Intelligent inquiry method and device based on artificial intelligence, equipment and medium

Abstract

The invention relates to the technical field of artificial intelligence, can be applied to medical health business scenes, and discloses an intelligent inquiry method, device and equipment based on artificial intelligence and a medium. Acquiring inquiry voice data of a patient and a doctor, and recognizing identities of the patient and the doctor in the inquiry voice data based on a voiceprint positioning technology; converting the inquiry voice data into inquiry text content according to an identity recognition result; constructing a large language model, and finely adjusting parameters of the large language model to obtain a diagnosis report generation model; and inputting the pre-inquiry information, the inquiry text content and the doctor face diagnosis input information into a diagnosis report generation model to obtain a diagnosis report. Patient information is obtained in advance through pre-inquiry, and the face diagnosis time of doctors is shortened; voice recognition is carried out based on the voiceprint positioning technology, and the dialogue process between a doctor and a patient can be clearly recognized; and the diagnosis report output based on the large model does not need manual input, so that the workload of doctors is reduced.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an artificial intelligence-based intelligent medical consultation method, device, equipment and medium. Background Art

[0002] In the existing technology, during the doctor's consultation process, the first step is to collect the patient's medical history, asking about the patient's basic information, current medical history, past medical history, personal history, drug allergy history, etc., and then issue an examination order based on the content of the consultation. After receiving the examination results, the doctor writes a medical record, makes a diagnosis and prescription, prescribes medicine, and reminds the patient of precautions for recovery.

[0003] Because consultations can be lengthy and inefficient, seeing a doctor becomes more difficult and hospital throughput is low. Entering medical records takes time, and doctors often streamline them to save time, or copy existing templates and modify them to speed things up. This results in incomplete records and the common duplication of medical records between different patients.

[0004] In response to the problems that arise in existing medical consultations, some intelligent medical consultation systems based on artificial intelligence have emerged. The existing intelligent medical consultation systems generally include the following functions: intelligent voice transcription and entry of medical records; auxiliary retrieval of medical knowledge bases, and auxiliary formation of diagnosis and treatment plans.

[0005] In existing intelligent medical consultation systems, when intelligent voice transcription is used to enter medical records, because the medical environment is usually noisy and there are multiple people on site, it is impossible to accurately distinguish the voices of doctors and patients, which brings inconvenience to the generation of intelligent medical records and reduces the efficiency of intelligent consultation. Summary of the Invention

[0006] In view of the above-mentioned deficiencies in the prior art, the present invention provides an artificial intelligence-based intelligent medical consultation method, device, equipment and medium, which aims to solve the problem in the prior art that when intelligent voice transcription is used to enter medical records in the intelligent medical consultation system, because the medical environment is usually noisy and there are multiple people on site, it is impossible to accurately distinguish the voices of doctors and patients, which brings inconvenience to the generation of intelligent medical records and reduces the efficiency of intelligent medical consultation.

[0007] The technical solutions of the present invention are as follows:

[0008] A first embodiment of the present invention provides an intelligent medical consultation method based on artificial intelligence, the method comprising:

[0009] Building a pre-diagnosis system, and obtaining pre-diagnosis information of the patient based on the pre-diagnosis system;

[0010] Constructing a voice preprocessing tool, collecting the patient and doctor's consultation voice data based on the voice preprocessing tool, and identifying the patient and doctor's identities in the consultation voice data based on voiceprint positioning technology;

[0011] Converting the medical consultation voice data into corresponding medical consultation text content according to the identity recognition result;

[0012] Building a large language model, and fine-tuning the parameters of the large language model using a supervised fine-tuning algorithm based on the usage corpus database to obtain a diagnosis report generation model;

[0013] The pre-questioning information, the text content of the questioning and the doctor's face-to-face consultation input information are obtained, and the pre-questioning information, the text content of the questioning and the doctor's face-to-face consultation input information are input into the diagnosis report generation model to obtain a diagnosis report.

[0014] Another embodiment of the present invention provides an intelligent medical consultation device based on artificial intelligence, the device comprising:

[0015] A pre-diagnosis system building module is used to build a pre-diagnosis system and obtain the patient's pre-diagnosis information based on the pre-diagnosis system;

[0016] A speech preprocessing module is used to build a speech preprocessing tool, collect the patient and doctor's consultation voice data based on the speech preprocessing tool, and identify the patient and doctor in the consultation voice data based on voiceprint positioning technology;

[0017] A voice conversion module is used to convert the medical consultation voice data into corresponding medical consultation text content according to the identity recognition result;

[0018] A model fine-tuning module is used to build a large language model and fine-tune the parameters of the large language model based on the usage corpus database using a supervised fine-tuning algorithm to obtain a diagnosis report generation model;

[0019] The diagnosis report generation module is used to obtain the pre-questioning information, the text content of the consultation and the doctor's face-to-face consultation input information, and input the pre-questioning information, the text content of the consultation and the doctor's face-to-face consultation input information into the diagnosis report generation model to obtain a diagnosis report.

[0020] Another embodiment of the present invention provides a computer device, the computer device comprising at least one processor; and

[0021] a memory communicatively connected to the at least one processor; wherein,

[0022] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can perform the steps of the above-mentioned artificial intelligence-based intelligent diagnosis method.

[0023] Another embodiment of the present invention further provides a computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are executed by one or more processors, the one or more processors can execute the steps of the above-mentioned artificial intelligence-based intelligent diagnosis method.

[0024] Beneficial effects: The intelligent medical consultation method, device, equipment and medium based on artificial intelligence of the embodiments of the present invention can obtain the patient's pre-consultation information; collect the patient and doctor's consultation voice data, and identify the patient and doctor's identities in the consultation voice data based on voiceprint positioning technology; convert the consultation voice data into corresponding consultation text content according to the identity recognition result; construct a large language model, fine-tune the parameters of the large language model, and obtain a diagnosis report generation model; input the pre-consultation information, consultation text content and doctor's face-to-face input information into the diagnosis report generation model to obtain a diagnosis report. The present invention obtains patient information in advance through pre-consultation, shortens the doctor's face-to-face consultation time; performs voice recognition based on voiceprint positioning technology, and can clearly identify the conversation process between the doctor and the patient; and outputs the diagnosis report based on the large model, which does not require manual input and reduces the doctor's workload. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments of the present invention. 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 any creative labor.

