Intelligent medical assistant system based on large language model

Abstract

The invention discloses an intelligent medical assistant system based on a large language model, which relates to the technical field of medical assistants and comprises an information management module, an intelligent inquiry module, an identification analysis module and a personalized feedback module. The information management module comprises a reading unit and a classification unit, the reading unit integrates a medical card reading device and the OCR technology, when a patient uses a medical assistant and inserts a medical card, medical card information of the patient is recognized and read, specifically, the medical card information comprises the age, the gender and the past medical history of the patient, and the medical card information is transmitted to the classification unit; through the recognition and analysis module and the personalized feedback module, the recognition and response ability of the system to the complex emotion appeal of the patient is remarkably improved, the problem that an existing medical assistant cannot recognize the emotion appeal uncertainty of the patient in time and is inaccurate is solved, the recognition and analysis module captures voice and text input of the patient in real time, and the recognition and response ability of the system to the emotion appeal uncertainty of the patient is improved. And emotion states such as anxiety and fear can be accurately identified.

Description

Technical Field

[0001] The present invention relates to the field of medical assistant technology, and more specifically, to an intelligent medical assistant system based on a large language model. Background Art

[0002] With the improvement of people's health awareness and the continuous advancement of medical technology, communication and exchanges between doctors and patients have become more and more important. Traditional medical consultation methods have many limitations, such as limited time for face-to-face communication between patients and doctors, and geographical restrictions that make it difficult for some patients to obtain professional medical advice in a timely manner. Doctors also face tremendous pressure in their busy work and find it difficult to fully communicate and guide every patient. In order to effectively solve these problems, intelligent medical assistants have now emerged; Intelligent medical assistants are based on large language models and have powerful natural language understanding and generation capabilities. They can process and analyze large amounts of text data. They can not only provide patients with convenient online medical consultation services, help them gain a preliminary understanding of their conditions, and reduce unnecessary hospital visits, but can also serve as guides in hospitals, helping patients understand the distribution of hospital departments, registration procedures, and other information, thereby improving hospital service efficiency. However, in actual use, hospital patients are diverse and complex. The age, gender, constitution, genetic background, past medical history, living habits and other factors of different patients will affect the pathogenesis, symptoms and treatment effects of the disease. The system is often unable to accurately identify and adapt to these differences, resulting in insufficient personalization of the medical information and suggestions provided, and unable to meet the specific needs of different patients. In the face of complex disease conditions, such as coexistence of multiple diseases, rare diseases, and rapid changes in disease conditions, the processing capabilities of single-mode intelligent medical assistants are limited; the system may find it difficult to accurately identify and analyze complex pathophysiological processes, as well as the mutual influence between different diseases, thus affecting the accuracy of diagnosis and the rationality of treatment plans; During the consultation process, in addition to their concern about their condition, patients often also have various complex emotional demands, such as anxiety, fear, loneliness, helplessness, etc. However, existing medical assistants tend to focus more on providing medical information and advice. Patients' questions and anxiety are random and unpredictable, and may change at any time due to changes in their condition, environmental factors, psychological state, etc. When faced with this uncertainty, smart medical assistants are unable to promptly and accurately identify the exact demands of medical patients. Summary of the invention

[0003] To solve the above problems, the present invention provides an intelligent medical assistant system based on a large language model.

[0004] The present invention provides an intelligent medical assistant system based on a large language model, including an information management module, an intelligent consultation module, a recognition and analysis module, and a personalized feedback module; The information management module includes a reading unit and a classification unit. The reading unit integrates a medical card reading device and OCR technology. When a patient uses a medical assistant and inserts a medical card, the medical card information of the patient is identified and read, including the patient's age, gender and medical history, and the medical card information is transmitted to the classification unit. After receiving the medical card information from the reading unit, the classification unit classifies the patient according to the read medical card information, outputs the classification result, and transmits it to the intelligent consultation module; The intelligent consultation module includes a switching unit, an analysis and diagnosis unit, and a condition processing unit. The switching unit is used to receive the classification result of the classification unit, and switch the intelligent consultation module to a corresponding communication mode according to the classification result of the classification unit; The analysis and diagnosis unit is equipped with a large language model, which is used to receive the medical card information from the reading unit, and analyze the patient's condition according to the medical card information, determine the complexity of the patient's condition, and output the determination result to the condition processing unit; The condition processing unit is used to receive the judgment result of the analysis and diagnosis unit and the classification result of the classification unit, and to perform secondary classification on the patient according to the judgment result and the classification result, and to generate corresponding diagnosis suggestions according to the result of the secondary classification, and transmit them to the identification and analysis module and the display screen of the intelligent medical assistant device; The recognition and analysis module is used to receive input information from the patient, specifically including voice input or text input, and recognize the patient's emotional state information in real time according to the patient's input information, and transmit it; The personalized feedback module is used to receive the patient's emotional state information, generate real-time response information based on the patient's emotional state information, and transmit the response information to the condition processing unit to correct the diagnosis suggestions generated by the condition processing unit.

