed. Based on the social network, the social data of the target object and each social object is determined, wherein the social data includes social relationship and interaction frequency; based on the social data of the target object and each social object, the social network characteristics of the target object are calculated.

[0063] In an embodiment, the social network is a network structure composed of nodes (representing social objects) and edges (representing interaction relationships between social objects), which is used to describe and analyze the connection and interaction mode between different social objects.

[0064] For example, as Figure 3The social network relationship diagram shows that the homebound elderly are the research object, based on their multi-source data, including communication records, activity participation, smart device interaction information, etc. A dynamic social network relationship diagram is constructed. The diagram takes the elderly node as the center, and establishes connections with daughters, old friends, neighbors, community activities, etc. (such as various contacts or social entities).

[0065] For example, by analyzing the call records of the elderly and their daughters, the number of meetings with old friends, group chat interactions with neighbors, and the frequency of participating in community activities, etc. Data determines the corresponding connection relationship, constructs a social network, and intuitively reflects the actual social relationship and interaction frequency. Each line in the diagram represents a type of social interaction, and the number of interactions within a week is annotated next to the line. The nodes distinguish different types of social objects (such as family members, friends, neighbors, and service resources).

[0066] In the constructed social network, the social data of the target object and each social object is determined, including social relationship and interaction frequency. Among them, the social relationship refers to the association between two nodes in the social network, including relationship type (such as kinship, friendship, neighborhood relationship, etc.), relationship strength (such as closeness, interaction frequency, etc.) and other attributes, reflecting the interaction nature and closeness between social objects; the interaction frequency refers to the number of interactions between the elderly and a social object within a certain time range, which is used to measure the activity level of their interaction.

[0067] As Figure 3 shown in the social network relationship diagram, the social relationship refers to the closeness and interaction mode between the elderly and various objects, such as the close family relationship between the elderly and their daughters, the long-term friendship relationship with old friends, the neighborhood relationship with neighbors, and the relationship between participants and activities in community activities. The interaction frequency quantifies the number of interactions between the elderly and various objects with specific numerical values, such as the elderly and their daughters talking 12 times a week, meeting old friends 3 times a week, having 2 group chat interactions with neighbors a week, and participating in community activities once a week.

[0068] Through this graph modeling method, the social activity level, diversity and structural characteristics of the elderly can be comprehensively and dynamically quantified, providing a scientific basis for subsequent loneliness risk assessment and intervention decision-making. In addition, the social network relationship diagram can also be used to track the changing trend of the elderly's social circle, assist in discovering potential social isolation risks, and thus achieve more accurate care and service push.

[0069] Specifically, as Figure 3The node features shown can present the sociality, activity, and interest diversity of the target object. The sociality can measure the number of direct connections of the target object with other objects, reflecting the activity and social range of the target object in the social network; the activity can be calculated based on the interaction frequency and duration, reflecting the participation degree of the target object in social activities; the interest diversity can be measured by analyzing the types of activities participated in by the target object, the topics concerned, and the like, measuring the extent of the interest field of the target object.

[0070] As shown in the social network relationship diagram, Figure 3 The edge features shown in the social network relationship diagram can present the interaction intensity and emotional tendency of the target object. The interaction intensity can be calculated by comprehensively considering the interaction frequency, duration, and the like, and is used to reflect the closeness of the interaction between the target object and other specific objects. The emotional tendency can analyze the emotional color of the interaction content, and evaluate whether the emotional state of the target object in the interaction with others is positive, negative, or neutral.

[0071] Through the construction of the social network and the extraction and calculation of the social network features, the social state of the old people can be comprehensively and quantitatively understood, providing a scientific basis for subsequent loneliness risk assessment and intervention decision-making. At the same time, the social network relationship diagram can also be used to track the change trend of the social circle of the old people, assist in discovering potential social isolation risks, and thus realize more accurate care and service pushing.

[0072] S103: Based on an emotional analysis model, analyze the sentence emotion in the voice data of the target object, and obtain voice emotion features;

[0073] In order to accurately capture the emotional state of the target object (such as an old person at home), the present scheme uses advanced voice analysis technology to deeply analyze the emotional information in the voice data of the target object, and then obtains voice emotion features that can reflect the emotional state.