[0026] Figure 1 A schematic diagram of an application environment of an embodiment of an artificial intelligence-based intelligent medical consultation method of the present invention;

[0027] Figure 2 This is a flow chart of a preferred embodiment of an intelligent medical consultation method based on artificial intelligence of the present invention;

[0028] Figure 3 for Figure 2 A flowchart of a specific embodiment of step S100;

[0029] Figure 4 for Figure 2 A flowchart of a specific embodiment of step S200;

[0030] Figure 5 for Figure 2 A flowchart of a specific embodiment of step S300;

[0031] Figure 6 for Figure 2A flowchart of a specific embodiment of step S400;

[0032] Figure 7 for Figure 2 A flowchart of a specific embodiment of step S500;

[0033] Figure 8 This is a functional module diagram of a preferred embodiment of an intelligent medical consultation device based on artificial intelligence of the present invention;

[0034] Figure 9 A schematic structural diagram of a preferred embodiment of a computer device of the present invention;

[0035] Figure 10 This is another structural diagram of a preferred embodiment of a computer device of the present invention. DETAILED DESCRIPTION

[0036] To make the purpose, technical solution and effect of the present invention clearer and more specific, the present invention is further described in detail below. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0037] The following describes the embodiments of the present invention in conjunction with the accompanying drawings. Those skilled in the art will appreciate that, with the development of technology and the emergence of new scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0038] The terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequential order. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances, and this is merely a way of distinguishing the objects of the same attributes when describing them in the embodiments of the present application. Here, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, so that the process, method, system, product or equipment comprising a series of units need not be limited to those units, but may include other units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0039] The method provided in this application belongs to the field of artificial intelligence (AI) and can be applied to medical and health business scenarios. AI is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. In other words, artificial intelligence is a branch of computer science that attempts to understand the essence of intelligence and produce a new type of intelligent machine that can respond in a similar way to human intelligence. Artificial intelligence is to study the design principles and implementation methods of various intelligent machines so that machines have the functions of perception, reasoning and decision-making. Research in the field of artificial intelligence includes robotics, natural speech processing, computer vision, decision-making and reasoning, human-computer interaction, recommendation and search, basic AI theory, etc.

[0040] The intelligent diagnosis method based on artificial intelligence provided by the embodiment of the present invention can be applied in Figure 1 In an application environment, the client constructs a pre-consultation system and obtains the patient's pre-consultation information based on the pre-consultation system; constructs a voice pre-processing tool, collects the patient and doctor's consultation voice data based on the voice pre-processing tool, and identifies the patient and doctor in the consultation voice data based on the voiceprint positioning technology; converts the consultation voice data into corresponding consultation text content according to the identity recognition result; constructs a large language model, fine-tunes the parameters of the large language model through a supervised fine-tuning algorithm based on a purpose corpus database to obtain a diagnosis report generation model; obtains the pre-consultation information, consultation text content and doctor's face-to-face input information, and inputs the pre-consultation information, consultation text content and doctor's face-to-face input information into the diagnosis report generation model to obtain a diagnosis report. In the present invention, patient information is obtained in advance through pre-consultation, shortening the doctor's face-to-face consultation time; voice recognition is performed based on voiceprint positioning technology, and the conversation process between the doctor and the patient can be clearly identified; the diagnosis report output based on the large model does not require manual input, reducing the doctor's workload. The client can be, but is not limited to, various personal computers, laptops, smart phones, tablet computers and portable wearable devices. The server side can be implemented by an independent server or a server cluster composed of multiple servers. The present invention will be described in detail below through specific embodiments.

[0041] In response to the above problems, the present invention provides an intelligent diagnosis method based on artificial intelligence. Figure 2 , Figure 2 This is a flow chart of a preferred embodiment of an intelligent medical consultation method based on artificial intelligence of the present invention. Figure 2 As shown, it includes:

[0042] Step S100: constructing a pre-diagnosis system, and obtaining pre-diagnosis information of the patient based on the pre-diagnosis system.

[0043] The AI-based intelligent diagnosis method in the embodiments of the present invention is applied to realize an intelligent diagnosis platform in intelligent medical scenarios. The intelligent diagnosis platform is usually implemented through a server, which can obtain pre-diagnosis information and process it according to the corresponding algorithm.

[0044] For example, a pre-diagnosis system is built to obtain the patient's pre-diagnosis information, which includes but is not limited to the chief complaint information (patient's basic information), current medical history, past medical history, personal history and drug allergy history.

[0045] The chief complaint information: The doctor first asks the patient about the most prominent symptoms or discomfort, such as "Where do you feel uncomfortable?" and "How long has this discomfort lasted?"

[0046] History of current illness: Detailed information on the onset of the patient's current illness, the development of symptoms, and accompanying symptoms. For example, for a patient with a fever, the doctor will ask, "Did you have chills when you had a fever?" and "Did you sweat after having a fever?"

[0047] Medical history: Ask the patient whether he or she has had similar illnesses in the past, or suffers from other chronic diseases, such as hypertension, diabetes, etc.

[0048] Personal history: including lifestyle habits (such as diet and exercise), family medical history (whether there are hereditary diseases), etc.

[0049] Drug allergy history: To find out if the patient is allergic to certain drugs so as to avoid using drugs that may trigger allergic reactions.

[0050] For example, a jump link can be established so that when a patient connects to the hospital's official account, they can access the pre-consultation system. The patient or their family can enter the corresponding pre-consultation information through the pre-consultation system.