[0005] Preferably, the specific working mode of the reading unit is as follows: When using the smart medical assistant, the patient first inserts the medical card into the card slot of the smart medical assistant device; The smart medical assistant device uses built-in optical character recognition technology to recognize the information on the medical card; Identify and read the patient's medical card information, and transmit the medical card information to the classification unit.

[0006] Preferably, the specific working steps of the classification unit are as follows: Receiving the medical card information read by the reading unit; If the patient's age is between 0 and 12 years old, it is marked as a patient in category 1; If the patient is aged between 13 and 18 years, they are marked as Category II patients; If the patient’s age is between 19 and 60 years old, they are marked as category three patients; If the patient’s age is over 60 years old, they are marked as category 4 patients; The patient's marking result is transmitted to the intelligent consultation module as a classification result.

[0007] Preferably, the specific working steps of the switching unit are as follows: Formulate a corresponding communication mode according to the classification result of the classification unit; For patients in the first category, the font size of the smart medical assistant device was set to 18, and the output speech speed was controlled at 120 words per minute; For patients in category II, the font size of the smart medical assistant device was set to 16, and the output speech speed was controlled at 160 words per minute; For the third category of patients, the font size of the smart medical assistant device was set to 14, and the output speech speed was controlled at 180 words per minute; For the four types of patients, the font size of the smart medical assistant device was set to 22, and the output speech speed was controlled at 110 words per minute; The classification result of the classification unit is received, and a corresponding communication mode is selected according to different classification results.

[0008] Preferably, the specific working steps of the analysis and diagnosis unit are as follows: Determining a large language model built into the analysis and diagnosis unit; The analysis and diagnosis unit first receives the medical card information transmitted by the reading unit; According to the built-in large language model, the read medical card information is analyzed to obtain the judgment result of the patient's condition; The specific output judgment results of the patient's condition include mild results, moderate results and severe results, and the judgment results are output to the corresponding condition processing unit.

[0009] Preferably, the specific working steps of the condition processing unit are as follows: First, receiving the judgment result of the analysis and diagnosis unit and the classification result of the classification unit, and performing secondary classification on the patient according to the judgment result and the classification result to obtain a secondary classification result, wherein the secondary classification result specifically includes a first-class management requirement, a second-class management requirement, a third-class management requirement and a fourth-class management requirement; According to the secondary classification results, corresponding diagnostic suggestions are generated and transmitted to the identification and analysis module.

[0010] Preferably, the specific working steps of performing secondary classification of patients according to the judgment result and the classification result to obtain the secondary classification result are as follows; For patients in category 1, category 2, and category 3, all patients whose results are judged to be mild are marked as category 1 management needs; For patients in category 1, category 2, all patients in category 3 with moderate results, and patients in category 4 with mild results, they are marked as category 2 management needs; For patients in category 1, category 2, and all patients in category 3 who are judged to have severe results, as well as patients in category 4 who have moderate results, they are marked as category 3 management needs; For severe results in the four categories of patients, four categories of management needs were marked; The marked results are taken as the secondary classification results.

[0011] Preferably, the specific working steps of generating corresponding diagnostic suggestions according to the secondary classification results are as follows; For patients with type 1 management needs, patients are advised to perform self-care at home; For patients with Class II management needs, it is recommended that they go to a specialist clinic for disease assessment and treatment; For patients with the third type of management needs, contact the relevant departments of the hospital including the emergency department, intensive care unit and operating room in advance to inform the patient's condition and arrival time; For patients with category 4 management needs, based on the advice for patients with category 3 management needs, it is recommended that patients go to the hospital's nursing department.

[0012] Preferably, the specific working steps of the identification and analysis module are as follows: Obtain text information input by the patient; Obtaining the speech rate R input by the patient, the average strength of the patient's voice, the number of texts input by the patient, and the number of emotional words E in the text input by the patient; The speech rate R input by the patient, the average strength V of the patient's voice, the number of texts input by the patient T, and the number of emotional words E in the text input by the patient are standardized to obtain the standardized speech rate R1 input by the patient, the average strength V1 of the patient's voice, the number of texts T1 input by the patient, and the number of emotional words E1 in the text input by the patient; According to the formula , calculate and obtain the total score of the patient's negative emotions ; According to the formula , calculate and obtain the total score of the patient's positive emotions ; According to the formula 100, to get the total score of the patient's emotional state , the total score of the patient's emotional state is transmitted as the patient's emotional state information.

[0013] Preferably, the specific working steps of the personalized feedback module are as follows: Get the total score of the patient's emotional state , if 0≤S3<50, it means that the patient is very calm; If 50≤S3<100, it means the patient is very calm; If 100≤S3<150, it means the patient is emotionally agitated; If 150≤S3≤200, it means the patient is very emotional; Generate real-time response information based on the patient's emotional state information; If the patient's emotional state is very calm, no special adjustments are needed in the diagnostic recommendations; If the patient's emotional state is calm: further confirmation of the patient's emotions is added to the diagnostic recommendations. If the patient's emotional state is agitated: add emotional soothing measures to the diagnostic recommendations; If the patient's emotional state is very agitated, the information will be transmitted to the medical staff's mobile terminal for further face-to-face communication.