[0074] Generally, the emotional analysis model is an algorithm model that uses natural language processing, speech signal processing, and machine learning technology to automatically identify and classify the emotional tendency in text or voice data. In the field of voice, it analyzes the acoustic features (such as pitch, speech rate, timbre, etc.) and language content (analyzes the vocabulary and semantics after converting the voice into text) of the voice to determine whether the emotion expressed by the speaker is positive, negative, or neutral, and gives the corresponding confidence score. These models are usually trained based on a large amount of voice data labeled with emotional tendency, learning the performance patterns of different types of emotions on voice features, so as to be able to accurately analyze the emotions of new voice data.

[0075] Further, based on the multi-dimensional social behavior data of the target object, voice text data of the target object is extracted; based on the sentiment analysis model, sentiment analysis is performed on each piece of voice text in the voice text data to obtain a sentiment score of each piece of voice text; and based on the sentiment scores of each piece of voice text, voice emotion features of the target object are determined.

[0076] In an embodiment, voice text data is accurately extracted from multi-dimensional social behavior data of the target object. These data are derived from the target object's voice communication in various social scenarios, such as call records with family and friends, voice interaction during community activities, and instructions and conversations when using a smart voice assistant, etc. The comprehensiveness and representativeness of the extracted data are ensured to accurately reflect the emotional state of the target object in different social situations.

[0077] The collected voice data is preprocessed, including removing background noise, standardizing volume, unifying sampling rate, etc., to improve data quality and ensure the accuracy of subsequent sentiment analysis. For example, a filtering algorithm is used to remove environmental noise interference, making the voice signal clearer and purer.

[0078] The preprocessed voice data is input into each piece of voice text data, which is sequentially input into a sentiment analysis model that has been professionally trained. Based on advanced natural language processing techniques and machine learning algorithms, the model can automatically identify the emotional features in the voice, deeply understand the semantic information and emotional expression in the voice text.

[0079] The sentiment analysis model can extract acoustic features (such as pitch, timbre, speech rate, volume, etc.) and language content features (by converting voice to text through voice recognition technology, analyzing emotional words, semantic expressions, etc.) in the voice data, comprehensively judge the emotional tendency expressed by each piece of voice, and output the sentiment score of each piece of voice text, usually represented by a value between 0 and 1, with a higher score indicating a more positive emotion. For example, a piece of voice text expressing happiness and satisfaction may have a sentiment score of 0.89, while a piece of text revealing sadness or depression may have a score of 0.32.

[0080] In an embodiment, the output of the sentiment analysis model can also be a confidence score. For example, a piece of voice may be determined to have a positive emotion with a confidence of 0.85, or a negative emotion with a confidence of 0.7, etc.

[0081] According to the sentiment score of each piece of voice text, multiple key indicators are calculated to determine the voice emotion features of the target object, construct a voice emotion feature vector, and comprehensively quantify the voice emotion state of the target object.

[0082] Exemplarily, the voice emotion features can include, but are not limited to, an average sentiment score, an emotion fluctuation range, a positive emotion proportion, a negative emotion proportion, an emotion extreme value occurrence frequency, an emotion fluctuation intensity, a dominant emotion type, and an emotion stability, etc.

[0083] Specifically, the average sentiment score can be an arithmetic mean of all voice text sentiment scores, reflecting the overall tendency of the target object's emotion in a statistical period; the positive emotion proportion refers to the proportion of the number of positive emotion voice segments to the total number of voice segments; the negative emotion proportion refers to the proportion of the number of negative emotion voice segments to the total number of voice segments; the emotion fluctuation range refers to the difference between the maximum value and the minimum value of the sentiment score, embodying the stability of the target object's emotion; the emotion extreme value occurrence frequency refers to the number of times that the voice text with a sentiment score close to 1 (such as greater than 0.9) or close to 0 (such as less than 0.1) occurs, and the proportion thereof in the total text is counted, so as to understand the occurrence frequency of the target object's extreme emotion; the emotion fluctuation intensity refers to the degree of change in the sentiment tendency between adjacent voice segments, which is measured by calculating the average value of the absolute value of the difference between the sentiment scores of adjacent segments; the dominant emotion type refers to the emotion type with the highest proportion in the emotion analysis result, such as positive, negative or neutral; and the emotion stability refers to the dispersion degree of the sentiment tendency in the entire voice data, such as a smaller variance indicating a more stable emotion and a larger variance indicating a more variable emotion.

[0084] These voice emotion features can quantitatively evaluate the emotional state of the target object from multiple dimensions, providing strong support for the emotional evaluation, psychological care, and loneliness risk intervention of the target object, helping to timely detect abnormal emotional fluctuations and take corresponding measures in advance.

