An interactive AI education system

Abstract

The present invention provides an interactive AI education system, including: a front-end user interface module for building a friendly user interface; a speech recognition module for converting a patient's speech input signal into text information; a medical knowledge base construction module for building a medical knowledge base containing medical knowledge and education materials to support dynamic retrieval and content generation; a natural language processing module for understanding the patient's inquiry and intention recognition according to the text information and generating an answer text; an education content generation module for generating education content in real time according to the answer text; a speech synthesis module for converting the answer text into speech output; a video generation module for generating content and speech, referring to the patient information and the context of interaction, and generating an education video in real time; and sending the final education content output to the front-end user interface module to enhance the patient's interactive experience. The present invention can enhance the patient experience.

Description

Technical Field

[0001] The present invention relates to the field of medical information technology, and in particular to an interactive AI education system. Background Art

[0002] With the rapid development of medical technology, patients' demand for medical information is growing. However, traditional education methods are usually limited by manpower and time, and it is difficult to meet the personalized needs of patients. In addition, due to the complexity of medical knowledge and individual differences among patients, passive education methods have many shortcomings in practical applications. For example, patients may have doubts or demands when receiving education, which requires further explanation and clarification.

[0003] Therefore, it is of great practical significance to develop an intelligent AI education system that can interact with patients, dynamically generate education content, and answer questions. Summary of the invention

[0004] In order to overcome the shortcomings of the prior art, the purpose of the present invention is to provide an interactive AI education system, which shows significant beneficial effects by improving patient experience, improving information acquisition efficiency, enhancing knowledge dissemination effect, reducing the burden on medical staff, promoting the updating and maintenance of medical knowledge, supporting data analysis and decision-making, and improving the accessibility of medical services. It has broad application prospects and practical significance.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] An interactive AI education system, comprising:

[0007] The front-end user interface module is used to build a friendly user interface to display educational content and support patients to interact with the system through voice or text;

[0008] A speech recognition module, used to convert the patient's speech input signal into text information;

[0009] Medical knowledge base construction module, used to build a medical knowledge base containing medical knowledge and educational materials to support dynamic retrieval and content generation;

[0010] A natural language processing module, for understanding the patient's inquiry and intention based on the text information, and retrieving relevant information from the medical knowledge base to generate a response text;

[0011] A propaganda content generation module, used to generate initial propaganda content in real time according to the answer text based on a preset artificial intelligence decision-making body;

[0012] A speech synthesis module, used for converting the answer text into speech output based on speech synthesis technology;

[0013] The video education module is used to combine the context of the patient's interaction process, generate a final education video in real time according to the initial education content and the voice output, and send the final education video to the front-end user interface module to enhance the patient's interaction experience.

[0014] Preferably, the speech recognition module comprises:

[0015] A speech denoising unit, used to perform denoising on the collected speech input signal to obtain a denoised signal;

[0016] A normalization unit, used for performing normalization processing on the denoised signal to obtain a normalized signal;

[0017] A framing unit, used for performing framing processing on the normalized signal to obtain individual frame signals;

[0018] A feature extraction unit, used for extracting a speech feature vector from each of the frame signals;

[0019] An acoustic model processing unit, configured to determine the probability distribution of phonemes or words according to the speech feature vector based on a preset acoustic model;

[0020] A language model processing unit, used to determine a text sequence according to a probability distribution of phonemes or words based on a preset language model;

[0021] The post-processing unit is used to perform spelling check, punctuation completion and formatting processing on the generated text sequence to obtain the text information.

[0022] Preferably, the speech denoising unit comprises:

[0023] A signal translation subunit, used for performing translation on the voice input signal to obtain a translated collected signal;

[0024] A wavelet decomposition subunit, used for performing wavelet decomposition on the translated collected signal to obtain a plurality of wavelet coefficients;

[0025] A threshold determination subunit, used to determine a filtering threshold according to a decomposition scale and length of a signal;

[0026] A function determination subunit, used for constructing a denoising function according to the filtering threshold;

[0027] The denoising subunit is used to remove the noise of the translated collected signal by using the denoising function to obtain the denoised signal.

[0028] Preferably, the filtering threshold comprises: Among them, λ is the filtering threshold, j is the decomposition scale of the NMR acquisition signal after translation, and dj is the decomposition of the wavelet coefficients with a scale of j, N represents the length of the NMR acquisition signal after translation, and median represents the median operation.

