Personalized psychological counseling interactive decision-making system and method based on artificial intelligence

Abstract

The invention relates to the technical field of psychological counseling, in particular to a personalized psychological counseling interactive decision-making system and method based on artificial intelligence, and the system comprises a data collection end, a data processing end, an auxiliary decision-making end and a behavior supervision end, and can achieve the collection of image, sound and physiological state data during psychological counseling. Through data processing, data flows to a network model combining the psychological problem identification and classification neural network and reinforcement learning, the model can automatically ask questions and continuously optimize the psychological problem identification precision, and finally the effect of assisting a consultant in psychological consultation is achieved. The behavior supervision end of the system can collect and analyze data in the psychological rehabilitation period in real time, the rehabilitation condition of a consultant is clear, and related personnel are reminded to pay attention in time.

Description

Technical Field

[0001] The present invention relates to the field of psychological counseling technology, and in particular to an artificial intelligence-based personalized psychological counseling interactive decision-making system and method. Background Art

[0002] The significance of psychological counseling in modern society is profound and extensive. It can improve interpersonal relationships, promote personal growth and development, enhance individual psychological quality and quality of life, and solve emotional problems and adaptation problems. Among them, the psychological problems of special people in special occasions need to be given due attention. According to the results of relevant psychological tests, a considerable number of special people in special occasions have psychological disorders to varying degrees, especially in the early stage of entering special occasions, which is the "hardest hit area" for the outbreak of psychological problems of special people.

[0003] Psychological counseling for special persons refers to the process in which special occasions or related institutions use psychological knowledge, methods and techniques to intervene and guide the psychological problems of special persons in special occasions in order to promote their mental health and enhance the effect of reform. This process is of great significance for the resocialization of special persons and reducing the recidivism rate.

[0004] Conventional psychological counseling methods are that the counselor listens carefully to the counselee's narration and constantly explores the counselee's true psychological demands, and then through professional guidance, the counselee can begin to look at the problem from a new perspective, slowly understand the cause and effect of the incident, understand why he is trapped, and finally take positive actions to eliminate the corresponding psychological problems. However, in the process of psychological counseling for special people, there are the following problems: 1. Traditional psychological counseling methods lack attention to the consultant's expression, movements, tone of voice, etc., and due to nervousness, the consultant may narrate in a trivial manner without focus. If the counselor cannot quickly understand and accurately get to the topic, it may take more time and energy to find the consultant's real psychological problem; 2. During the rehabilitation period after psychological counseling, due to the special nature of special personnel, frequent telephone communication cannot be achieved, making it difficult to track the counselor's psychological condition in real time. Summary of the invention

[0005] In view of this, the purpose of the present invention is to provide a personalized psychological counseling interactive decision-making system and method based on artificial intelligence, which can assist counselors to quickly locate the counselors' psychological problems and timely detect and track the psychological rehabilitation effects.

[0006] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: an artificial intelligence-based personalized psychological consultation interactive decision-making system, comprising a data collection end, a data processing end, a decision-making assistance end, and a behavior supervision end; The data collection end includes image data collection, voice data collection and physiological data collection: image data collection collects the consultant's facial expressions and body movement pictures and feeds them back to the data processing end; voice data collection collects the consultant's narrative content and captures the consultant's speech speed and tone information, and the relevant content and information are fed back to the data processing end; physiological data collection uses wearable devices to collect data implicitly, and the collected data includes at least heart rate information and body temperature information; The data processing end receives and processes the data collected by the data collection end. The narrative content of the consultant needs to be processed. The semantic key words are extracted and converted into word vectors by word embedding. The consultant's speaking speed and tone are represented by numbers and normalized. The consultant's facial expressions and body movement pictures are identified by neural networks and the consultant's emotions are labeled. The corresponding word vectors are associated with the corresponding speaking speed, tone, consultant's emotions, current heart rate, and body temperature information to form a feature vector. The deep learning neural network model is used to receive the feature vector input for psychological problem identification and the reinforcement learning method is used to ask questions to the consultant, improve the reliability of psychological problem identification, clarify the consultant's psychological condition, and give auxiliary suggestions based on the consultant's situation. The decision-making support end receives data from the data processing end, including: original narrative content, key vocabulary, corresponding speech speed, intonation, facial expressions, body movements, heart rate, body temperature, and the consultant's emotions, psychological problems, and consulting suggestions determined by the neural network. The relevant data is displayed to the consultant for reference; The behavior monitoring end is installed in the consultant's wearable device. Based on the psychological counseling suggestions finally confirmed by the counselor, it regularly reminds the consultant of his / her behavior and monitors the consultant's behavior and sends it to the data processing end to detect abnormal situations in time.