[0051] Among them, such as Figure 3 As shown, step S100, i.e., building a pre-diagnosis system, and obtaining the patient's pre-diagnosis information based on the pre-diagnosis system, includes:

[0052] Step S101: constructing a pre-diagnosis system based on artificial intelligence;

[0053] Step S102: Generate a basic questionnaire based on the patient's preliminary symptoms, department or disease type, and generate a pre-diagnosis questionnaire based on the search enhancement generation technology and the basic questionnaire;

[0054] Step S103: collecting the answers to the pre-diagnosis questionnaire input by the patient, generating pre-diagnosis information according to the answers, and sending the pre-diagnosis information to the doctor's operation terminal.

[0055] We have built an AI-based pre-consultation system. Patients can access the AI ​​pre-consultation system by linking to the hospital's official account. Patients or their accompanying personnel can conduct pre-consultation through the pre-consultation system.

[0056] The AI ​​pre-diagnosis system provides a large-scale AI agent with voice or text models. Each department (or disease category) designs a basic questionnaire, plus RAG (Retrieval-Augmented Generation). The content of RAG is pathological manifestations and key points of diagnosis, which can be written by specialized experts organized by the hospital.

[0057] AI Agents (Artificial Intelligence) are intelligent entities capable of perceiving their environment, making autonomous decisions, and executing tasks. Unlike traditional AI systems, AI Agents not only understand and generate language but also dynamically interact with users and the environment through multimodal interactions (such as text, voice, and images) to proactively complete complex tasks. Their core features are autonomy, goal-drivenness, and continuous learning, enabling them to adapt to complex and ever-changing scenarios and becoming a key technology driving intelligent upgrades across various industries.

[0058] Retrieval-Augmented Generation (RAG) is a cutting-edge framework that combines retrieval technology with generative AI. It aims to address the limitations of large language models (LLMs) in terms of knowledge updating, factual accuracy, and domain adaptability. Its core logic is to dynamically retrieve external knowledge bases and inject real-time information into the generation process, thereby improving the reliability, timeliness, and professionalism of model output.

[0059] Furthermore, a reward mechanism can be designed in the smart medical platform. For example, points can be obtained by using the pre-consultation system, and the points can be exchanged for some small gifts to encourage patients to use the system.

[0060] The results of the pre-consultation are synchronized to the doctor's operating platform. Before the face-to-face consultation, you can spend one or two minutes to familiarize yourself with it. When the patient comes for the face-to-face consultation, you can only conduct a face-to-face consultation on issues that concern you or issues that need to be clarified in depth, saving time.

[0061] Step S200: construct a voice preprocessing tool, collect the patient and doctor's consultation voice data based on the voice preprocessing tool, and identify the patient and doctor's identities in the consultation voice data based on voiceprint positioning technology.

[0062] To reduce manual input time for doctors, voice recognition tools can be used to collect and process the voices of doctors and patients during face-to-face consultations. A voice preprocessing tool can be pre-built within the doctor's department to perform voice recognition on the collected conversations between patients and doctors. To facilitate voice recognition, voiceprint positioning technology can be used to distinguish the identities of patients and doctors in the consultation voice data, allowing for appropriate voice data processing based on their identities.

[0063] Voiceprint positioning technology is a cutting-edge technology that combines voiceprint recognition and sound source positioning. It achieves identity authentication by analyzing individual voice characteristics and determines the speaker's location by using the propagation characteristics of sound waves.

[0064] Voiceprint recognition builds a speaker's voiceprint model by analyzing biometric features such as the spectral characteristics, fundamental frequency, and formant peaks in speech signals. It uses Mel-Frequency Cepstral Coefficients (MFCCs) and deep learning networks (such as CNN and LSTM) to extract features and compares them with pre-stored templates for identity authentication.

[0065] Sound source localization is based on microphone array technology and is achieved through the following methods:

[0066] Time Difference of Arrival (TDOA): Calculates the time difference between sound waves arriving at different microphones and uses the array geometry to determine the direction of the sound source.

[0067] Beamforming: Enhances signals in specific directions and suppresses noise interference through digital signal processing.

[0068] High-resolution spectrum estimation: Use algorithms such as MUSIC and ESPRIT to improve positioning accuracy.

[0069] By adopting the AI ​​pre-consultation system, the efficiency of medical consultation will be greatly improved. Doctors can already have almost sufficient information before the formal consultation, and can focus on the key points during the formal consultation, thereby improving diagnostic efficiency.

[0070] Among them, Figure 4 As shown, step S200, i.e., constructing a voice preprocessing tool, collecting the patient and doctor's consultation voice data based on the voice preprocessing tool, and identifying the patient and doctor's identities in the consultation voice data based on voiceprint positioning technology, includes:

[0071] Step S201: constructing a speech preprocessing tool based on an array microphone;

[0072] Step S202: collecting the patient and doctor's consultation voice data based on the voice preprocessing tool, and obtaining the patient and doctor's position information through the array microphone;

[0073] Step S203: Processing the consultation voice data using voiceprint positioning technology combined with the position information to identify the identities of the patient and the doctor;

[0074] Step S204: performing noise recognition on the medical consultation voice data based on the voice preprocessing tool, and performing denoising on the recognized noise to obtain target medical consultation voice data.

[0075] When building a speech preprocessing tool, it is combined with an array microphone to identify different accesses during recording to distinguish different speakers. Through voiceprint recognition technology, it is possible to clearly identify who is speaking the voice, so that subsequent speech recognition can correctly convert the voice into text.

[0076] Noise recognition, elimination and noise reduction technologies are applied to minimize noise interference, eliminate non-human voices, and obtain the target medical consultation voice data.

[0077] Noise identification and elimination, as well as noise reduction technologies, improve signal quality in scenarios such as voice communication, audio recording, and industrial monitoring by sensing, analyzing, and suppressing ambient noise. Noise identification distinguishes target signals (such as voice and mechanical vibration) from background noise, providing a basis for subsequent processing.

[0078] Key identification methods include:

[0079] Spectral feature analysis: Short-time Fourier transform (STFT) is used to convert time-domain signals into the frequency domain, extracting noise spectrum distribution characteristics (such as the flat spectrum of broadband noise and the harmonic components of periodic noise). For example, in factory equipment noise, motor harmonics (integer multiples of 50 Hz) can be identified through spectrum peak detection.