[0029] Preferably, the denoising subunit comprises:

[0030] A smoothing component is used to remove the corresponding wavelet coefficients using the denoising function to obtain a smoothed collected signal; wherein the denoising function is: Among them, X represents the wavelet coefficient, n represents the adjustment coefficient;

[0031] An inverse translation component, used for inversely translating the smoothed collected signal to obtain an inversely translated collected signal;

[0032] The denoising component is used to obtain the average of multiple inverse-translated collected signals to obtain the denoised signal.

[0033] Preferably, the acoustic model is a CNN-LSTM combined network structure.

[0034] Preferably, the language model is an N-gram model.

[0035] Preferably, the natural language processing module includes:

[0036] A text preprocessing unit, used for preprocessing the text information to obtain preprocessed text data;

[0037] An intention recognition unit, used for inputting the preprocessed text data into a trained intention recognition model to obtain the patient's intention;

[0038] An entity recognition unit, used for inputting the preprocessed text data into a trained named entity recognition model to obtain key entities;

[0039] A query unit, used to construct a query statement according to the patient's intention and the key entity, and execute a query in the medical knowledge base according to the query statement to obtain retrieval information;

[0040] The answer generation unit is used to generate the answer text according to the search information.

[0041] Preferably, the education content generation module includes:

[0042] A content analysis unit, configured to analyze the answer text to identify key information of the answer text; the key information includes: medical terms, specific needs of the patient and background information;

[0043] A template selection unit, used to select a suitable template from a preset content template library according to the analysis results; the templates are classified according to different medical topics, patient types and interaction scenarios to ensure that the generated content meets the needs of patients;

[0044] A content generation unit, used to combine the answer text with the selected template and dynamically generate personalized initial propaganda content using natural language generation technology;

[0045] The formatted data unit is used to format the personalized initial education content to meet the display requirements of the front-end user interface.

[0046] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0047] The present invention provides an interactive AI education system, comprising: a front-end user interface module, which is used to build a friendly user interface to display education content and support patients to interact with the system through voice or text; a voice recognition module, which is used to convert the patient's voice input signal into text information; a medical knowledge base construction module, which is used to build a medical knowledge base containing medical knowledge and education materials to support dynamic retrieval and content generation; a natural language processing module, which is used to understand the patient's inquiry and intention based on text information, and retrieve relevant information from the medical knowledge base to generate an answer text; an education content generation module, which is used to generate initial education content in real time based on the answer text based on a preset artificial intelligence decision body; a speech synthesis module, which is used to convert the answer text into speech output based on speech synthesis technology; a video education module, which is used to combine the contextual relationship of the patient's interaction process, generate a final education video in real time based on the initial education content and the speech output, and send the final education video to the front-end user interface module to enhance the patient's interactive experience. The present invention has significant beneficial effects by improving patient experience, increasing information acquisition efficiency, enhancing knowledge dissemination effect, reducing the burden on medical staff, promoting the updating and maintenance of medical knowledge, supporting data analysis and decision-making, and improving the accessibility of medical services. It has broad application prospects and practical significance. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0049] Figure 1 A schematic diagram of the system structure provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0050] 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.

[0051] The purpose of the present invention is to provide an interactive AI education system, which shows significant beneficial effects by improving patient experience, improving information acquisition efficiency, enhancing knowledge dissemination effect, reducing the burden on medical staff, promoting the updating and maintenance of medical knowledge, supporting data analysis and decision-making, and improving the accessibility of medical services. It has broad application prospects and practical significance.

[0052] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0053] Figure 1 A schematic diagram of the system structure provided by an embodiment of the present invention, such as Figure 1 As shown, the present invention provides an interactive AI education system, comprising:

[0054] The front-end user interface module is used to build a friendly user interface to display educational content and support patients to interact with the system through voice or text;

[0055] A speech recognition module, used to convert the patient's speech input signal into text information;

[0056] Medical knowledge base construction module, used to build a medical knowledge base containing medical knowledge and educational materials to support dynamic retrieval and content generation;

[0057] A natural language processing module, for understanding the patient's inquiry and intention based on the text information, and retrieving relevant information from the medical knowledge base to generate a response text;

[0058] A propaganda content generation module, used to generate initial propaganda content in real time according to the answer text based on a preset artificial intelligence decision-making body;

[0059] A speech synthesis module, used for converting the answer text into speech output based on speech synthesis technology;

[0060] The video education module is used to combine the context of the patient's interaction process, generate a final education video in real time according to the initial education content and the voice output, and send the final education video to the front-end user interface module to enhance the patient's interaction experience.