[0007] The data processing of the consultant's narrative content at least includes: data cleaning, word segmentation, stem extraction, standardization, and word embedding.

[0008] The questions asked in data processing are those summarized based on expert knowledge, and the questions asked on the data processing end are selected from the question bank using a reinforcement learning strategy based on the currently collected information and the results of psychological problem identification.

[0009] Psychological problem identification is obtained by reasoning with a separate neural network model. The identification result data set is a psychological problem data set based on expert knowledge summary. Psychological problems include at least: anxiety, depression, anger, fear, resentment, inferiority, and disappointment.

[0010] In particular, the behavior monitoring end automatically processes psychological counseling advice information and refines reminder timing and reminder content, where the reminder timing is the counselor's recommended reminder time, the corresponding reminder timing preset for psychological problems, and the timing when the behavior monitoring end detects abnormal data.

[0011] Furthermore, the behavior monitoring terminal detects abnormal data by at least detecting the consultant's heart rate, body temperature, and voice decibel level in real time and drawing corresponding time-varying curves. When the gradient of the heart rate, body temperature data, and voice decibel level changes exceeds a set threshold at a certain moment, it is determined that the current consultant's behavior is abnormal and needs to be reminded. The reminder can be a reminder to the consultant himself or to people related to the consultant.

[0012] Furthermore, the behavior monitoring end also collects relevant authorization data of the consultant during the psychological counseling period and feeds it back to the auxiliary decision-making end for psychological rehabilitation assessment.

[0013] Finally, a personalized psychological counseling interactive decision-making method based on artificial intelligence is proposed. The processing logic of this method is as follows: the data processing end uses a reinforcement learning method to ask questions to the consultant, with the consultant's narration and answer information as the basic content, and the consultant's speech speed, tone, expression, body movement, heart rate, body temperature collected by the data collection end, and the consultant's emotions and psychological problems judged by the neural network as auxiliary content, which are jointly used as the reinforcement learning current state information State. The data processing end is the Agent, and the question to be asked is the Action. The final result accuracy value of the psychological problem recognition is the Reward. The policy policy derived from the state State is a psychological problem recognition method based on deep learning. During data processing, the questioning Agent based on reinforcement learning selects the Action according to the current State and obtains the consultant's psychological problem recognition results based on the deep learning method for psychological problem recognition and outputs the confidence score conf corresponding to each judged psychological problem i , where i represents the i-th psychological problem, conf i Represents the confidence score corresponding to the i-th psychological problem. All psychological problem results are summarized to calculate the reward, where .

[0014] The present invention also provides a computer-readable storage medium, which stores a program, and the program is executed by a processor to implement the above-mentioned personalized psychological counseling interactive decision-making method based on artificial intelligence.

[0015] Compared with the prior art, the present invention has the following beneficial effects: By collecting data such as images and sounds through the data collection terminal, the personalized psychological consultation interactive decision-making system and method based on artificial intelligence of the present invention can comprehensively use the consultant's narrative content, narrative tone, and micro-expressions and micro-movements during the narration to assist in determining the consultant's psychological state.

[0016] The present invention adopts a reinforcement learning method to automatically select questions to ask the consultant, adopts a deep learning neural network model to classify and judge the consultant's psychological problems, and uses this as a strategy model for reinforcement learning to guide the optimization of reinforcement learning, so that the final psychological problem judgment model makes a judgment result that the true psychological problem judgment has a high confidence degree, and the non-psychological problem judgment has a low confidence degree, thereby ensuring the accuracy of the consultant's psychological problem confirmation.