[0080] Joint Time-Frequency Analysis: Wavelet transform and Mel-spectrogram are used to capture the time-varying characteristics of noise. Application: Burst noise (such as a door slamming) can be located by identifying the sudden change in time-frequency energy.

[0081] Deep learning classification: Uses convolutional neural networks (CNNs) or recurrent neural networks (RNNs) to classify noise types (such as traffic noise and wind noise). Advantage: No need to manually design features; noise patterns can be learned directly from the raw signal.

[0082] Noise cancellation technology is used to separate or suppress noise from mixed signals and restore the target signal.

[0083] Noise cancellation technologies include:

[0084] Traditional signal processing algorithms: Spectral subtraction estimates the noise spectrum and subtracts it from the mixed signal. It is suitable for stationary noise (such as white noise). Limitations: It is prone to generating "musical noise" (residual random spectral spikes). Wiener filtering: Based on the minimum mean square error criterion, it estimates the optimal estimate of the target signal in the frequency domain. Adaptive filtering: It dynamically tracks noise changes by iteratively updating filter coefficients. Application: Eliminating feedback noise from speakers to microphones in acoustic echo cancellation (AEC).

[0085] Deep Learning Denoising: Supervised Learning Models: DNN / CNN Denoiser: Direct end-to-end mapping from noisy signals to clean signals (e.g., SEGAN, DCCRN). U-Net Architecture: Preserves signal details through an encoder-decoder structure combined with skip connections. Unsupervised / Self-Supervised Learning: Noise2Noise: Trains models using pairs of noisy signals without the need for clean signal labels. Contrastive Learning (SimCLR): Improves generalization by comparing the representation differences between noise and clean signals. Time and Frequency Domain Fusion: Fully Convolutional Time-Frequency Network (Conv-TasNet): Models signals directly in the time domain, avoiding the phase distortion problem of STFT.

[0086] The embodiment of the present invention can clearly identify the conversation process between doctors and patients through powerful noise processing and person separation technology.

[0087] Step S300: converting the medical consultation voice data into corresponding medical consultation text content according to the identity recognition result.

[0088] Among them, such as Figure 5 As shown, step S300, that is, converting the medical consultation voice data into corresponding medical consultation text content according to the identity recognition result, includes:

[0089] Step S301: Pre-build a speech recognition model that is optimized and trained using corpus data from a hospital's vertical field, wherein the speech recognition model includes a dialect recognition model and a non-dialect recognition model;

[0090] Step S302: Acquire target medical consultation voice data, and determine whether the target medical consultation voice data is in a dialect. If it is in a dialect, execute step S303; if it is not in a dialect, execute step S304;

[0091] Step S303: Obtain a corresponding dialect recognition model, and convert the medical consultation voice data into corresponding medical consultation text content based on the dialect recognition model, identity recognition results, and hot word technology;

[0092] Step S304: Obtain a non-dialect recognition model, and convert the medical consultation voice data into corresponding medical consultation text content based on the non-dialect recognition model, identity recognition results, and hot word technology.

[0093] ASR (Automatic Speech Recognition) vertical domain corpus training, which is optimized and trained using corpus data from the hospital's vertical field, can more accurately output text related to medical diagnosis. For dialects in specific regions, such as Sichuanese and Cantonese, specialized dialect ASR is used.

[0094] In the process of converting speech to text, the use of hot word technology enables ASR to more accurately identify words and sentences in hospital diagnosis scenarios.

[0095] Hot words refer to a set of words prioritized in automatic speech recognition (ASR) systems to improve the recognition accuracy of specific keywords or phrases. These hot words are typically high-frequency words in business scenarios, proper nouns, or core commands that require precise recognition. Technical measures are used to enhance the model's sensitivity to these words, thereby optimizing user experience and task execution.

[0096] The core of ASR hotwords is to use contextual biasing technology to give higher weight to preset hotwords during the decoding stage. The specific implementation methods include:

[0097] Acoustic model weighting: Adjust the acoustic feature scores of hot words to give them a greater advantage in decoding competition. For example, in smart speaker scenarios, increasing the acoustic model score of "play music" can reduce the probability of misidentification as "play news."

[0098] Language model enhancement: This increases the prior probability of hot words in the language model, giving sentences containing these words a higher overall score. For example, in a customer service system, the probability of hot words such as "manual customer service" and "order inquiry" appearing in the language model is increased.

[0099] Decoder intervention: Use a finite state machine (FST) or neural network decoder to prioritize candidate results containing hot words in the decoding path. For example, in a meeting record system, FST can be used to force the decoding path to include hot words such as "meeting minutes" and "next steps."

[0100] In the embodiment of the present invention, the ASR trained by using corpus data from the vertical field of hospitals can more accurately output text related to medical diagnosis.

[0101] Step S400: construct a large language model, and fine-tune the parameters of the large language model through a supervised fine-tuning algorithm based on the usage corpus database to obtain a diagnosis report generation model.

[0102] Large language models fine-tuned based on SFT, such as the Deepseek tuned version, are more suitable for the organization and output of hospital diagnosis processes and result documents.

[0103] Supervised fine-tuning (SFT), also known as instruction fine-tuning, refers to further training and adjusting the model based on the trained language model using labeled specific task data.

[0104] Among them, such as Figure 6 As shown, in step S400, a large language model is constructed, and the parameters of the large language model are fine-tuned by a supervised fine-tuning algorithm based on the usage corpus database to obtain a diagnosis report generation model, including:

[0105] Step S401: Build a large language model based on deep learning;

[0106] Step S402: constructing a usage corpus database for supervised fine-tuning based on the manual standard corpus data of doctors and the corpus data output by the large language model;

[0107] Step S403: Fine-tune the parameters of the large language model based on the usage corpus database using a supervised fine-tuning algorithm to obtain a diagnosis report generation model.

[0108] Build a large language model based on deep learning, for example, the Deepseek large speech model can be used.