[0061] Specifically, the design of the front-end user interface of this embodiment is user-centric, ensuring that the interface is simple, intuitive and easy to operate. The needs of different user groups, including patients, medical staff and managers, need to be considered during the design process. The interface contains a clear navigation bar to facilitate users to quickly find the required educational content and functional modules. Color matching and icon design are used to enhance visual appeal, and this embodiment ensures the readability and accessibility of information. In addition, the interface should support responsive design to adapt to the screen sizes of different devices (such as mobile phones, tablets and computers) to ensure that users can get a good user experience on various devices.

[0062] In order to support patients to interact with the system through voice or text, the front-end user interface needs to integrate voice recognition and text input functions. The text input box should have automatic completion and suggestion functions to help users quickly enter questions or requests. At the same time, the voice interaction function is realized by integrating voice recognition API (such as Google Speech-to-Text or other related technologies), and users only need to click the microphone icon to start voice input. The system should convert speech to text in real time and display it on the interface to ensure that users can confirm the accuracy of the input. In addition, the interface should provide a feedback mechanism to respond to user input in a timely manner to ensure the smoothness and naturalness of the interaction.

[0063] The front-end user interface module of this embodiment can realize the dynamic content display function so that the educational content can be updated in real time according to the needs and inputs of the users. The system should retrieve relevant information from the medical knowledge base and present it to the users in an easy-to-understand way, such as using pictures and texts, video tutorials or interactive questions and answers. In order to improve the user experience, the interface should support personalized recommendations of content and intelligently push relevant educational materials based on the user's historical queries and preferences. In addition, the interface should have good scalability so that the displayed content can be flexibly adjusted and expanded in the future according to the update of medical knowledge and changes in user needs.

[0064] Preferably, the speech recognition module comprises:

[0065] A speech denoising unit, used to perform denoising on the collected speech input signal to obtain a denoised signal;

[0066] A normalization unit, used for performing normalization processing on the denoised signal to obtain a normalized signal;

[0067] A framing unit, used for performing framing processing on the normalized signal to obtain individual frame signals;

[0068] A feature extraction unit, used for extracting a speech feature vector from each of the frame signals;

[0069] An acoustic model processing unit, configured to determine the probability distribution of phonemes or words according to the speech feature vector based on a preset acoustic model;

[0070] A language model processing unit, used to determine a text sequence according to a probability distribution of phonemes or words based on a preset language model;

[0071] The post-processing unit is used to perform spelling check, punctuation completion and formatting processing on the generated text sequence to obtain the text information.

[0072] Preferably, the speech denoising unit comprises:

[0073] A signal translation subunit, used for performing translation on the voice input signal to obtain a translated collected signal;

[0074] A wavelet decomposition subunit, used for performing wavelet decomposition on the translated collected signal to obtain a plurality of wavelet coefficients;

[0075] A threshold determination subunit, used to determine a filtering threshold according to a decomposition scale and length of a signal;

[0076] A function determination subunit, used for constructing a denoising function according to the filtering threshold;

[0077] The denoising subunit is used to remove the noise of the translated collected signal by using the denoising function to obtain the denoised signal.

[0078] Preferably, the filtering threshold comprises: Among them, λ is the filtering threshold, j is the decomposition scale of the NMR acquisition signal after translation, and d j is the decomposition of the wavelet coefficients with a scale of j, N represents the length of the NMR acquisition signal after translation, and median represents the median operation.

[0079] Specifically, in the wavelet threshold denoising process, the quality of threshold selection directly affects the effect of denoising. Since the wavelet coefficient amplitude of noise has the characteristic of decreasing with the increase of decomposition scale, this embodiment introduces the decomposition scale to adaptively adjust the filtering threshold.