[0017] The behavior monitoring terminal of the present invention can monitor the status of the consultant after psychological consultation in real time, and can promptly remind the consultant of the cross-border behavior and the consultant's precautions. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of the system framework of the present invention; Figure 2 This is a schematic diagram of the personalized psychological consultation interactive decision-making method based on artificial intelligence of the present invention; Figure 3 This is a schematic diagram of optimizing the confidence level of psychological problem identification according to the present invention; Figure 4 This is an example interface of the decision-making assistance terminal of the present invention. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] Embodiment 1: Figure 1 to Figure 2 As shown, this embodiment provides a personalized psychological consultation interactive decision-making system based on artificial intelligence, including a data collection end, a data processing end, a decision-making assistance end, and a behavior supervision end; The data collection end includes image data collection, voice data collection and physiological data collection: image data collection collects the consultant's facial expressions and body movement pictures and feeds them back to the data processing end; voice data collection collects the consultant's narrative content and captures the consultant's speech speed and tone information, and the relevant content and information are fed back to the data processing end; physiological data collection uses wearable devices to collect data implicitly, and the collected data includes at least heart rate information and body temperature information; The data processing end receives and processes the data collected by the data collection end. The narrative content of the consultant needs to be processed. The semantic key words are extracted and converted into word vectors by word embedding. The consultant's speaking speed and tone are represented by numbers and normalized. The consultant's facial expressions and body movement pictures are identified by neural networks and the consultant's emotions are labeled. The corresponding word vectors are associated with the corresponding speaking speed, tone, consultant's emotions, current heart rate, and body temperature information to form a feature vector. The deep learning neural network model is used to receive the feature vector input for psychological problem identification and the reinforcement learning method is used to ask questions to the consultant, improve the reliability of psychological problem identification, clarify the consultant's psychological condition, and give auxiliary suggestions based on the consultant's situation. The decision-making support end receives data from the data processing end, including: original narrative content, key vocabulary, corresponding speech speed, intonation, facial expressions, body movements, heart rate, body temperature, and the consultant's emotions, psychological problems, and consulting suggestions determined by the neural network. The relevant data is displayed to the consultant for reference; The behavior monitoring end is installed in the consultant's wearable device. Based on the psychological counseling suggestions finally confirmed by the counselor, it regularly reminds the consultant of his / her behavior and monitors the consultant's behavior and sends it to the data processing end to detect abnormal situations in time.

[0021] The data processing of the consultant's narrative content at least includes: data cleaning, word segmentation, stem extraction, standardization, and word embedding.

[0022] The questions asked in data processing are those summarized based on expert knowledge, and the questions asked on the data processing end are selected from the question bank using a reinforcement learning strategy based on the currently collected information and the results of psychological problem identification.

[0023] Psychological problem identification is obtained by reasoning with a separate neural network model. The identification result data set is a psychological problem data set based on expert knowledge summary. Psychological problems include at least: anxiety, depression, anger, fear, resentment, inferiority, and disappointment.

[0024] At the same time, the present invention proposes a personalized psychological consultation interactive decision-making method based on artificial intelligence. The processing logic of the method is: the data processing end adopts a reinforcement learning method to the consultant's questions, with the consultant's narration content and the information of the answer as the basic content, and the consultant's speech speed, tone, expression, body movement, heart rate, body temperature collected by the data collection end, and the consultant's emotions and psychological problems judged by the neural network as auxiliary content, which are jointly used as the reinforcement learning current state information State, the data processing end is Agent, the question to be asked is Action, the final result accuracy value of psychological problem recognition is Reward, and the policy Policy of deriving Action from the state State is a psychological problem recognition method based on deep learning. During data processing, the questioning Agent based on reinforcement learning selects Action according to the current State and obtains the consultant's psychological problem recognition result based on the psychological problem recognition deep learning method and outputs the confidence score conf corresponding to each judged psychological problem i , where i represents the i-th psychological problem, conf i Represents the confidence score corresponding to the i-th psychological problem. All psychological problem results are summarized to calculate the reward, where .