[0014] Beneficial effects: Through the information management module and the intelligent consultation module, it is possible to achieve accurate matching of the patient's personalized medical information needs, effectively solving the problem of insufficient personalization of medical information and suggestions due to differences in patient age, gender, physical constitution and other factors. The reading unit and classification unit of the information management module work together to fully obtain and classify the patient's personal information, and provide detailed patient background data for the intelligent consultation module; the intelligent consultation module flexibly switches the communication mode according to the classification results, and combines the in-depth condition analysis of the analysis and diagnosis unit to generate diagnostic suggestions that are highly consistent with the patient's specific situation, ensuring that each patient can obtain tailored medical services to meet their specific needs, thereby improving the personalization level of medical services and patient satisfaction; Through the recognition and analysis module and the personalized feedback module, the system's ability to recognize and respond to patients' complex emotional demands has been significantly improved, solving the problem of untimely and inaccurate recognition of existing medical assistants when facing the uncertainty of patients' emotional demands. The recognition and analysis module captures the patient's voice and text input in real time, accurately identifies emotional states such as anxiety and fear, and transmits emotional information to the personalized feedback module. The unit quickly generates targeted emotional responses and personalized feedback, promptly meets the patient's emotional needs, and relieves their psychological pressure. The feedback information also corrects the diagnosis suggestions of the condition treatment unit, making the medical suggestions more comprehensive and humane, taking into account the patient's mental health, enhancing the patient's trust and dependence on the system, and providing patients with a better medical experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0016] Application scenarios: In actual use, the patients in hospitals are diverse and complex. The age, gender, physique, genetic background, medical history, living habits and other factors of different patients will affect the pathogenesis, symptoms and treatment effects of the disease. The system is often unable to accurately identify and adapt to these differences, resulting in the medical information and suggestions provided are not personalized enough to meet the specific needs of different patients. In the face of complex disease conditions, such as coexistence of multiple diseases, rare diseases, and rapid changes in disease conditions, the processing capabilities of single-mode intelligent medical assistants are limited; the system may find it difficult to accurately identify and analyze complex pathophysiological processes, as well as the mutual influence between different diseases, thereby affecting the accuracy of diagnosis and the rationality of treatment plans.

[0017] like Figure 1 As shown: An intelligent medical assistant system based on a large language model, including an information management module, an intelligent consultation module, a recognition and analysis module, and a personalized feedback module; The information management module includes a reading unit and a classification unit. The reading unit integrates a medical card reading device and OCR technology. When a patient uses a medical assistant and inserts a medical card, the medical card information of the patient is identified and read, including the patient's age, gender and medical history, and the medical card information is transmitted to the classification unit. After receiving the medical card information from the reading unit, the classification unit classifies the patient according to the read medical card information, outputs the classification result, and transmits it to the intelligent consultation module; it should be noted that the intelligent consultation module has multiple communication modes built in, and can switch to the corresponding communication mode according to the classification result of the classification unit to better meet the communication needs of different patients and improve communication efficiency and patient satisfaction; The intelligent consultation module includes a switching unit, an analysis and diagnosis unit, and a condition processing unit. The switching unit is used to receive the classification result of the classification unit, and switch the intelligent consultation module to a corresponding communication mode according to the classification result of the classification unit; The analysis and diagnosis unit is equipped with a large language model, which is used to receive the medical card information from the reading unit, and analyze the patient's condition according to the medical card information, determine the complexity of the patient's condition, and output the determination result to the condition processing unit; The condition processing unit is used to receive the judgment result of the analysis and diagnosis unit and the classification result of the classification unit, and to perform secondary classification on the patient according to the judgment result and the classification result, and to generate corresponding diagnosis suggestions according to the result of the secondary classification, and transmit them to the identification and analysis module and the display screen of the intelligent medical assistant device; The recognition and analysis module is used to receive the patient's input information, specifically including voice input or text input, and recognize the patient's emotional state information in real time according to the patient's input information, and transmit it; it should be noted that after the patient receives the diagnosis suggestion on the display screen of the smart medical assistant device, if he has any questions or other needs, he will input them into the smart medical assistant device; The personalized feedback module is used to receive the patient's emotional state information, generate real-time response information based on the patient's emotional state information, and transmit the response information to the condition processing unit to correct the diagnosis suggestions generated by the condition processing unit.