[0080] Preferably, the denoising subunit comprises:

[0081] A smoothing component is used to remove the corresponding wavelet coefficients using the denoising function to obtain a smoothed collected signal; wherein the denoising function is: Among them, X represents the wavelet coefficient, n represents the adjustment coefficient;

[0082] An inverse translation component, used for inversely translating the smoothed collected signal to obtain an inversely translated collected signal;

[0083] The denoising component is used to obtain the average of multiple inverse-translated collected signals to obtain the denoised signal.

[0084] Specifically, this embodiment uses wavelet coefficients to measure the attenuation deviation, so that the attenuation of wavelet coefficients with larger absolute values ​​decreases as their absolute values ​​increase, thereby effectively avoiding the loss of high-frequency information and improving the signal-to-noise ratio of the signal.

[0085] Optionally, although the use of a denoising function is a simple and effective denoising method, in some cases, such as in the neighborhood of a signal singular point, the reconstructed signal may have large upper and lower peaks at that location. These peaks are not inherent to the original signal, but due to the lack of translation invariance of the threshold method, the denoised signal may have distortions such as pseudo-Gibbs effect. Therefore, this embodiment can not only effectively suppress the pseudo-Gibbs effect by continuously translating the signal, but also obtain a smaller mean square error than the threshold method denoising, further improving the signal-to-noise ratio.

[0086] Specifically, the construction of the medical knowledge base of this embodiment first involves comprehensive data collection. This process includes obtaining medical knowledge and educational materials from various sources, such as medical literature, clinical guidelines, patient education manuals, scientific research papers, and expert interviews. The collected data needs to be sorted and cleaned to ensure the accuracy and consistency of the information. Specifically, data sorting includes removing duplicate information, processing missing values, labeling data categories (such as diseases, treatment plans, drug information, etc.), and converting data into a unified format for subsequent storage and retrieval.

[0087] After the data is organized, the cleaned data is then converted into a knowledge representation suitable for storage and retrieval. This process usually involves building a knowledge graph or database model to define entities, attributes, and relationships. For example, diseases, symptoms, treatment plans, etc. can be used as entities and connected through relationships to form a structured knowledge network. Using ontology or semantic network to describe these entities and their relationships can improve the scalability and flexibility of the knowledge base, thereby supporting complex queries and reasoning.

[0088] Finally, the constructed knowledge base needs to be stored in a suitable database management system for efficient retrieval and access. You can choose a relational database (such as MySQL, PostgreSQL) or a graph database (such as Neo4j), and the specific choice depends on the structure of the knowledge and the query requirements. In order to support dynamic retrieval, the system should implement an efficient query interface that allows users to search by keywords, topics, or entities. At the same time, the knowledge base should have an update mechanism to regularly add new medical knowledge and education materials to ensure the timeliness and accuracy of the information. Through these steps, the medical knowledge base can provide rich knowledge support for the interactive AI education system and enhance the patient's educational experience.

[0089] Preferably, the acoustic model is a CNN-LSTM combined network structure; and the language model is an N-gram model.

[0090] Specifically, the CNN-LSTM combined network structure combines the advantages of convolutional neural networks (CNN) and long short-term memory networks (LSTM) in the acoustic model to improve the accuracy and efficiency of speech recognition. First, CNN is used to extract local features of the input speech signal. Through multiple convolutional layers and pooling layers, CNN can effectively capture the time-frequency features in the speech signal, reduce the dimension of the input data and enhance the expressiveness of the features. Next, the extracted features are passed to the LSTM layer, which can process sequence data and capture the temporal dependencies and contextual information in the speech signal. Through this combination, CNN is responsible for extracting short-term features, while LSTM focuses on long-term dependencies, thereby achieving efficient modeling of speech signals and ultimately outputting the probability distribution of phonemes or words, providing a basis for subsequent decoding and recognition.

[0091] Furthermore, the N-gram model is a statistical method for language model construction and text generation. Its working process mainly includes two stages: training and application. In the training stage, the N-gram model constructs N-grams by analyzing a large amount of text data, counting the frequency of occurrence of N consecutive words, and calculating the probability of each N-gram. These probabilities reflect the possibility of the next word appearing given the previous N-1 words. In the application stage, when the system receives input text, the N-gram model predicts the next most likely word based on the trained probability distribution, thereby generating a coherent text sequence. In this way, the N-gram model can effectively capture local context information in the language, support text generation and speech recognition in natural language processing tasks, and improve the system's language understanding ability.