[0085] For example, after analysis and calculation, the voice emotion features of the old person on a certain day are: the positive emotion proportion is 0.6, the negative emotion proportion is 0.3, the emotion fluctuation intensity is 0.45, the dominant emotion type is positive, and the emotion stability is high (the variance is 0.03). These voice emotion features can provide strong basis for subsequent emotional state evaluation, loneliness risk analysis, and targeted care intervention, helping to timely discover potential emotional problems and loneliness risks of the target object, and realizing precise care.

[0086] S104: performing feature analysis on the social network features and the voice emotion features based on a loneliness risk assessment model, to generate a loneliness risk prediction value of the target object;

[0087] The loneliness risk assessment model is a data-driven prediction model designed to predict the likelihood of an individual falling into a state of loneliness by analyzing multi-dimensional features related to loneliness, such as social network features and voice emotion features. This model is typically trained based on historical data using machine learning algorithms and can output a loneliness risk prediction value for each target object based on input feature data, helping relevant personnel identify potential loneliness risks and take targeted prevention and intervention measures.

[0088] The loneliness risk assessment model can be trained using historical data.

[0089] Specifically, a large amount of historical data is collected, including social network features (such as sociality, interaction intensity, social diversity, etc.) and voice emotion features (such as average emotional score, emotional fluctuation range, etc.) of target objects or multiple different evaluation objects, as well as corresponding labels of whether they are lonely (such as loneliness state information obtained through psychological assessment questionnaires, etc.). The data is preprocessed, such as cleaning and normalization, to ensure data quality.

[0090] Highly relevant features related to loneliness risk are selected from social network features and voice emotion features for integration to build a more comprehensive social feature vector as input features for the model. For example, through correlation analysis, feature importance evaluation, etc., it is determined that sociality, interaction intensity, average emotional score, and emotional fluctuation range have strong predictive ability for loneliness risk.

[0091] Select appropriate machine learning algorithms (such as logistic regression, random forest, support vector machine, etc.) to build a loneliness risk assessment model. Divide the preprocessed data set into training set and test set, use the training set to train the model, adjust the hyperparameters of the model to optimize the performance of the model, so that it can accurately predict the loneliness risk on the training set. Use the test set to validate the trained model and evaluate its accuracy, recall rate, F1 value, etc. According to the validation results, further optimize the model, such as adjusting feature engineering, selecting different algorithms, using ensemble learning, etc., to improve the prediction accuracy and generalization ability of the model.

[0092] The pre-trained loneliness risk assessment model is used to analyze the social network features and voice emotion features of the target object to predict the loneliness risk of the target object.

[0093] Further, based on the social network features and the voice emotion features, a social feature vector of the target object is constructed; based on the loneliness risk assessment model, the social feature vector is predicted to output the loneliness risk prediction value.

[0094] The social feature vector is a multidimensional vector that comprehensively represents the social status and emotional state of the target object. It quantifies and integrates social network features and voice emotion features, and can fully reflect various attributes and behavior patterns of the target object in social and emotional aspects. By constructing the social feature vector, rich input information can be provided for the loneliness risk assessment model, improving the prediction accuracy and reliability of the model.

[0095] Specifically, the social network features (such as social degree, interaction intensity, social diversity, interaction duration distribution, etc.) and voice emotion features (such as average emotional score, emotional fluctuation range, positive emotion proportion, negative emotion proportion, etc.) of the target object are integrated to construct a comprehensive social feature vector.

[0096] The constructed social feature vector is input into the loneliness risk assessment model. The loneliness risk assessment model will comprehensively analyze the social feature vector and evaluate the contribution of each social feature to the loneliness risk. For example, low social degree, weak interaction intensity, low average emotional score, and large emotional fluctuation range may increase the risk of loneliness.

[0097] The loneliness risk assessment model generates a loneliness risk prediction value for the target object based on the feature analysis results through internal mathematical operations and logical judgments. This loneliness risk prediction value can be a probability value between 0 and 1, indicating the likelihood of the target object being in a state of loneliness. For example, a prediction value of 0.8 indicates an 80% likelihood of the target object being in a state of loneliness. Based on this prediction value, appropriate intervention measures can be taken, such as providing social activity recommendations, psychological counseling, etc., to reduce the risk of loneliness of the target object.

[0098] S105: Based on the loneliness risk prediction value, generate a loneliness intervention strategy for the target object.

[0099] In an embodiment, based on the evaluated loneliness risk prediction value, combined with the interests and historical records of the elderly, appropriate intervention measures are recommended to construct a loneliness intervention strategy for the target object.