[0018] It should be noted that the information management module is responsible for reading the patient's medical card information, including age, gender, past medical history, etc., and transmits this information to the classification unit through OCR technology and medical card reading equipment. The classification unit classifies the patient according to the information and outputs the classification results, laying the foundation for subsequent services; The intelligent consultation module receives classification results and medical card information, has multiple built-in communication modes, switches modes according to patient classification, conducts in-depth analysis of the patient's condition through the switching unit, analysis and diagnosis unit, and condition processing unit, determines the complexity, generates personalized diagnosis suggestions, and assists doctors in improving diagnostic efficiency and accuracy; The recognition and analysis module receives the patient's voice or text input, identifies the emotional state in real time, and transmits it to the personalized feedback module. This unit generates real-time responses and personalized feedback information based on the emotional state, and transmits it to the condition treatment unit to correct the diagnosis suggestions. At the same time, it provides emotional support and psychological counseling for patients, relieves emotional pressure, improves patients' mental health and satisfaction, enhances their trust and reliance on the system, and fully meets patients' medical needs; It should also be noted that the information management module and the intelligent consultation module can achieve accurate matching of patients' personalized medical information needs, effectively solving the problem of insufficient personalization of medical information and suggestions due to differences in patient age, gender, physical constitution and other factors. The reading unit and classification unit of the information management module work together to comprehensively obtain and classify patients' personal information, and provide detailed patient background data for the intelligent consultation module; the intelligent consultation module flexibly switches the communication mode according to the classification results, and combines the in-depth condition analysis of the analysis and diagnosis unit to generate diagnostic suggestions that are highly consistent with the patient's specific situation, ensuring that each patient can obtain tailored medical services to meet their specific needs, thereby improving the personalization level of medical services and patient satisfaction; Through the recognition and analysis module and the personalized feedback module, the system's ability to recognize and respond to patients' complex emotional demands has been significantly improved, solving the problem of untimely and inaccurate recognition of existing medical assistants when facing the uncertainty of patients' emotional demands. The recognition and analysis module captures the patient's voice and text input in real time, accurately identifies emotional states such as anxiety and fear, and transmits the emotional information to the personalized feedback module. The unit quickly generates targeted emotional responses and personalized feedback to meet the patient's emotional needs in a timely manner and relieve their psychological pressure. The feedback information also corrects the diagnostic recommendations of the disease treatment unit, making the medical recommendations more comprehensive and humane, taking into account the patient's mental health, enhancing the patient's trust and dependence on the system, and providing patients with a better medical experience; As a further embodiment, the specific working mode of the reading unit is as follows: When using the smart medical assistant, the patient first inserts the medical card into the card slot of the smart medical assistant device. It should be noted that modern hospitals will equip patients with a dedicated medical card, which stores the patient's identity information and historical medical records. The smart medical assistant device is generally placed in the lobby on the first floor of the hospital. When using it, the patient only needs to insert the medical card into the card slot on the smart medical assistant device to start using it. The smart medical assistant device uses built-in optical character recognition technology to identify the information on the medical card. It should be noted that OCR (optical character recognition) technology can convert text, numbers and other information on the card into electronic data format. As a common existing technology, the device will identify the information on the card line by line or block by block, such as name, ID number, age, gender, medical history and other key data; Identify and read the patient's medical card information, and transmit the medical card information to the classification unit.

[0019] As a further embodiment, the specific working steps of the classification unit are as follows: Receiving the medical card information read by the reading unit; If the patient's age is between 0 and 12 years old, it is marked as a patient in category 1; If the patient is aged between 13 and 18 years, they are marked as Category II patients; If the patient’s age is between 19 and 60 years old, they are marked as category three patients; If the patient’s age is over 60 years old, they are marked as category 4 patients; The patient's marking result is transmitted to the intelligent consultation module as a classification result.

[0020] It should be noted that the classification unit needs to receive the medical card information transmitted by the reading unit, including the patient's age, gender, medical history and other personal information, so as to carry out subsequent classification processing, in order to ensure that the system can obtain complete basic information of the patient and provide data support for personalized services; Patients of different age groups have significant differences in communication methods, information acceptance capabilities and psychological needs, and preliminary screening is required to better adapt to these differences; patients are classified by age grouping so that the system can select the most appropriate communication mode according to different age groups.

[0021] As a further embodiment, the specific working steps of the switching unit are as follows: Formulate a corresponding communication mode according to the classification result of the classification unit; For the first type of patients, the font size of the smart medical assistant device is set to 18, and the output speech speed is controlled at 120 words per minute. It should be noted that children's vision is more sensitive than that of adults, and larger fonts help them read information better. Children's comprehension ability and language response speed are relatively slow, and a slower speech speed can give them more time to understand the content being conveyed. It should also be noted that the output speech speed is the speed at which the smart medical assistant device outputs fonts or voices. For the second type of patients, the font size of the smart medical assistant device is set to 16, and the output speech speed is controlled at 160 words per minute. It should be noted that the vision and cognitive ability of teenagers are gradually approaching that of adults. A medium font size can ensure reading comfort without appearing too childish. Patients in this age group have a strong comprehension ability. A moderate speech speed can ensure the efficiency of information transmission without making them feel that it is too fast and difficult to understand. For the three types of patients, the font size of the smart medical assistant device is set to 14, and the output speech speed is controlled at 180 words per minute. It should be noted that adults have a better vision and cognitive ability, and regular font size can meet daily reading needs. Adults have a faster understanding and reaction speed, and normal-speed speech output can efficiently complete information communication. For the four types of patients, the font size of the smart medical assistant device is set to 22, and the output speech speed is controlled at 110 words per minute. It should be noted that the eyesight of the elderly generally declines, and larger fonts help them see information more clearly and reduce visual fatigue. The hearing and comprehension abilities of the elderly may decline, and a slower speech speed allows them to better hear and understand the content being conveyed. The classification result of the classification unit is received, and a corresponding communication mode is selected according to different classification results.