[0109] Based on SFT fine-tuning, we update the parameters of the large language model. We pre-acquire standard terms manually input by doctors and obtain the medical corpus output by existing excellent models. We then combine the standard user and medical corpus, combining the manual standards of doctors with the corpus output by excellent models, to construct a corpus for fine-tuning. Fine-tuning the large model makes it more adaptable to hospital scenarios and enables professional and effective organization and output of hospital diagnostic processes and result documents.

[0110] Step S500: Obtain the pre-questionnaire information, the text content of the questioning, and the doctor's face-to-face consultation input information, and input the pre-questionnaire information, the text content of the questioning, and the doctor's face-to-face consultation input information into the diagnosis report generation model to obtain a diagnosis report.

[0111] The pre-consultation information, the written content of the consultation, and the doctor's face-to-face consultation information are obtained and, according to the doctor's instructions, are input into the diagnostic report generation model. The diagnostic report generation model processes the input information and generates and outputs a diagnostic report. The diagnostic report may include medical records, prescriptions, medication orders, patient instructions, rehabilitation guides, healthy diet plans, and other data.

[0112] The RAG-based diagnosis and treatment model can greatly assist doctors in diagnosis, especially the diagnosis and treatment plans for typical cases. The diagnosis and treatment plans for typical cases can be expert consultation records or records of authoritative typical cases. These references can improve the diagnosis and treatment level of each doctor.

[0113] Among them, such as Figure 7 As shown, in step S500, the pre-questioning information, the text content of the questioning, and the doctor's face-to-face consultation input information are obtained, and the pre-questioning information, the text content of the questioning, and the doctor's face-to-face consultation input information are input into the diagnosis report generation model to obtain a diagnosis report, including:

[0114] Step S501: obtaining the pre-diagnosis information, the text content of the medical consultation, and the doctor's face-to-face consultation input information; if a doctor's diagnosis report generation instruction is detected, inputting the pre-diagnosis information, the text content of the medical consultation, and the doctor's face-to-face consultation input information into the diagnosis report generation model;

[0115] Step S502: Obtain the diagnostic report generated by the diagnostic report generation model, and display the diagnostic report on the doctor's operation terminal;

[0116] Step S503: Determine whether the doctor has modified the diagnosis report. If the doctor has not modified the diagnosis report, execute step S504. If the doctor has modified the diagnosis report, execute step S505.

[0117] Step S504: The current diagnostic report is used as a target diagnostic report, and the target diagnostic report is sent to the user operation terminal;

[0118] Step S505: Use the modified diagnostic report as a target diagnostic report, and send the target diagnostic report to the user operation terminal.

[0119] Based on the information obtained from the pre-interview and the information input by the doctor during the face-to-face consultation, when the doctor believes that the diagnostic information is comprehensive, he / she instructs the diagnostic system to generate a diagnostic report based on the RAG diagnosis and treatment model. The doctor reads the report and makes manual adjustments as needed.

[0120] In some other embodiments, image recognition technology is combined to analyze the patient's medical images, and the analysis results are integrated with the text content of the medical interview to generate a more comprehensive diagnosis report;

[0121] Combine wearable device data (such as heart rate and blood sugar monitoring) to fuse multimodal data with the text of the medical interview to generate a more comprehensive diagnostic report;

[0122] Based on the patient's genomic data, AI technology is used to generate personalized treatment plans, such as recommending targeted drugs or predicting disease risks.

[0123] It can also push appropriate exercise plans, diet therapy plans and other data to patients based on their smart medical records.

[0124] In some other embodiments, step S500 obtains the pre-questionnaire information, the text content of the question, and the doctor's face-to-face consultation input information, and inputs the pre-questionnaire information, the text content of the question, and the doctor's face-to-face consultation input information into the diagnosis report generation model to obtain a diagnosis report, further comprising:

[0125] Obtaining the pre-questionnaire information, the text content of the questioning, and the doctor's face-to-face consultation input information, and inputting the pre-questionnaire information, the text content of the questioning, and the doctor's face-to-face consultation input information into the diagnosis report generation model;

[0126] Obtaining a diagnostic report output by a diagnostic report generation model, wherein the diagnostic report includes an intelligent medical record, prescription, medication list, patient instructions, rehabilitation guide, and healthy recipes;

[0127] The usage corpus database of the diagnosis report generation model is updated based on the diagnosis report.

[0128] Large model generation documentation can include the following:

[0129] Smart medical records, prescriptions, medicine lists, patient instructions, rehabilitation guides, and healthy recipes can all be implemented through customized AI agents. Different hospitals may have different requirements, which can be completed according to the following steps: knowledge base organization, typical copywriting organization, large-scale model customized AI agent (prompt, RAG and other technical assistance can be used here), and the large-scale AI agent collaborates with the diagnostic platform to complete the work.

[0130] Rich output documents give patients a sense of gain. For example, rehabilitation instructions can be illustrated with pictures and texts, and healthy recipes can be easier for patients to understand and follow, replacing the previous limited verbal instructions from doctors.

[0131] Furthermore, the multiple AI agents used in the present invention can be further optimized in the application, so that each subtask can be performed more professionally and the level can be continuously improved.

[0132] Compared with the existing technology, the intelligent medical consultation method based on artificial intelligence in the embodiment of the present invention has the following technical advantages:

[0133] 1) Improve the efficiency of medical treatment: Pre-consultation allows patients to "see a doctor" in advance, which shortens the time doctors spend and increases the overall turnover rate of the hospital, which is beneficial to patients, doctors and hospitals.

[0134] 2) Improve medical reference cases: A multi-AI agent system supported by a large model can use a rich professional knowledge system to help doctors diagnose and prescribe medicine, allowing all doctors to achieve the same level of medical treatment as senior experts.