[0025] Due to the special nature of special people, they may have strong psychological barriers. Therefore, when special people conduct psychological inquiries actively or passively, counselors need to quickly understand what the counselors express and explore their psychological demands. However, counselors usually cannot express themselves clearly, and some special people may even have residual sins and selectively ignore or mislead counselors. This greatly increases the difficulty for counselors to find the real psychological problems of counselors. According to the system and method of the present invention, when a consultant is consulting, a voice acquisition device acquires the voice of the consultant and converts it into text, and at the same time, the voice acquisition acquires the speaker's speech speed and intonation and corresponds to the corresponding text; the camera acquires images and uses a neural network model to recognize the facial expressions and movements of the person when speaking, and then judges the emotions of the person when speaking, thereby giving corresponding weights to the content of the consultant; finally, wearable devices (including smart bracelets, special smart clothing, etc.) worn by special personnel are used to collect data such as the consultant's heart rate and body temperature at the time, so as to grasp the consultant's key content more accurately and quickly. At the same time, due to the particularity of special occasions, cameras, recording devices, and wearable devices are not special. Relatively speaking, special personnel for consultation are not prone to nervousness, which is more conducive to the normal emotional expression of the consultant. Therefore, when the consultant unconsciously speeds up the speech speed, emphasizes the tone, and has a faster heartbeat and a higher body temperature, the consultant may be more excited at this time, and the content of the expression may be the key factor of the psychological problem, which can be further communicated after being comforted. When the eyes are wagging around and the consultant is pretending to talk about something else, the consultant in special occasions may be covering up certain content. The narrated content needs to be deducted in weight and further questions need to be asked. Moreover, if some quick questions and answers can be used to ask questions, the consultant may not have time to think too much and make more realistic emotional expressions. Therefore, using the system and method of the present invention, during consultation, the collection system continuously collects data and sends relevant original information, stems, and auxiliary information to the consultant. At the same time, the internal artificial intelligence algorithm adopts a combination of reinforcement learning and a psychological problem classification model (a kind of intention recognition) to continuously find the questions preset by the expert according to the current content, context and the speaker's performance, and automatically ask questions. Since reinforcement learning takes the confidence of the psychological problem classification model as the optimization goal, it continuously asks questions and continuously adds the content obtained after asking questions to the input of the classification model. The final goal of reinforcement learning after asking questions is to make the confidence score of possible psychological problems higher (close to 1) and the confidence score of psychological problems lower (close to 0). Figure 3As shown, the judgment results given by the psychological problem classification model after this method will have higher credibility, and due to the powerful computing power of computers, their data processing speed is more real-time than manual processing. Therefore, the system and method of the present invention can quickly lock possible psychological problems in the early stage and quickly learn through reinforced questioning, allowing the consultant to express real emotions in quick answers, and reinforced learning is trained in continuous trial and error. The goal of reinforcement is the confirmation degree (confidence) of psychological problems. Therefore, the questions of reinforced learning can be directed (targeted questions), so as to quickly confirm psychological problems and give consulting suggestions set in advance according to expert knowledge based on relevant psychological problems and emotional expressions. All this information will be dynamically synchronized to the consultant's auxiliary system interface in real time, such as Figure 4 As shown, the consultant can quickly refer to and give consulting suggestions. At the same time, the computer operations such as reinforcement learning questions here can be authorized by the consultant according to the current situation to allow the computer to automatically execute, automatically ask questions and give auxiliary conclusions, or the consultant can ask questions and summarize by himself based on the auxiliary conclusions.

[0026] Embodiment 2: The behavior monitoring terminal of the present invention automatically processes psychological counseling advice information, and refines reminder timing and reminder content, wherein the reminder timing is the counselor's recommended reminder time, the corresponding reminder timing preset for the psychological problem, and the timing when the behavior monitoring terminal detects abnormal data.