[0100] Further, based on the comparison result of the preset loneliness risk threshold and the loneliness risk prediction value, the loneliness risk level of the target object is determined; based on the loneliness risk level of the target object, the loneliness intervention strategy for the target object is determined.

[0101] In an embodiment, one or more loneliness risk thresholds can be set, such as a medium risk threshold and a high risk threshold. The loneliness risk prediction value of the target object is compared with the preset threshold to determine its loneliness risk level.

[0102] For example, two isolation risk thresholds are set, a medium risk threshold (e.g., 0.5) and a high risk threshold (e.g., 0.7). The isolation risk prediction value of the target object is compared with the preset thresholds to determine its isolation risk level.

[0103] If the prediction value is less than the medium risk threshold (0.5), it is determined that the target object is in a low isolation risk state, and its social and emotional state is relatively good. If the prediction value is between the medium risk threshold (0.5) and the high risk threshold (0.7), it is determined that the target object is in a medium isolation risk state, and there may be a certain tendency to be lonely, which needs to be concerned and appropriate intervention measures are needed. If the prediction value is greater than or equal to the high risk threshold (0.7), it is determined that the target object is in a high isolation risk state, and is likely to be experiencing a more serious sense of loneliness, and urgent and effective intervention measures are needed.

[0104] Further, historical intervention data and historical interest data of the target object are collected; at least one intervention measure is determined based on the historical intervention data and the historical interest data; and an isolation intervention strategy for the target object is determined based on the isolation risk level of the target object and the at least one intervention measure.

[0105] In an embodiment, detailed information of all intervention measures that the target object has accepted in the past is collected from relevant databases or record systems. This includes the type of intervention measures (such as psychological counseling, community activity invitation, volunteer accompanying, etc.), implementation time, implementation frequency, duration, and effect evaluation results after each intervention (such as target object feedback, emotional improvement degree, social behavior change, etc.).

[0106] The historical interest and hobby information of the target object is sorted out, which may come from the self-report of the target object, the information provided by the family members, and the record of past participation in activities, etc. The degree of interest of the target object in various activities, topics, hobbies, etc. in the past is understood, such as once liked painting, interested in historical stories, and keen on gardening, etc.

[0107] Based on the collected historical intervention data and historical interest data, intervention measures suitable for the target object are preliminarily screened out. For the types of measures with good effects in the historical intervention data, they can be considered for priority inclusion; at the same time, combined with the historical interests and hobbies of the target object, activities or services related to them are selected as potential intervention measures. For example, according to the condition that the target object has positive feedback after participating in the community art performance and has been interested in painting, community art workshops (including painting courses) and cultural lectures (historical theme) are preliminarily selected as candidate intervention measures.

[0108] The initial intervention measures can be further refined and expanded to consider different forms and content of implementation to increase the richness and targeting of the intervention. For example, for the community art workshop intervention, it can be refined into a weekly painting group activity guided by a professional art teacher, with a duration of two hours per activity; at the same time, online painting work sharing and exchange activities can be expanded to facilitate target objects to exchange experiences with other painting enthusiasts during their spare time.

[0109] According to the interests, health status and social needs of the elderly, suitable activities, psychological counseling, volunteer companionship and other resources are automatically matched and pushed, supporting multi-channel push such as smart speakers, Apps, and mini programs. Monitor behavior changes after intervention, continuously optimize risk models and intervention strategies.

[0110] For example, if the elderly have difficulties in daily life care, such as difficulty in moving around, inability to take care of themselves, etc., they can be provided with home nursing, housekeeping services, etc. to help them solve practical problems in life, improve their quality of life, and reduce feelings of loneliness and helplessness due to life difficulties. Some elderly people may be lonely due to lack of social skills or poor communication skills, and can be provided with social skills training courses to teach them some basic social skills and communication methods to enhance their social confidence and help them better integrate into social circles. Alternatively, community activity centers, senior activity rooms, etc. can be set up to provide entertainment facilities such as chess, table tennis, and table tennis to facilitate leisure and entertainment and social activities for the elderly within the community; community mutual aid organizations can be established to encourage community residents to care for and look after the elderly around them, creating a good atmosphere of neighborhood mutual assistance.

[0111] According to the target object's risk level of loneliness, combined with the target object's historical intervention data (including intervention measures and feedback effects), at least one intervention measure is determined to build a loneliness intervention strategy.