[0022] It should be noted that patients of different ages have their own characteristics and needs both physically and psychologically. The purpose of classified switching of communication modes is to better adapt to these differences and match the communication method with the patient's actual situation, so that patients can have a smoother medical treatment process. Through accurate classification and targeted switching of communication modes, misunderstandings and information omissions caused by poor communication can be reduced, so that patients can fully understand medical information.

[0023] As a further embodiment, the specific working steps of the analysis and diagnosis unit are as follows: Determine the built-in big language model of the analysis and diagnosis unit; it should be noted that commonly used medical big language models include but are not limited to: BioNeMo™: developed by NVIDIA, specifically for analysis and prediction of biomedical data; ChatOpenAI: Based on OpenAI's GPT model, fine-tuned for the medical field, suitable for medical text analysis; The analysis and diagnosis unit first receives the medical card information transmitted by the reading unit; it should be noted that this information includes the patient's age, gender, medical history and current symptoms, etc. This information will be used as input data for subsequent condition analysis; According to the built-in large language model, the read medical card information is analyzed to obtain the judgment result of the patient's condition; it should be noted that the patient's past medical history and current symptoms can be deeply analyzed. For example, using BioNeMo™'s protein structure prediction model ESMFold or protein language model ESM2, disease-related proteins can be analyzed to help judge the complexity of the disease; ChatOpenAI can perform text analysis on the patient's medical card information. Through dialogue generation, tool calling and other functions, ChatOpenAI can understand the patient's symptom description and generate a preliminary judgment on the condition in combination with the medical knowledge base. The specific output judgment results of the patient's condition include mild results, moderate results and severe results, and the judgment results are output to the corresponding condition processing unit.

[0024] It should be noted that the specific judgment results are classified into: Mild results: Disease type: common diseases, such as the common cold, mild gastroenteritis, mild skin allergies, etc., duration: the disease lasts for a short time, usually within 1 week, symptoms and signs: the symptoms are mild and do not affect daily life and work. No hospitalization is required and can be treated with conventional medication or rest and rehabilitation.

[0025] Moderate results: Disease type: moderate disease, such as moderate asthma, moderate hypertension, moderate diabetes, etc., duration: the disease lasts between 1 week and 1 month, symptoms and signs: the symptoms are more obvious and may require hospitalization, close monitoring and special treatment. Patients may need regular review and adjustment of treatment plans; Severe results: Disease type: serious disease, such as acute myocardial infarction, severe pneumonia, severe stroke, etc., duration: the disease exists for a long time, usually more than 1 month, or the condition deteriorates rapidly, symptoms and signs: severe symptoms, seriously affecting daily life and work, requiring emergency hospitalization, and may require intensive care and multidisciplinary consultation.

[0026] As a further embodiment, the specific working steps of the condition processing unit are as follows: First, receiving the judgment result of the analysis and diagnosis unit and the classification result of the classification unit, and performing secondary classification on the patient according to the judgment result and the classification result to obtain a secondary classification result, wherein the secondary classification result specifically includes a first-class management requirement, a second-class management requirement, a third-class management requirement and a fourth-class management requirement; According to the secondary classification results, corresponding diagnostic suggestions are generated and transmitted to the identification and analysis module.

[0027] It should be noted that by carefully classifying patients according to age and severity of illness, this method can more accurately identify the specific needs of different patients, thereby providing highly personalized medical advice and treatment plans. It not only helps to rationally allocate medical resources and ensure that emergency and critically ill patients receive timely treatment, but also prevents the deterioration of mild and moderate conditions and reduces complications through early intervention.

[0028] As a further embodiment, the specific working steps of performing secondary classification of the patient according to the judgment result and the classification result to obtain the secondary classification result are as follows; For patients in category 1, category 2, and category 3, all patients whose results are judged to be mild are marked as category 1 management needs; For patients in category 1, category 2, all patients in category 3 with moderate results, and patients in category 4 with mild results, they are marked as category 2 management needs; For patients in category 1, category 2, and all patients in category 3 who are judged to have severe results, as well as patients in category 4 who have moderate results, they are marked as category 3 management needs; For the severe results among the four categories of patients, they are marked as four categories of management needs; it should be noted that the first category of management needs is routine management needs, the second category of management needs is specialist treatment needs, the third category of management needs is emergency treatment needs, and the fourth category of management needs is supportive care needs; The marked results are taken as the secondary classification results.