[0135] 3) Detailed Medical Information: The AI ​​agent's output, combined with physician review, significantly enhances the level of detail in medical records, eliminating the need for simplification due to busy physicians. Comprehensive and diverse information is required. For example, not only medical records and prescriptions, but also patient guides, rehabilitation guidelines, and health recipes are included. Previously, medical consultations often contained scant information, leaving patients with a constant stream of questions and inquiries, resulting in a poor patient experience. However, documents like rehabilitation guides and health recipes are crucial during the patient's post-hospital recovery process, significantly improving patient satisfaction.

[0136] 4) Benign accumulation of diagnostic knowledge: As the completeness of medical records is gradually guaranteed, clinical diagnosis and subsequent patient feedback form a closed loop of the diagnostic process, effectively forming a process of benign accumulation of diagnostic knowledge. It can accumulate, summarize and refine an excellent diagnostic knowledge base, and keep the base evergreen, which is conducive to the positive improvement of the overall level of the health system.

[0137] It should be noted that there is not necessarily a certain order between the above steps. A person skilled in the art can understand, based on the description of the embodiments of the present invention, that in different embodiments, the above steps may have different execution orders, that is, they may be executed in parallel, or may be executed interchangeably, etc.

[0138] Another embodiment of the present invention provides an intelligent medical consultation device based on artificial intelligence, which corresponds one-to-one with the intelligent medical consultation method based on artificial intelligence in the above embodiment. Figure 8 As shown, the device 1 includes:

[0139] A pre-diagnosis system construction module 100 is used to construct a pre-diagnosis system and obtain pre-diagnosis information of the patient based on the pre-diagnosis system;

[0140] The speech preprocessing module 200 is used to construct a speech preprocessing tool, collect the patient and doctor's consultation voice data based on the speech preprocessing tool, and identify the patient and doctor in the consultation voice data based on voiceprint positioning technology;

[0141] The voice conversion module 300 is used to convert the medical consultation voice data into corresponding medical consultation text content according to the identity recognition result;

[0142] A model fine-tuning module 400 is used to construct a large language model and fine-tune the parameters of the large language model using a supervised fine-tuning algorithm based on a usage corpus database to obtain a diagnosis report generation model;

[0143] The diagnosis report generation module 500 is used to obtain the pre-questioning information, the text content of the consultation and the doctor's face-to-face consultation input information, and input the pre-questioning information, the text content of the consultation and the doctor's face-to-face consultation input information into the diagnosis report generation model to obtain a diagnosis report.

[0144] The specific implementation method is shown in the method embodiment, which will not be repeated here.

[0145] In one embodiment, the pre-diagnosis system construction module 100 is specifically used to:

[0146] Build an AI-based pre-diagnosis system;

[0147] Generate a basic questionnaire based on the patient's preliminary symptoms, department or disease type, and generate a pre-diagnosis questionnaire based on the retrieval enhancement generation technology and the basic questionnaire;

[0148] The patient's answer to the pre-diagnosis questionnaire is collected, pre-diagnosis information is generated based on the answer, and the pre-diagnosis information is sent to the doctor's operation terminal.

[0149] The specific implementation method is shown in the method embodiment, which will not be repeated here.

[0150] In one embodiment, the speech pre-processing module 200 is specifically configured to:

[0151] Build a speech preprocessing tool based on array microphones;

[0152] The speech preprocessing tool is used to collect the patient and doctor's consultation voice data, and the array microphone is used to obtain the patient and doctor's position information;

[0153] Using voiceprint positioning technology in combination with the position information to process the consultation voice data and identify the identities of the patient and the doctor;

[0154] Noise recognition is performed on the medical consultation voice data based on the voice preprocessing tool, and the recognized noise is denoised to obtain target medical consultation voice data.

[0155] The specific implementation method is shown in the method embodiment, which will not be repeated here.

[0156] In one embodiment, the voice conversion module 300 is specifically configured to:

[0157] Pre-build a speech recognition model that is optimized and trained using corpus data from the hospital's vertical field, including a dialect recognition model and a non-dialect recognition model;

[0158] Obtain target medical consultation voice data, and determine whether the target medical consultation voice data is in a dialect;

[0159] If it is a dialect, obtain the corresponding dialect recognition model, and convert the medical consultation voice data into corresponding medical consultation text content based on the dialect recognition model, identity recognition results and hot word technology;

[0160] If it is not a dialect, a non-dialect recognition model is obtained, and the medical consultation voice data is converted into corresponding medical consultation text content based on the non-dialect recognition model, identity recognition results and hot word technology.

[0161] The specific implementation method is shown in the method embodiment, which will not be repeated here.

[0162] In one embodiment, the model fine-tuning module 400 is specifically configured to:

[0163] Build a large language model based on deep learning;

[0164] Based on the manual standard corpus data of doctors and the corpus data output by the large language model, a purpose corpus database for supervised fine-tuning is constructed;

[0165] Based on the usage corpus database, the parameters of the large language model are fine-tuned using a supervised fine-tuning algorithm to obtain a diagnosis report generation model.

[0166] The specific implementation method is shown in the method embodiment, which will not be repeated here.

[0167] In one embodiment, the diagnostic report generating module 500 is specifically configured to:

[0168] Obtaining the pre-interview information, the text content of the interview, and the doctor's face-to-face consultation input information; if a doctor's diagnosis report generation instruction is detected, inputting the pre-interview information, the text content of the interview, and the doctor's face-to-face consultation input information into the diagnosis report generation model;

[0169] Obtaining a diagnostic report generated by the diagnostic report generation model, and displaying the diagnostic report on a doctor's operation terminal;

[0170] Determining whether the doctor modifies the diagnosis report;

[0171] If the doctor has not modified the diagnosis report, the current diagnosis report is used as the target diagnosis report, and the target diagnosis report is sent to the user operation terminal;

[0172] If the doctor modifies the diagnosis report, the modified diagnosis report is used as the target diagnosis report and the target diagnosis report is sent to the user operation terminal.

[0173] The specific implementation method is shown in the method embodiment, which will not be repeated here.