[0027] Furthermore, the behavior monitoring end detects abnormal data by at least detecting the consultant's heart rate, body temperature, and voice decibel level in real time and drawing corresponding time-varying curves. When the gradient of the heart rate, body temperature data, and voice decibel level changes exceeds a set threshold at a certain moment, it is determined that the current consultant's behavior is abnormal and needs to be reminded. The reminder can be a reminder to the consultant himself or to related people.

[0028] Furthermore, the behavior monitoring end also collects relevant authorization data of the consultant during the psychological counseling period and feeds it back to the auxiliary decision-making end for psychological rehabilitation assessment.

[0029] The present invention also provides a computer-readable storage medium, which stores a program, and the program is executed by a processor to implement the above-mentioned personalized psychological counseling interactive decision-making method based on artificial intelligence.

[0030] When applied, the behavior monitoring end mainly relies on wearable devices, which can be the same as the physiological data collection device when collecting data. When the counselor finally confirms the counselor's psychological problem, the behavior monitoring end can query the recommended solution corresponding to a certain psychological problem preset in the database on the one hand, and receive the psychological reminder plan formulated by the counselor according to the actual situation on the other hand, and extract the reminder timing and reminder content therein. The reminder timing can be the counselor or a preset time point, or it can be triggered when a specified situation occurs. For example, when the psychological problem of a special person is confirmed to be fear, he or she has serious discomfort due to the new control. At this time, the behavior monitoring end regularly plays soothing text content and soothing music on the one hand, and receives the heart rate, temperature, sound decibel number and other information of the person at all times on the other hand. When the heart rate, temperature and speaking sound decibel number change significantly, the staff is reminded in time to prevent accidents from happening; when the special person has a violent problem, the behavior monitoring end detects the decibel number of the counselor's speaking volume. When the sound decibel number in a certain period of time is greater than a certain proportion of the mean sound decibel number (such as 2 times), at this time, the behavior monitoring end issues an alarm to remind the staff, and on the other hand, issues a preset reminder content to tell the counselor to calm down.

[0031] In addition, the behavior monitoring end also collects information such as the consultant's heart rate, temperature, and sound decibel level in real time, and compares it with historical data in the database and relevant data of normal people. Through analysis, it confirms whether the relevant consultant is getting better and completes the psychological rehabilitation assessment. For example, when the frequency of the decibel level of a violent person exceeding the threshold when communicating with others gradually decreases and the peak value of the decibel level gradually weakens, the reminder frequency of the behavior monitoring trigger mechanism is significantly reduced, indicating that the consultant is getting better.

[0032] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A personalized psychological consultation interactive decision-making system based on artificial intelligence, characterized by: The system includes data collection end, data processing end, decision support end and behavior supervision end; The data collection end includes image data collection, voice data collection and physiological data collection: the image data collection collects the consultant's expression and body movement pictures and feeds them back to the data processing end; Voice data collection collects the consultant's narrative content and captures the consultant's speech speed and tone information, and the relevant content and information are fed back to the data processing end; Physiological data collection is performed implicitly using wearable devices, and the collected data includes at least heart rate information and body temperature information; The data processing end receives and processes the data collected by the data collection end. The narrative content of the consultant needs to be processed, and the semantic key words are extracted and converted into word vectors using word embedding; The consultant's speaking speed and intonation were expressed in numbers and normalized; The consultant's facial expressions and body movements are identified using a neural network and the consultant's emotions are labeled. The corresponding word vectors are associated with the corresponding speech speed, tone, consultant's emotions, current heart rate, and body temperature information to form a feature vector. A deep learning neural network model is used to receive the feature vector input to identify psychological problems and use reinforcement learning methods to ask questions to the consultant, improve the reliability of psychological problem identification, clarify the consultant's psychological condition, and give auxiliary suggestions based on the consultant's situation. The decision-making support end receives data from the data processing end, including: original narrative content, key vocabulary, corresponding speech speed, intonation, facial expressions, body movements, heart rate, body temperature, and the consultant's emotions, psychological problems, and consulting suggestions determined by the neural network. The relevant data is displayed to the consultant for reference; The behavior monitoring end is installed in the consultant's wearable device. Based on the psychological counseling suggestions finally confirmed by the counselor, it reminds the consultant of the behavior and monitors the consultant's behavior and sends it to the data processing end to detect abnormal situations in time and issue alarms.