[0029] As a further embodiment, the specific working steps of generating corresponding diagnostic suggestions according to the secondary classification results are as follows; For patients with Class I management needs, it is recommended that they perform self-care at home, including rest, drinking plenty of water, local cleaning and disinfection, etc. If symptoms do not improve within 48 hours or show signs of worsening, seek medical attention in a timely manner. Use of over-the-counter drugs: Common over-the-counter drugs, such as antipyretics and analgesics (acetaminophen, ibuprofen, etc.), may be used as appropriate, but the applicable age and dosage guidelines must be strictly followed. Lifestyle adjustments: Patients are encouraged to maintain a reasonable diet, moderate exercise, adequate sleep, and avoid bad habits such as smoking and excessive drinking. Regular health check-ups: Especially for patients with a family history of chronic diseases or bad lifestyle habits, it is recommended to undergo a comprehensive physical examination every 6 months to 1 year to detect potential health problems early. For patients with Class II management needs, it is recommended that they go to a specialist clinic for disease assessment and treatment; the frequency of follow-up visits depends on the specific condition, and it is generally recommended to follow up every 1-3 months. Drug treatment compliance: patients should strictly follow the doctor's prescription, take the medicine on time and in the right amount, and pay close attention to the possible side effects of the medicine. If any discomfort symptoms occur, they need to communicate with the doctor in time. Psychological support and health education: patients are encouraged to participate in psychological support groups or consult psychologists to enhance their confidence in treatment and improve their self-management ability, so as to better cope with the psychological pressure brought by the disease. Participation in rehabilitation training: For patients who need rehabilitation training, it is recommended to go to a professional rehabilitation institution for training, or to exercise at home according to rehabilitation guidance to promote the recovery of physical functions; For patients with the third type of management needs, contact the relevant departments of the hospital including the emergency department, intensive care unit and operating room in advance to inform the patient's condition and arrival time; so that the hospital can make full preparations for receiving the patient and ensure that the patient can quickly receive further diagnosis and treatment after arriving at the hospital; Advance notification of special conditions: If the patient has a history of severe allergies or other special diseases, be sure to inform the emergency personnel and the attending doctor in advance so that they can prepare targeted response measures in advance to avoid possible medical risks; For patients with the fourth category of management needs, on the premise of the recommendations for patients with the third category of management needs, it is recommended that patients go to the hospital's special nursing department. It should be noted that a professional nursing team may be arranged to provide door-to-door services regularly, and a personalized supportive care plan will be formulated based on the patient's overall condition, covering pain management, nutritional support, psychological comfort and other aspects to improve the patient's living comfort.

[0030] As a further embodiment, the specific working steps of the identification and analysis module are as follows: Obtaining text information input by the patient; it should be noted that the voice input by the patient is recognized and converted into text information; The speech rate R input by the patient, the average strength of the patient's voice, the number of texts input by the patient, and the number of emotional words E in the text input by the patient are obtained; it should be noted that the speech rate, that is, the number of words spoken per minute, the voice level, that is, the average strength of the voice, can be expressed in decibels (dB), the number of texts, that is, the number of words in the input text, and the number of emotional words, that is, the number of times emotion-related words appear in the text; The speech rate R input by the patient, the average strength V of the patient's voice, the number of texts input by the patient T, and the number of emotional words E in the text input by the patient are standardized to obtain the standardized speech rate R1 input by the patient, the average strength V1 of the patient's voice, the number of texts input by the patient T1, and the number of emotional words E1 in the text input by the patient; it should be noted that, in this embodiment, the standardization processing method is to divide the difference between the feature value of the current patient input text and the historical minimum value by the difference between the historical maximum value and the historical minimum value; According to the formula , calculate and obtain the total score of the patient's negative emotions ; It should be noted that high values ​​of speech rate and voice intensity usually indicate excitement or tension, and high values ​​of the number of emotional words usually indicate negative emotions. Therefore, taking these factors into consideration can more accurately assess the patient's negative emotions; According to the formula , calculate and obtain the total score of the patient's positive emotions ;It should be noted that a high value of the number of texts usually indicates that the patient describes his or her situation in detail, and a low value of the number of emotional words usually indicates positive emotions. Therefore, taking these factors into consideration can more accurately assess the patient's positive emotions; According to the formula 100, to get the total score of the patient's emotional state , the total score of the patient's emotional state is transmitted as the patient's emotional state information. It should be noted that by subtracting the total score of positive emotions from the total score of negative emotions and adding 100 for correction, a comprehensive score of the emotional state can be obtained; It should also be noted that by comprehensively considering speech rate, speech intensity, amount of text, and number of emotional words, a more comprehensive assessment of the patient's emotional state can be achieved.

[0031] As a further embodiment, the specific working steps of the personalized feedback module are as follows: Get the total score of the patient's emotional state , if 0≤S3<50, it means that the patient is very calm; If 50≤S3<100, it means the patient is very calm; If 100≤S3<150, it means the patient is emotionally agitated; If 150≤S3≤200, it means the patient is very emotional; Generate real-time response information based on the patient's emotional state information; it should be noted that the specific response information is very calm: Response message: Thank you for your patience and detailed description. If you have any questions or need further help, please feel free to let me know; Calm: Response message: Thank you for your description. If you have any questions or need further assistance, please feel free to let me know; Agitation: Response message: I understand you may be feeling a little agitated. Please try to calm down and I will do my best to help you resolve the issue; Very excited: Response message: I see you may be very excited. Take a deep breath. I am here to help you. We can work through this together. If the patient's emotional state is very calm, no special adjustments are needed in the diagnostic recommendations; If the patient's emotional state is calm: further confirmation of the patient's emotions is added to the diagnostic recommendations. If the patient's emotional state is agitated: add emotional soothing measures to the diagnostic recommendations; emotional soothing measures include further asking the patient about their specific concerns to ensure the comprehensiveness of the diagnosis; If the patient's emotional state is very excited, the information will be transmitted to the medical staff's mobile terminal for further face-to-face communication. Ensure that the patient's emotions are properly handled and obtain more accurate information about the condition. It should be noted that through the above steps, the intelligent medical assistant device can generate real-time response information and personalized feedback information based on the patient's emotional state, so as to better understand the patient's emotional state, and revise the diagnostic recommendations to improve the quality of medical services.