[0112] Specifically, if the elderly have a low risk of loneliness, such as a risk probability of less than 0.7, regular care reminders can be set for their family members to encourage them to communicate with the elderly in a timely manner, understand their living and psychological conditions, and provide them with more family care to prevent the occurrence of loneliness.

[0113] At the same time, on the basis of maintaining the existing regular family care reminders and interest group recommendations, measures related to the target object's interests determined by the intervention measures can be appropriately supplemented. Online resource recommendations based on their historical interests, such as sending painting tutorial video links, historical story audiobooks, etc. can also be added to further enrich their lives and prevent the risk from rising.

[0114] For example, for target objects who like painting and historical stories, in addition to recommending community painting workshops, high-quality painting tutorial videos and historical story audio book resources are regularly pushed, at least twice a week, and after each push, the target object is reminded to view through the intelligent device reminder function.

[0115] For middle-risk elderly people, such as risk probability between 0.7 and 0.85, in addition to retaining regular volunteer visits and community activity invitations and other original measures, measures that are historically effective and in line with the interests of the target object can be implemented more intensively or adjusted in detail.

[0116] For example, the elderly can be invited to participate in various activities organized by the community, such as cultural performances, health lectures, interest groups, etc., to increase their opportunities for interaction with the outside world and enrich their daily lives. In addition, suitable interest groups or clubs can be recommended for the elderly based on their interests, such as if the target object has previously participated in cultural lecture activities and has given positive feedback and is interested in historical themes, the frequency of cultural lectures can be increased from once a month to once every two weeks, and the target object is reminded one week in advance through both phone and text messages to ensure their awareness and participation. At the same time, online communication activities are combined to encourage the target object to participate in online discussions on related topics after the lecture, share their insights and feelings, and enhance their social interaction.

[0117] When the elderly are assessed to be at high risk of loneliness, such as a risk probability greater than 0.85, in addition to continuing emergency and in-depth emotional intervention measures such as psychological counseling and one-on-one volunteer companionship, historical intervention data and interest data should be fully utilized to strengthen and integrate intervention measures.

[0118] For example, a professional psychologist can be arranged for one-on-one communication with the elderly to help them vent their feelings of loneliness and relieve psychological stress; at the same time, volunteers can be organized to establish long-term companionship with the elderly, regularly visiting or calling to provide emotional support and companionship. If the target object has experienced temporary emotional improvement due to participation in a mutual support group in historical intervention, and is interested in gardening activities, the mutual support group activity plan can be adjusted and strengthened by combining it with gardening activities, organizing a gardening-themed mutual support group activity twice a week, each lasting at least three hours. The activity location can be chosen in the community garden or indoor gardening area, guided by professional gardeners and psychologists, allowing the target object to exchange and support each other in the process of gardening and plant care, satisfying their interest and providing a platform for emotional release and social interaction, thereby more effectively alleviating their feelings of loneliness.

[0119] Through multi-source data fusion and intelligent analysis technology, the scientificity and practicality of social health management of the elderly can be significantly improved. First, the system can collect and integrate multi-dimensional behavior data such as call records, social media, activity participation, address book interaction, smart home, etc. in real time, and comprehensively restore the real social network structure and interaction dynamics of the elderly. Second, through advanced data mining and machine learning algorithms, the system can automatically identify weak links and loneliness risks in the social network, realize dynamic and accurate monitoring of individual social status, and overcome the limitations of traditional isolated and static evaluation. In addition, the system has a personalized intervention engine, which can intelligently recommend a variety of social activities, psychological care or volunteer services according to the interests, social preferences and actual needs of each elderly person, significantly improving the pertinence and effectiveness of intervention. At the same time, the platform supports unified management and visualization of multi-source data, providing intuitive and scientific decision support for service personnel and family members. Thus, the early detection and intervention capacity for social loneliness of the elderly can be greatly enhanced, which helps to improve the quality of life and mental health of the elderly and promote the innovative development of smart elderly care services.

[0120] The embodiments of the present application are based on multi-dimensional social behavior data of the elderly in daily life with family, friends, neighbors, service personnel, etc. The social network structure, interaction frequency, interaction content and emotional tendency are comprehensively analyzed, combined with psychological health questionnaire, voice tone, etc. information, and with the help of artificial intelligence model, the loneliness risk is actively identified and quantified. The system automatically pushes personalized social activities, psychological care or professional intervention measures, realizes early intervention and accurate management of social loneliness.