2. According to claim 1, a personalized psychological consultation interactive decision-making system based on artificial intelligence is characterized by: The data processing of the consultant's narrative content includes at least: data cleaning, word segmentation, stem extraction, standardization, and word embedding.

3. According to claim 2, a personalized psychological consultation interactive decision-making system based on artificial intelligence is characterized by: The questions asked in data processing are those summarized based on expert knowledge, and the questions asked on the data processing end are selected from the question bank using a reinforcement learning strategy based on the currently collected information and the results of psychological problem identification.

4. According to claim 3, a personalized psychological consultation interactive decision-making system based on artificial intelligence is characterized by: Psychological problem identification is obtained by reasoning with a separate neural network model. The identification result data set is a psychological problem data set based on expert knowledge summary. Psychological problems include at least: anxiety, depression, anger, fear, resentment, inferiority, and disappointment.

5. According to claim 4, a personalized psychological consultation interactive decision-making system based on artificial intelligence is characterized by: The consulting suggestions given by the decision-making support end are preset based on expert experience.

6. The personalized psychological consultation interactive decision-making system based on artificial intelligence according to claim 5 is characterized by: The behavior monitoring end automatically processes psychological counseling advice information and refines reminder timing and reminder content, where the reminder timing is the counselor's recommended reminder time, the corresponding reminder timing preset for psychological problems, and the timing when the behavior monitoring end detects abnormal data.

7. The personalized psychological consultation interactive decision-making system based on artificial intelligence according to claim 6 is characterized by: The behavior monitoring terminal detects abnormal data by at least detecting the consultant's heart rate, body temperature, and voice decibel level in real time and drawing the corresponding curve over time. When the gradient of the heart rate, body temperature, and voice decibel level data changes exceeds the set threshold at a certain moment, it is determined that the current consultant's behavior is abnormal and a reminder is required.

8. The personalized psychological consultation interactive decision-making system based on artificial intelligence according to claim 7 is characterized by: The reminders on the behavior supervision side include reminders to the consultant himself and reminders to the consultant's related persons.

9. The personalized psychological consultation interactive decision-making system based on artificial intelligence according to claim 8 is characterized by: The behavior monitoring end also collects relevant authorization data of the consultant during the psychological counseling period and feeds it back to the auxiliary decision-making end for psychological rehabilitation assessment.

10. An artificial intelligence-based personalized psychological consultation interactive decision-making method, the method is applied to an artificial intelligence-based personalized psychological consultation interactive decision-making system according to any one of claims 1 to 9, characterized in that: The reinforcement learning method used by the data processing end for the consultant's questions is: the consultant's narration and answer information are used as the basic content, and the consultant's speech speed, tone, expression, body movement, heart rate, body temperature collected by the data collection end, and the consultant's emotions and psychological problems judged by the neural network are used as auxiliary content, which are together used as the reinforcement learning current state information State. The data processing end is the Agent, and the question to be asked is the Action. The final result accuracy value of the psychological problem recognition is the Reward. The policy policy that derives the Action from the state State is the psychological problem recognition method based on deep learning. During data processing, the questioning Agent based on reinforcement learning selects the Action according to the current State and obtains the consultant's psychological problem recognition results based on the deep learning method for psychological problem recognition and outputs the confidence score conf corresponding to each judged psychological problem i , where i represents the i-th psychological problem, conf i Represents the confidence score corresponding to the i-th psychological problem. All psychological problem results are summarized to calculate the reward, where .

11. A computer-readable storage medium, characterized in that: The storage medium stores a program, and the program is executed by the processor to implement the personalized psychological consultation interactive decision-making method based on artificial intelligence as described in claim 10.

Images