[0092] Preferably, the natural language processing module includes:

[0093] A text preprocessing unit, used for preprocessing the text information to obtain preprocessed text data;

[0094] An intention recognition unit, used for inputting the preprocessed text data into a trained intention recognition model to obtain the patient's intention;

[0095] An entity recognition unit, used for inputting the preprocessed text data into a trained named entity recognition model to obtain key entities;

[0096] A query unit, used to construct a query statement according to the patient's intention and the key entity, and execute a query in the medical knowledge base according to the query statement to obtain retrieval information;

[0097] The answer generation unit is used to generate the answer text according to the search information.

[0098] Optionally, the implementation of the natural language processing (NLP) module of this embodiment first involves preprocessing the patient's input text. This process includes removing extra spaces, punctuation marks, and special characters to ensure the neatness of the input data. Next, the system will perform word segmentation and part-of-speech tagging on the text, breaking the input text into analyzable basic units and labeling each unit with its grammatical role. These preprocessing steps lay the foundation for subsequent intent recognition and entity recognition, ensuring that the system can accurately understand the patient's inquiry.

[0099] After completing the text preprocessing, the NLP module will perform intent recognition and entity recognition. The purpose of intent recognition is to determine the type of patient inquiry, such as asking about the disease, treatment plan, or drug information. By training machine learning models (such as support vector machines, deep learning models, etc.), the system is able to extract key intents from the patient's input. At the same time, entity recognition focuses on identifying specific information in the text, such as disease names, symptoms, and drugs. By combining pre-trained models (such as BERT, SpaCy, etc.), the system can efficiently identify these entities.

[0100] Finally, the NLP module will retrieve relevant information from the medical knowledge base based on the identified intents and entities. The system will construct a query statement and use the knowledge base's retrieval interface to obtain relevant medical knowledge and educational materials. The retrieved information will be integrated and a natural language answer will be generated to ensure the accuracy and coherence of the answer. Through these steps, the NLP module can effectively understand the patient's inquiry and provide timely and accurate answers.

[0101] Specifically, the implementation steps of the propaganda content generation module of this embodiment specifically include:

[0102] First, the module of this embodiment receives the answer text generated by the natural language processing module, which is generated based on the patient's inquiry and the relevant information in the medical knowledge base. The answer text is analyzed to identify the key information and themes therein, including medical terms, the patient's specific needs and background information. According to the analysis results, a suitable template is selected from the preset content template library. These templates can be classified according to different medical themes, patient types and interaction scenarios to ensure that the generated content meets the needs of the patient. The answer text is combined with the selected template, and personalized propaganda content is dynamically generated using natural language generation technology (NLG). This process may involve reorganizing, expanding or simplifying the text to ensure the understandability and applicability of the content. The generated initial propaganda content is formatted to meet the display requirements of the front-end user interface, including text layout, image embedding, etc. The module supports real-time update function, which can dynamically adjust and optimize the generated propaganda content according to further inquiries or feedback from patients to enhance interactivity and user experience.

[0103] In this embodiment, the initial education content generated in real time can be customized according to the patient's specific problems and needs to provide personalized medical information. This flexibility enables patients to obtain more relevant and targeted answers, thereby improving patient satisfaction and understanding. The present invention can adjust and optimize the generated content in real time based on patient feedback and further inquiries. The dynamic adaptability of the present invention ensures the timeliness and accuracy of the information, and can reflect the latest medical knowledge and treatment plans in a timely manner. The present invention does not need to maintain a huge original education content library, reducing the complexity of content updates and management. The system can automatically generate the required content through medical knowledge base and natural language processing technology, reducing the need for manual intervention; by automatically generating content, the system can quickly respond to patients' inquiries and reduce waiting time. This efficient interactive method can enhance the patient's experience, especially in a medical environment, where timely information delivery is crucial. Exemplarily, the final education video of this embodiment can be in the format of ppt, voice, video, etc.

[0104] Through real-time generated content, patients can more actively participate in their own health management, enhancing their understanding and participation in the medical process. This sense of participation helps improve patient compliance and health management results.