[0032] Working principle: The recognition and analysis module receives the patient's voice or text input, identifies the emotional state in real time, and transmits it to the personalized feedback module. This unit generates real-time responses and personalized feedback information based on the emotional state, and transmits it to the condition treatment unit to correct the diagnosis suggestions. At the same time, it provides emotional support and psychological counseling for patients, relieves emotional pressure, improves patients' mental health and satisfaction, enhances their trust and reliance on the system, and fully meets patients' medical needs; It should also be noted that the information management module and the intelligent consultation module can achieve accurate matching of patients' personalized medical information needs, effectively solving the problem of insufficient personalization of medical information and suggestions due to differences in patient age, gender, physical constitution and other factors. The reading unit and classification unit of the information management module work together to comprehensively obtain and classify patients' personal information, and provide detailed patient background data for the intelligent consultation module; the intelligent consultation module flexibly switches the communication mode according to the classification results, and combines the in-depth condition analysis of the analysis and diagnosis unit to generate diagnostic suggestions that are highly consistent with the patient's specific situation, ensuring that each patient can obtain tailored medical services to meet their specific needs, thereby improving the personalization level of medical services and patient satisfaction; Through the recognition and analysis module and the personalized feedback module, the system's ability to recognize and respond to patients' complex emotional demands has been significantly improved, solving the problem of untimely and inaccurate recognition of existing medical assistants when facing the uncertainty of patients' emotional demands. The recognition and analysis module captures the patient's voice and text input in real time, accurately identifies emotional states such as anxiety and fear, and transmits emotional information to the personalized feedback module. The unit quickly generates targeted emotional responses and personalized feedback, promptly meets the patient's emotional needs, and relieves their psychological pressure. The feedback information also corrects the diagnosis suggestions of the condition treatment unit, making the medical suggestions more comprehensive and humane, taking into account the patient's mental health, enhancing the patient's trust and dependence on the system, and providing patients with a better medical experience.

[0033] The above are only preferred implementations of the present invention. The protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technical staff in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of this template.

Claims

1. An intelligent medical assistant system based on a large language model, characterized in that: It includes information management module, intelligent consultation module, identification and analysis module and personalized feedback module; The information management module includes a reading unit and a classification unit. The reading unit integrates a medical card reading device and OCR technology. When a patient uses a medical assistant and inserts a medical card, the medical card information of the patient is identified and read, including the patient's age, gender and medical history, and the medical card information is transmitted to the classification unit. After receiving the medical card information from the reading unit, the classification unit classifies the patient according to the read medical card information, outputs the classification result, and transmits it to the intelligent consultation module; The intelligent consultation module includes a switching unit, an analysis and diagnosis unit, and a condition processing unit. The switching unit is used to receive the classification result of the classification unit, and switch the intelligent consultation module to a corresponding communication mode according to the classification result of the classification unit; The analysis and diagnosis unit is equipped with a large language model, which is used to receive the medical card information from the reading unit, and analyze the patient's condition according to the medical card information, determine the complexity of the patient's condition, and output the determination result to the condition processing unit; The condition processing unit is used to receive the judgment result of the analysis and diagnosis unit and the classification result of the classification unit, and to perform secondary classification on the patient according to the judgment result and the classification result, and to generate corresponding diagnosis suggestions according to the result of the secondary classification, and transmit them to the identification and analysis module and the display screen of the intelligent medical assistant device; The recognition and analysis module is used to receive input information from the patient, specifically including voice input or text input, and recognize the patient's emotional state information in real time according to the patient's input information, and transmit it; The personalized feedback module is used to receive the patient's emotional state information, generate real-time response information based on the patient's emotional state information, and transmit the response information to the condition processing unit to correct the diagnosis suggestions generated by the condition processing unit.

2. The intelligent medical assistant system based on a large language model according to claim 1, characterized in that: The specific working mode of the reading unit is as follows: When using the smart medical assistant, the patient first inserts the medical card into the card slot of the smart medical assistant device; The smart medical assistant device uses built-in optical character recognition technology to recognize the information on the medical card; Identify and read the patient's medical card information, and transmit the medical card information to the classification unit.