[0174] In one embodiment, the diagnostic report generating module 500 is further configured to:

[0175] Obtaining the pre-questionnaire information, the text content of the questioning, and the doctor's face-to-face consultation input information, and inputting the pre-questionnaire information, the text content of the questioning, and the doctor's face-to-face consultation input information into the diagnosis report generation model;

[0176] Obtaining a diagnostic report output by a diagnostic report generation model, wherein the diagnostic report includes an intelligent medical record, prescription, medication list, patient instructions, rehabilitation guide, and healthy recipes;

[0177] The usage corpus database of the diagnosis report generation model is updated based on the diagnosis report.

[0178] The specific implementation method is shown in the method embodiment, which will not be repeated here.

[0179] The present invention provides an intelligent medical consultation device based on artificial intelligence, which obtains patient information in advance through pre-interview, shortening the doctor's face-to-face consultation time; performs speech recognition based on voiceprint positioning technology, and can clearly identify the conversation process between the doctor and the patient; and outputs a diagnostic report based on a large model, which does not require manual input and reduces the doctor's workload.

[0180] Another embodiment of the present invention provides a computer device, which may be a server, and its internal structure diagram may be as shown in FIG. Figure 9 As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile and / or volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external client via a network connection. When the computer program is executed by the processor, it realizes the functions or steps on the server side of an intelligent diagnosis method based on artificial intelligence.

[0181] In one embodiment, a computer device is provided. The computer device may be a client, and its internal structure diagram may be as follows: Figure 10As shown. The computer device includes a processor, memory, network interface, display screen and input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external server via a network connection. When the computer program is executed by the processor, it implements the functions or steps on the client side of an intelligent diagnosis method based on artificial intelligence.

[0182] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the following steps are performed:

[0183] Building a pre-diagnosis system, and obtaining pre-diagnosis information of the patient based on the pre-diagnosis system;

[0184] Constructing a voice preprocessing tool, collecting the patient and doctor's consultation voice data based on the voice preprocessing tool, and identifying the patient and doctor's identities in the consultation voice data based on voiceprint positioning technology;

[0185] Converting the medical consultation voice data into corresponding medical consultation text content according to the identity recognition result;

[0186] Building a large language model, and fine-tuning the parameters of the large language model using a supervised fine-tuning algorithm based on the usage corpus database to obtain a diagnosis report generation model;

[0187] The pre-questioning information, the text content of the questioning and the doctor's face-to-face consultation input information are obtained, and the pre-questioning information, the text content of the questioning and the doctor's face-to-face consultation input information are input into the diagnosis report generation model to obtain a diagnosis report.

[0188] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0189] Building a pre-diagnosis system, and obtaining pre-diagnosis information of the patient based on the pre-diagnosis system;

[0190] Constructing a voice preprocessing tool, collecting the patient and doctor's consultation voice data based on the voice preprocessing tool, and identifying the patient and doctor's identities in the consultation voice data based on voiceprint positioning technology;

[0191] Converting the medical consultation voice data into corresponding medical consultation text content according to the identity recognition result;

[0192] Building a large language model, and fine-tuning the parameters of the large language model using a supervised fine-tuning algorithm based on the usage corpus database to obtain a diagnosis report generation model;

[0193] The pre-questioning information, the text content of the questioning and the doctor's face-to-face consultation input information are obtained, and the pre-questioning information, the text content of the questioning and the doctor's face-to-face consultation input information are input into the diagnosis report generation model to obtain a diagnosis report.

[0194] It should be noted that the above functions or steps that can be implemented by the computer-readable storage medium or computer device can be found in the relevant descriptions of the server side and the client side in the aforementioned method embodiment. To avoid repetition, they will not be described one by one here.

[0195] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchl ink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0196] The embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules may be selected to achieve the objectives of the present embodiments as needed.

[0197] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a general hardware platform, or of course, can also be implemented by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the relevant technology can be embodied in the form of a software product. This computer software product can be present in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of each embodiment or certain parts of the embodiment.

[0198] It should be noted that if software tools or components other than those of our company appear in the embodiments of this application, they are only used for illustration and do not represent actual use.

[0199] Conditional language such as "can," "may," or "might," among others, unless specifically stated otherwise or otherwise understood within the context as used, is generally intended to convey that particular embodiments can include, while other embodiments do not, particular features, elements, and / or operations. Thus, such conditional language is also generally intended to imply that features, elements, and / or operations are anyway required for one or more embodiments or that one or more embodiments must include logic for determining, with or without input or prompting, whether such features, elements, and / or operations are included or to be performed in any particular embodiment.

[0200] What has been described herein in this specification and the accompanying drawings includes examples that can provide intelligent medical consultation methods and devices based on artificial intelligence. Of course, it is not possible to describe every conceivable combination of elements and / or methods for the purpose of describing the various features of the present disclosure, but it is recognized that many other combinations and permutations of the disclosed features are possible. Therefore, it is obvious that various modifications can be made to the present disclosure without departing from the scope or spirit of the present disclosure. In addition, or in the alternative, other embodiments of the present disclosure may be obvious from consideration of this specification and the accompanying drawings and the practice of the present disclosure as presented herein. It is intended that the examples set forth in this specification and the accompanying drawings are considered to be illustrative and not restrictive in all respects. Although specific terms are employed herein, they are used in a general and descriptive sense and are not used for limiting purposes.

Claims

1. An intelligent medical consultation method based on artificial intelligence, characterized by , the method comprises: Building a pre-diagnosis system, and obtaining pre-diagnosis information of the patient based on the pre-diagnosis system; Constructing a voice preprocessing tool, collecting the patient and doctor's consultation voice data based on the voice preprocessing tool, and identifying the patient and doctor's identities in the consultation voice data based on voiceprint positioning technology; Converting the medical consultation voice data into corresponding medical consultation text content according to the identity recognition result; Building a large language model, and fine-tuning the parameters of the large language model using a supervised fine-tuning algorithm based on the usage corpus database to obtain a diagnosis report generation model; The pre-questioning information, the text content of the questioning and the doctor's face-to-face consultation input information are obtained, and the pre-questioning information, the text content of the questioning and the doctor's face-to-face consultation input information are input into the diagnosis report generation model to obtain a diagnosis report.