[0105] The speech synthesis module converts the generated text answers into natural and fluent speech output to enhance the patient's interactive experience. First, the system receives the answer text generated by the natural language processing module and formats it to ensure that the text content is suitable for speech synthesis. This step may include spelling checking, punctuation completion, and adjustment of intonation and stress to improve the naturalness and comprehensibility of the speech. The speech synthesis module converts the formatted text into speech using text-to-speech (TTS) technology. Modern speech synthesis technology is usually based on deep learning models such as WaveNet or Tacotron, which can generate high-quality and natural speech output. The system selects appropriate voice features (such as gender, speaking speed, pitch, etc.) based on the user's preferences to provide a personalized voice experience. During the synthesis process, the system decomposes the text into phonemes and generates corresponding audio waveforms.

[0106] In addition, the specific implementation steps of the video education module of this embodiment are as follows:

[0107] During the interaction between the patient and the system, the video education module first needs to collect the patient's input information in real time, including voice or text content, the patient's emotional state, the topic of the question, etc. This information will help the system understand the patient's needs and background, thereby generating more targeted education videos.

[0108] Based on the collected contextual information and relevant data in the medical knowledge base, the video education module will use natural language processing technology to generate preliminary education content. This process includes extracting the patient's main questions, relevant medical knowledge and suggestions, and integrating this information into a coherent text framework as the basis for the video script.

[0109] After generating the preliminary educational content, the system will select appropriate images, animations, and video clips from the pre-prepared video material library based on the content's theme and the patient's needs. These materials should match the generated text content to ensure the accuracy and effectiveness of the information conveyed.

[0110] The video education module will use video editing technology to combine the selected footage with the generated text content to create a complete video. This process includes adding the speech output generated by the speech synthesis module to ensure that the speech in the video is synchronized with the visual content. In addition, the system can also add subtitles, graphics and other visual effects as needed to enhance the comprehensibility and appeal of the video.

[0111] Finally, the final educational video generated in this embodiment will be sent to the front-end user interface module for patients to watch. During the viewing process, the system can collect feedback information from patients in real time, such as viewing time, understanding, and satisfaction. These feedbacks will be used to optimize the content and form of subsequent videos to ensure that the system can be continuously improved to better meet the needs of patients.

[0112] Through the above steps, the video education module of this embodiment can effectively transform medical knowledge into visual content that is easy to understand and absorb, thereby improving the patient's interactive experience and learning effect.

[0113] The beneficial effects of the present invention are as follows:

[0114] (1) The present invention can provide personalized educational content based on the specific needs and problems of patients, thereby enhancing patients' sense of participation and satisfaction. The present invention supports both voice and text interaction modes to meet the preferences of different patients, and is especially more friendly to elderly patients or patients with language barriers.

[0115] (2) Through speech recognition and natural language processing technology, the present invention enables patients to quickly obtain the information they need, reduce waiting time, and improve the efficiency of information acquisition; the dynamic retrieval function of the medical knowledge base ensures that patients can obtain the latest medical information and educational materials, thereby improving the timeliness of information.

[0116] (3) The present invention provides information through a standardized medical knowledge base to ensure the accuracy and consistency of the educational content and reduce misunderstandings caused by information asymmetry; combined with speech synthesis technology, the system can convert text information into speech output, which is suitable for patients with different learning styles and enhances the effect of knowledge dissemination.

[0117] (4) The present invention uses an automated education system, allowing medical staff to spend more time on personalized care of patients and handling of complex problems, thereby improving work efficiency; the present invention can handle the education of common problems and basic knowledge, reducing the repetitive work of medical staff in these areas.

[0118] (5) The medical knowledge base of the present invention can be updated according to the latest research results and clinical practices to ensure that the information obtained by patients is always the latest; the present invention can collect feedback from patients, help medical institutions identify deficiencies in the knowledge base, and promote continuous improvement of knowledge.

[0119] (6) The present invention can record patients' inquiries and feedback, provide data support for medical institutions, help analyze patients' needs and common problems, and optimize medical services; by analyzing patient data, medical institutions can better formulate education strategies and improve medical services.

[0120] (7) The present invention can be applied in telemedicine scenarios, so that patients can obtain professional medical knowledge and education services at home, thereby improving the accessibility of medical services.