3. The intelligent medical assistant system based on a large language model according to claim 2, characterized in that: The specific working steps of the classification unit are as follows: Receiving the medical card information read by the reading unit; If the patient’s age is between 0 and 12 years old, it is marked as a patient in category 1; If the patient is aged between 13 and 18 years, they are marked as Category II patients; If the patient’s age is between 19 and 60 years old, they are marked as category three patients; If the patient’s age is over 60 years old, they are marked as category 4 patients; The patient's marking result is transmitted to the intelligent consultation module as a classification result.

4. The intelligent medical assistant system based on a large language model according to claim 3, characterized in that: The specific working steps of the switching unit are as follows: Formulate a corresponding communication mode according to the classification result of the classification unit; For patients in the first category, the font size of the smart medical assistant device was set to 18, and the output speech speed was controlled at 120 words per minute; For patients in category II, the font size of the smart medical assistant device was set to 16, and the output speech speed was controlled at 160 words per minute; For the third category of patients, the font size of the smart medical assistant device was set to 14, and the output speech speed was controlled at 180 words per minute; For the four types of patients, the font size of the smart medical assistant device was set to 22, and the output speech speed was controlled at 110 words per minute; The classification result of the classification unit is received, and a corresponding communication mode is selected according to different classification results.

5. The intelligent medical assistant system based on a large language model according to claim 1, characterized in that: The specific working steps of the analysis and diagnosis unit are as follows: Determining a large language model built into the analysis and diagnosis unit; The analysis and diagnosis unit first receives the medical card information transmitted by the reading unit; According to the built-in large language model, the read medical card information is analyzed to obtain the judgment result of the patient's condition; The specific output judgment results of the patient's condition include mild results, moderate results and severe results, and the judgment results are output to the corresponding condition processing unit.

6. The intelligent medical assistant system based on a large language model according to claim 5, characterized in that: The specific working steps of the condition processing unit are as follows: First, receiving the judgment result of the analysis and diagnosis unit and the classification result of the classification unit, and performing secondary classification on the patient according to the judgment result and the classification result to obtain a secondary classification result, wherein the secondary classification result specifically includes a first-class management requirement, a second-class management requirement, a third-class management requirement and a fourth-class management requirement; According to the secondary classification results, corresponding diagnostic suggestions are generated and transmitted to the identification and analysis module.

7. The intelligent medical assistant system based on a large language model according to claim 6, characterized in that: The specific working steps of performing secondary classification of patients according to the judgment results and classification results to obtain the secondary classification results are as follows; For patients in category 1, category 2, and category 3, all patients whose results are judged to be mild are marked as category 1 management needs; For patients in category 1, category 2, all patients in category 3 with moderate results, and patients in category 4 with mild results, they are marked as category 2 management needs; For patients in category 1, category 2, and all patients in category 3 with severe results, as well as patients in category 4 with moderate results, they are marked as category 3 management needs; For severe results in the four categories of patients, four categories of management needs were marked; The marked results are taken as the secondary classification results.

8. The intelligent medical assistant system based on a large language model according to claim 6, characterized in that: The specific working steps of generating corresponding diagnostic suggestions according to the secondary classification results are as follows; For patients with type 1 management needs, patients are advised to perform self-care at home; For patients with Class II management needs, it is recommended that they go to a specialist clinic for disease assessment and treatment; For patients with the third type of management needs, contact the relevant departments of the hospital including the emergency department, intensive care unit and operating room in advance to inform the patient's condition and arrival time; For patients with category 4 management needs, based on the advice for patients with category 3 management needs, it is recommended that patients go to the hospital's nursing department.

9. The intelligent medical assistant system based on a large language model according to claim 1, characterized in that: The specific working steps of the identification and analysis module are as follows: Obtain text information input by the patient; Obtaining the speech rate R input by the patient, the average strength of the patient's voice, the number of texts input by the patient, and the number of emotional words E in the text input by the patient; The speech rate R input by the patient, the average strength V of the patient's voice, the number of texts input by the patient T, and the number of emotional words E in the text input by the patient are standardized to obtain the standardized speech rate R1 input by the patient, the average strength V1 of the patient's voice, the number of texts T1 input by the patient, and the number of emotional words E1 in the text input by the patient; According to the formula , calculate and obtain the total score of the patient's negative emotions ; According to the formula , calculate the total score of the patient's positive emotions ; According to the formula 100, to get the total score of the patient's emotional state , the total score of the patient's emotional state is transmitted as the patient's emotional state information.

10. The intelligent medical assistant system based on a large language model according to claim 9, characterized in that: The specific working steps of the personalized feedback module are as follows: Get the total score of the patient's emotional state , if 0≤S3<50, it means that the patient is very calm; If 50≤S3<100, it means the patient is very calm; If 100≤S3<150, it means the patient is emotionally agitated; If 150≤S3≤200, it means the patient is very emotional; Generate real-time response information based on the patient's emotional state information; If the patient's emotional state is very calm, no special adjustments are needed in the diagnostic recommendations; If the patient's emotional state is calm: further confirmation of the patient's emotions is added to the diagnostic recommendations. If the patient's emotional state is agitated: add emotional soothing measures to the diagnostic recommendations; If the patient's emotional state is very agitated, the information will be transmitted to the medical staff's mobile terminal for further face-to-face communication.

Images