[0121] It can be seen that in the above scheme, by collecting multi-dimensional social behavior data of the target object, the data dimension is enriched, and the social status of the target object is more comprehensively reflected, avoiding the wrong evaluation of the social risk of the target object due to one-sided data, providing accurate and comprehensive data support for social risk assessment. By calculating the social network characteristics of the target object, the multi-dimensional social behavior data of the target object is converted into quantitative indicators with analysis value, which can more accurately identify the position and state of the target object in the social network, providing key quantitative basis for subsequent loneliness risk assessment. By analyzing the voice data through the emotion analysis model, the emotional state of the target object in social interaction can be captured, and the emotional dimension information is added for social risk assessment. Based on the loneliness risk assessment model, the social network characteristics and voice emotion characteristics are comprehensively analyzed, a reliable loneliness risk prediction value is generated, which intuitively reflects the possibility of the target object falling into a lonely state, and provides a key decision basis for developing loneliness intervention strategies. According to different loneliness risk levels, personalized loneliness intervention strategies are developed, which can ensure the pertinence and effectiveness of the intervention measures, realize early warning and timely intervention, and significantly improve the intervention accuracy and timeliness of the social health of the elderly in home-based care.

[0122] It should be understood that the size of the serial number of each step in the above embodiment does not mean the order of execution, and the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the application.

[0123] In an embodiment, a multi-dimensional data-based loneliness risk prediction device is provided, which corresponds to the multi-dimensional data-based loneliness risk prediction method in the above embodiment. As shown in the figure, the multi-dimensional data-based loneliness risk prediction device includes a multi-dimensional data acquisition module 201, a social network feature calculation module 202, a voice emotion feature analysis module 203, a loneliness risk assessment module 204, and a loneliness intervention policy generation module 205. The detailed description of each functional module is as follows: Figure 4

[0124] The multi-dimensional data acquisition module 201 is configured to acquire multi-dimensional social behavior data of a target object.

[0125] The social network feature calculation module 202 is configured to calculate social network features of the target object based on the multi-dimensional social behavior data.

[0126] The voice emotion feature analysis module 203 is configured to analyze the sentiment of a sentence in voice data of the target object based on a sentiment analysis model to obtain voice emotion features.

[0127] The loneliness risk assessment module 204 is configured to perform feature analysis on the social network features and the voice emotion features based on a loneliness risk assessment model to generate a loneliness risk prediction value of the target object.

[0128] The loneliness intervention policy generation module 205 is configured to generate a loneliness intervention strategy for the target object based on the loneliness risk prediction value.

[0129] In an embodiment, the multi-dimensional data acquisition module 201 includes:

[0130] A platform interfacing unit is configured to interface with at least one social data management platform.

[0131] A data acquisition unit is configured to acquire multi-dimensional social behavior data of the target object based on an identity of the target object.

[0132] In an embodiment, the social network feature calculation module 202 includes:

[0133] ​a social network construction unit configured to extract at least one social object of the target object based on the multidimensional social behavior data of the target object, and construct a social network between the target object and each of the social objects;

[0134] a social data determination unit configured to determine social data of the target object and each of the social objects based on the social network, wherein the social data comprises a social relationship and an interaction frequency;

[0135] a social network feature calculation unit configured to calculate a social network feature of the target object based on the social data of the target object and each of the social objects.

[0136] In an embodiment, the voice emotion feature analysis module 203 comprises:

[0137] a voice text data extraction unit configured to extract voice text data of the target object based on the multidimensional social behavior data of the target object;

[0138] an emotion analysis unit configured to perform emotion analysis on each piece of voice text in the voice text data based on the emotion analysis model, and obtain an emotion score of each piece of voice text;

[0139] a voice emotion feature determination unit configured to determine a voice emotion feature of the target object based on the emotion score of each piece of voice text.

[0140] In an embodiment, the loneliness risk assessment module 204 comprises:

[0141] a feature vector construction unit configured to construct a social feature vector of the target object based on the social network feature and the voice emotion feature;

[0142] a loneliness risk prediction unit configured to perform prediction on the social feature vector based on the loneliness risk assessment model, and output a loneliness risk prediction value.

[0143] In an embodiment, the loneliness intervention policy generation module 205 comprises:

[0144] a loneliness risk level determination unit configured to determine a loneliness risk level of the target object based on a comparison result of a preset loneliness risk threshold and the loneliness risk prediction value;

[0145] a loneliness intervention strategy determination unit configured to determine a loneliness intervention strategy of the target object based on the loneliness risk level of the target object.