[0121] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0122] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. An interactive AI education system, characterized in that: include: The front-end user interface module is used to build a friendly user interface to display educational content and support patients to interact with the system through voice or text; A speech recognition module, used to convert the patient's speech input signal into text information; Medical knowledge base construction module, used to build a medical knowledge base containing medical knowledge and educational materials to support dynamic retrieval and content generation; A natural language processing module, for understanding the patient's inquiry and intention based on the text information, and retrieving relevant information from the medical knowledge base to generate a response text; A propaganda content generation module, used to generate initial propaganda content in real time according to the answer text based on a preset artificial intelligence decision-making body; A speech synthesis module, used for converting the answer text into speech output based on speech synthesis technology; The video education module is used to combine the context of the patient's interaction process, generate a final education video in real time according to the initial education content and the voice output, and send the final education video to the front-end user interface module to enhance the patient's interaction experience.

2. The interactive AI education system according to claim 1, characterized in that: The speech recognition module comprises: A speech denoising unit, used to perform denoising on the collected speech input signal to obtain a denoised signal; A normalization unit, used for performing normalization processing on the denoised signal to obtain a normalized signal; A framing unit, used for performing framing processing on the normalized signal to obtain individual frame signals; A feature extraction unit, used for extracting a speech feature vector from each of the frame signals; An acoustic model processing unit, configured to determine the probability distribution of phonemes or words according to the speech feature vector based on a preset acoustic model; A language model processing unit, used to determine a text sequence according to a probability distribution of phonemes or words based on a preset language model; The post-processing unit is used to perform spelling check, punctuation completion and formatting processing on the generated text sequence to obtain the text information.

3. The interactive AI education system according to claim 2, characterized in that: The speech denoising unit comprises: A signal translation subunit, used for performing translation on the voice input signal to obtain a translated collected signal; A wavelet decomposition subunit, used for performing wavelet decomposition on the translated collected signal to obtain a plurality of wavelet coefficients; A threshold determination subunit, used to determine a filtering threshold according to a decomposition scale and length of a signal; A function determination subunit, used for constructing a denoising function according to the filtering threshold; The denoising subunit is used to remove the noise of the translated collected signal by using the denoising function to obtain the denoised signal.

4. The interactive AI education system according to claim 3, characterized in that: The filtering thresholds include: Among them, λ is the filtering threshold, j is the decomposition scale of the NMR acquisition signal after translation, and d j is the decomposition of the wavelet coefficients with a scale of j, N represents the length of the NMR acquisition signal after translation, and median represents the median operation.

5. The interactive AI education system according to claim 4, characterized in that: The denoising subunit comprises: A smoothing component is used to remove the corresponding wavelet coefficients using the denoising function to obtain a smoothed collected signal; wherein the denoising function is: Among them, X represents the wavelet coefficient, n represents the adjustment coefficient; An inverse translation component, used for inversely translating the smoothed collected signal to obtain an inversely translated collected signal; The denoising component is used to obtain the average of multiple inverse-translated collected signals to obtain the denoised signal.

6. The interactive AI education system according to claim 2, characterized in that: The acoustic model is a CNN-LSTM combined network structure.

7. The interactive AI education system according to claim 2, characterized in that: The language model is an N-gram model.

8. The interactive AI education system according to claim 1, characterized in that: The natural language processing module includes: A text preprocessing unit, used for preprocessing the text information to obtain preprocessed text data; An intention recognition unit, used for inputting the preprocessed text data into a trained intention recognition model to obtain the patient's intention; An entity recognition unit, used for inputting the preprocessed text data into a trained named entity recognition model to obtain key entities; A query unit, used to construct a query statement according to the patient's intention and the key entity, and execute a query in the medical knowledge base according to the query statement to obtain retrieval information; The answer generation unit is used to generate the answer text according to the search information.

9. The interactive AI education system according to claim 1, characterized in that: The education content generation module includes: A content analysis unit, configured to analyze the answer text to identify key information of the answer text; the key information includes: medical terms, specific needs of the patient and background information; A template selection unit, used to select a suitable template from a preset content template library according to the analysis results; the templates are classified according to different medical topics, patient types and interaction scenarios to ensure that the generated content meets the needs of patients; A content generation unit, used to combine the answer text with the selected template and dynamically generate personalized initial propaganda content using natural language generation technology; The formatted data unit is used to format the personalized initial education content to meet the display requirements of the front-end user interface.

Images