2. The intelligent medical consultation method based on artificial intelligence according to claim 1, characterized in that: The constructing of the pre-diagnosis system and obtaining the pre-diagnosis information of the patient based on the pre-diagnosis system includes: Build an AI-based pre-diagnosis system; Generate a basic questionnaire based on the patient's preliminary symptoms, department or disease type, and generate a pre-diagnosis questionnaire based on the retrieval enhancement generation technology and the basic questionnaire; The patient's answer to the pre-diagnosis questionnaire is collected, pre-diagnosis information is generated based on the answer, and the pre-diagnosis information is sent to the doctor's operation terminal.

3. The intelligent medical consultation method based on artificial intelligence according to claim 1, characterized in that: The method of constructing a voice preprocessing tool, collecting the patient and doctor's consultation voice data based on the voice preprocessing tool, and identifying the patient and doctor's identities in the consultation voice data based on voiceprint positioning technology includes: Build a speech preprocessing tool based on array microphones; The speech preprocessing tool is used to collect the patient and doctor's consultation voice data, and the array microphone is used to obtain the patient and doctor's position information; Using voiceprint positioning technology in combination with the position information to process the consultation voice data and identify the identities of the patient and the doctor; Noise recognition is performed on the medical consultation voice data based on the voice preprocessing tool, and the recognized noise is denoised to obtain target medical consultation voice data.

4. The intelligent medical consultation method based on artificial intelligence according to claim 1, characterized in that: The converting of the medical consultation voice data into corresponding medical consultation text content according to the identity recognition result includes: Pre-build a speech recognition model that is optimized and trained using corpus data from the hospital's vertical field, including a dialect recognition model and a non-dialect recognition model; Obtain target medical consultation voice data, and determine whether the target medical consultation voice data is in a dialect; If it is a dialect, obtain the corresponding dialect recognition model, and convert the medical consultation voice data into corresponding medical consultation text content based on the dialect recognition model, identity recognition results and hot word technology; If it is not a dialect, a non-dialect recognition model is obtained, and the medical consultation voice data is converted into corresponding medical consultation text content based on the non-dialect recognition model, identity recognition results and hot word technology.

5. The intelligent medical consultation method based on artificial intelligence according to claim 1, characterized in that: The method of constructing a large language model and fine-tuning the parameters of the large language model using a supervised fine-tuning algorithm based on a usage corpus database to obtain a diagnosis report generation model includes: Build a large language model based on deep learning; Based on the manual standard corpus data of doctors and the corpus data output by the large language model, a purpose corpus database for supervised fine-tuning is constructed; Based on the usage corpus database, the parameters of the large language model are fine-tuned using a supervised fine-tuning algorithm to obtain a diagnosis report generation model.

6. The intelligent medical consultation method based on artificial intelligence according to claim 1, characterized in that: The obtaining of the pre-questioning information, the text content of the questioning, and the doctor's face-to-face consultation input information, and inputting the pre-questioning information, the text content of the questioning, and the doctor's face-to-face consultation input information into the diagnosis report generation model to obtain a diagnosis report includes: Obtaining the pre-questionnaire information, the text content of the questioning, and the doctor's face-to-face consultation input information, and if a doctor's diagnosis report generation instruction is detected, inputting the pre-questionnaire information, the text content of the questioning, and the doctor's face-to-face consultation input information into the diagnosis report generation model; Obtaining a diagnostic report generated by the diagnostic report generation model, and displaying the diagnostic report on a doctor's operation terminal; Determining whether the doctor modifies the diagnosis report; If the doctor has not modified the diagnosis report, the current diagnosis report is used as the target diagnosis report, and the target diagnosis report is sent to the user operation terminal; If the doctor modifies the diagnosis report, the modified diagnosis report is used as the target diagnosis report and the target diagnosis report is sent to the user operation terminal.

7. The intelligent medical consultation method based on artificial intelligence according to claim 1, characterized in that: The obtaining of the pre-questioning information, the text content of the questioning, and the doctor's face-to-face consultation input information, and inputting the pre-questioning information, the text content of the questioning, and the doctor's face-to-face consultation input information into the diagnosis report generation model to obtain a diagnosis report includes: Obtaining the pre-questionnaire information, the text content of the questioning, and the doctor's face-to-face consultation input information, and inputting the pre-questionnaire information, the text content of the questioning, and the doctor's face-to-face consultation input information into the diagnosis report generation model; Obtaining a diagnostic report output by a diagnostic report generation model, wherein the diagnostic report includes an intelligent medical record, prescription, medication list, patient instructions, rehabilitation guide, and healthy recipes; The usage corpus database of the diagnosis report generation model is updated based on the diagnosis report.

8. An intelligent medical consultation device based on artificial intelligence, characterized in that: The device comprises: A pre-diagnosis system building module is used to build a pre-diagnosis system and obtain the patient's pre-diagnosis information based on the pre-diagnosis system; A speech preprocessing module is used to build a speech preprocessing tool, collect the patient and doctor's consultation voice data based on the speech preprocessing tool, and identify the patient and doctor in the consultation voice data based on voiceprint positioning technology; A voice conversion module is used to convert the medical consultation voice data into corresponding medical consultation text content according to the identity recognition result; A model fine-tuning module is used to build a large language model and fine-tune the parameters of the large language model based on the usage corpus database using a supervised fine-tuning algorithm to obtain a diagnosis report generation model; The diagnosis report generation module is used to obtain the pre-questioning information, the text content of the consultation and the doctor's face-to-face consultation input information, and input the pre-questioning information, the text content of the consultation and the doctor's face-to-face consultation input information into the diagnosis report generation model to obtain a diagnosis report.

9. A computer device, characterized in that: The computer device includes at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can perform the steps of the artificial intelligence-based intelligent diagnosis method described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer-executable instructions. When the computer-executable instructions are executed by one or more processors, the one or more processors can execute the steps of the artificial intelligence-based intelligent diagnosis method described in any one of claims 1-7.

Images