[0146] In an embodiment, the loneliness intervention strategy determination unit comprises:

[0147] a historical data collection subunit configured to collect historical intervention data and historical interest data of the target object;

[0148] an intervention measure determination subunit configured to determine at least one intervention measure based on the historical intervention data and the historical interest data;

[0149] a loneliness intervention strategy determination subunit configured to determine a loneliness intervention strategy of the target object based on the loneliness risk level of the target object and the at least one intervention measure.

[0150] The application provides a loneliness risk prediction device based on multi-dimensional data, which collects multi-dimensional social behavior data of a target object, enriches the data dimensions, more comprehensively reflects the social state of the target object, avoids the one-sided data from leading to an incorrect evaluation of the social risk of the target object, and provides accurate and comprehensive data support for the social risk evaluation. The multi-dimensional social behavior data of the target object is converted into quantitative indicators with analysis value by calculating the social network features of the target object, so that the position and state of the target object in the social network can be more accurately identified, and key quantitative basis is provided for subsequent loneliness risk evaluation. The voice data is analyzed by the sentiment analysis model, so that the emotional state of the target object in social interaction can be captured, and the information of the emotional dimension is added to the social risk evaluation. The social network features and the voice emotional features are comprehensively analyzed based on the loneliness risk evaluation model, so that a reliable loneliness risk prediction value can be generated, the possibility of the target object falling into a lonely state is intuitively reflected, key decision basis is provided for formulating a loneliness intervention strategy, personalized loneliness intervention strategies are formulated according to different loneliness risk levels, the pertinence and effectiveness of the intervention measures can be ensured, early warning and timely intervention are realized, and the intervention accuracy and timeliness of the social health of the elderly in home-based care are significantly improved.

[0151] The specific limitations of the loneliness risk prediction device based on multi-dimensional data can be referred to the limitations of the loneliness risk prediction method based on multi-dimensional data in the above, and will not be repeated here. Each module in the loneliness risk prediction device based on multi-dimensional data can be realized by software, hardware and a combination thereof in whole or in part. The above modules can be embedded in or independent of the processor in the computer device in hardware form, or can be stored in the memory in the computer device in software form, so as to be called and executed by the processor to perform the operations corresponding to the above modules.

[0152] In one embodiment, a computer device is provided, which can be a server, and the internal structure diagram thereof can be as shown in Figure 5As shown in the figure. The computer device includes a processor, a memory, a network interface and a database connected through a web application bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes non-volatile and / or volatile storage media, internal memory. The non-volatile storage medium stores the operating web application, computer program and database. The internal memory provides an environment for the running of the operating web application and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with the external client through the network connection. The computer program is executed by the processor to realize the functions or steps of the server side of the loneliness risk prediction method based on multi-dimensional data.

[0153] In one embodiment, a computer device is provided, which can be a client, and its internal structure diagram can be as shown in the figure. Figure 6 The computer device includes a processor, a memory, a network interface, a display screen and an input device connected through a web application bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes non-volatile storage media, internal memory. The non-volatile storage medium stores the operating web application and computer program. The internal memory provides an environment for the running of the operating web application and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with the external server through the network connection. The computer program is executed by the processor to realize the functions or steps of the client side of the loneliness risk prediction method based on multi-dimensional data.

[0154] In one embodiment, a computer device is provided, including a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor executes the computer program to realize the following steps:

[0155] Obtain the multi-dimensional social behavior data of the target object;

[0156] Based on the multi-dimensional social behavior data, calculate the social network features of the target object;

[0157] Based on the sentiment analysis model, analyze the sentence sentiment in the voice data of the target object to obtain the voice emotion features;

[0158] Based on the loneliness risk assessment model, analyze the social network features and the voice emotion features to generate the loneliness risk prediction value of the target object;

[0159] Based on the loneliness risk prediction value, generate the loneliness intervention strategy of the target object.

[0160] In one embodiment, a computer readable storage medium is provided, and a computer program is stored on the computer readable storage medium, and the computer program is executed by a processor to implement the following steps:

[0161] Obtaining multi-dimensional social behavior data of a target object;

[0162] Based on the multi-dimensional social behavior data, calculating social network features of the target object;

[0163] Based on an emotional analysis model, analyzing the sentence emotion in the voice data of the target object to obtain voice emotion features;

[0164] Based on a loneliness risk assessment model, performing feature analysis on the social network features and the voice emotion features to generate a loneliness risk prediction value of the target object;

[0165] Based on the loneliness risk prediction value, generating a loneliness intervention strategy for the target object.

[0166] It should be noted that the functions or steps that the computer readable storage medium or the computer device can implement correspond to the related descriptions of the server side and the client side in the foregoing method embodiments, and to avoid repetition, they will not be described one by one here.

[0167] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by a computer program instructing related hardware, and the computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments. In the embodiments provided in the present application, any reference to memory, storage, database or other medium can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0168] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above-mentioned division of each functional unit and module is exemplified, and in actual application, the above-mentioned functions can be completed by different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above.

[0169] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that the technical solutions recorded in the foregoing embodiments can be modified, or some technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A loneliness risk prediction method based on multidimensional data, characterized in that: The method comprises: Obtain multi-dimensional social behavior data of the target object; Calculating social network characteristics of the target object based on the multi-dimensional social behavior data; Analyzing the sentence emotions in the speech data of the target object based on the sentiment analysis model to obtain speech emotion features; Based on the loneliness risk assessment model, feature analysis is performed on the social network features and the voice emotion features to generate a loneliness risk prediction value for the target subject; Based on the loneliness risk prediction value, a loneliness intervention strategy for the target subject is generated.

2. The loneliness risk prediction method based on multidimensional data according to claim 1, characterized in that: The step of obtaining the multi-dimensional social behavior data of the target object includes: Connect to at least one social data management platform; Based on the identity identifier of the target object, social behavior data of at least one data type corresponding to the target object in each of the social data management platforms is collected to obtain multi-dimensional social behavior data of the target object.

3. The loneliness risk prediction method based on multidimensional data according to claim 1, characterized in that: The calculating the social network characteristics of the target object based on the multi-dimensional social behavior data includes: extracting at least one social object of the target object based on the multi-dimensional social behavior data, and building a social network between the target object and each of the social objects; Based on the social network, determining social data between the target object and each of the social objects, wherein the social data includes social relationships and interaction frequencies; Based on the social data of the target object and each of the social objects, a social network feature of the target object is calculated.

4. The loneliness risk prediction method based on multidimensional data according to claim 1, characterized in that: The emotion analysis model is used to analyze the emotion of the sentences in the speech data of the target object to obtain the emotion features of the speech, including: Extracting speech and text data of the target object based on the multi-dimensional social behavior data of the target object; Based on the sentiment analysis model, sentiment analysis is performed on each voice text in the voice text data to obtain a sentiment score for each voice text; Based on the emotion scores of the speech texts, the speech emotion characteristics of the target object are determined.

5. The loneliness risk prediction method based on multidimensional data according to claim 1, characterized in that: The loneliness risk assessment model is based on which the social network features and the voice emotion features are analyzed to generate a loneliness risk prediction value for the target subject, including: Constructing a social feature vector of the target object based on the social network features and the voice emotion features; Based on the loneliness risk assessment model, the social feature vector is predicted and the loneliness risk prediction value is output.

6. The loneliness risk prediction method based on multidimensional data according to claim 1, characterized in that: Generating a loneliness intervention strategy for the target subject based on the loneliness risk prediction value includes: Determining the loneliness risk level of the target subject based on a comparison result of a preset loneliness risk threshold and the loneliness risk prediction value; A loneliness intervention strategy for the target subject is determined based on the loneliness risk level of the target subject.

7. The loneliness risk prediction method based on multidimensional data according to claim 6, characterized in that: The step of determining a loneliness intervention strategy for the target subject based on the loneliness risk level of the target subject includes: Collecting historical intervention data and historical interest data of the target subject; determining at least one intervention measure based on the historical intervention data and the historical interest data; A loneliness intervention strategy for the target subject is determined based on the loneliness risk level of the target subject and the at least one intervention measure.

8. A loneliness risk prediction device based on multidimensional data, characterized in that: The loneliness risk prediction device based on multidimensional data includes: A multi-dimensional data acquisition module is used to obtain multi-dimensional social behavior data of the target object; A social network feature calculation module, configured to calculate the social network features of the target object based on the multi-dimensional social behavior data; A speech emotion feature analysis module is used to analyze the sentence emotions in the speech data of the target object based on the emotion analysis model to obtain speech emotion features; A loneliness risk assessment module, configured to perform feature analysis on the social network features and the voice emotion features based on a loneliness risk assessment model to generate a loneliness risk prediction value for the target subject; A loneliness intervention policy generation module is used to generate a loneliness intervention strategy for the target object based on the loneliness risk prediction value.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the loneliness risk prediction method based on multidimensional data as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the loneliness risk prediction method based on multidimensional data as claimed in any one of claims 1 to 7 are implemented.

Images