Intelligent inquiry large model technical method based on AI algorithm

Abstract

The invention relates to the technical field of artificial intelligence, in particular to an intelligent inquiry large model technical method based on an AI algorithm, which comprises the following steps: based on user inquiry basic information and medical field professional knowledge base data, extracting user symptom core features, and constructing a user exclusive symptom map and a standardized symptom feature library; on the basis of the user exclusive symptom map and a standardized symptom feature library, the feature matching degree and the disease cause integrating degree of the user symptom and each standardized symptom are calculated; and based on the feature matching degree and the disease cause integrating degree, generating a preliminary disease judgment result and a symptom supplementary checking list. By calculating the feature matching degree and the disease cause integrating degree, the intelligent system can generate a preliminary disease judgment result, and the judgment result is continuously optimized according to details supplemented by the user; by means of the iterative optimization process, the accuracy of diagnosis can be remarkably improved, and misdiagnosis and missed diagnosis are reduced.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, specifically to a large-scale intelligent diagnostic model technology method based on AI algorithms. Background Technology

[0002] Medical consultation refers to the process in which doctors or medical professionals collect information about a patient's condition, symptoms, medical history, and lifestyle habits through communication during the medical process, and then make a preliminary judgment, diagnosis, or treatment suggestion. Consultation is one of the important links in medical diagnosis. It helps doctors understand the patient's subjective symptoms and provides a basis for subsequent physical examinations, laboratory tests, or imaging examinations.

[0003] Currently, traditional diagnostic methods typically rely on single or limited symptom descriptions, lacking in-depth analysis of the time sequence and correlation between symptoms. This can easily lead to inaccurate symptom matching and an inability to fully understand the underlying causes behind the symptoms. Furthermore, traditional diagnoses often provide fixed results based on preliminary symptom information, lacking real-time feedback and iterative optimization processes for users to supplement details. As a result, diagnostic accuracy is low, which may lead to misdiagnosis or missed diagnosis.

[0004] Furthermore, in traditional diagnosis, doctors systematically investigate the causes of illness based on a list of symptoms. However, when faced with complex symptoms, traditional methods are prone to conflicting or missing symptom information and are difficult to correct and update the consistency of the cause in real time. In addition, traditional methods are difficult to effectively identify potential complications, which are often only discovered after the disease has progressed to a certain stage. The lack of an early warning mechanism leads to poor timeliness of intervention and prevention measures. Summary of the Invention

[0005] To achieve the above objectives, the present invention provides the following technical solution: a large-scale intelligent medical consultation model based on AI algorithms, comprising: Acquire basic information about user consultations and medical professional knowledge base data, wherein the basic information about user consultations includes symptom presentation, lifestyle habits and medical history, and the medical professional knowledge base data includes disease characteristics, treatment guidelines and drug compatibility information; Based on the user's basic consultation information and medical professional knowledge base data, the core features of the user's symptoms are extracted to construct a user-specific symptom atlas and a standardized disease feature library. Based on the user-specific symptom atlas and standardized disease feature library, the feature matching degree and etiological consistency between user symptoms and each standardized disease are calculated. Based on the feature matching degree and etiology fit degree, a preliminary symptom judgment result and a supplementary symptom verification list are generated. Based on the preliminary symptom diagnosis results and the symptom supplement checklist, collect the user's supplementary symptom details; combine the supplementary symptom details to update the user-specific symptom atlas and the etiology matching degree, and optimize the preliminary symptom diagnosis results; Based on the optimized symptom diagnosis, a complete consultation conclusion is generated, which includes symptom analysis, examination recommendations, and intervention plan.

[0006] Preferably, based on the user's basic consultation information and medical professional knowledge base data, core features of user symptoms are extracted to construct a user-specific symptom atlas and a standardized disease feature library, including: Based on the user's basic consultation information, symptom occurrence features, symptom association features, and symptom persistence features are extracted to obtain the core features of the user's symptoms. Based on the medical professional knowledge base data, typical symptom elements, atypical manifestation elements, and complication-related elements of various diseases are decomposed to obtain a standardized disease feature set; By integrating the core features of the user symptoms, a user-specific symptom atlas is constructed, and by integrating the standardized disease feature set, a standardized disease feature library is built.

[0007] Preferably, based on the user-specific symptom atlas and the standardized disease feature library, the feature matching degree and etiological consistency between the user's symptoms and each standardized disease are calculated, including: Map the symptom nodes in the user-specific symptom atlas to the typical symptom attributes in the standardized disease feature library; Calculate the symptom entity overlap ratio, symptom association edge strength similarity, and time series matching degree; The feature matching degree between user symptoms and each standardized disease is determined by integrating the overlap ratio of symptom entities, the similarity of the strength of symptom association edges, and the time series matching degree. Extract external influencing factors, symptom time series, and associated edge directions from the user-specific symptom atlas; Match the etiological transmission pathways, triggering factors, and pathological mechanisms in the standardized symptom feature database; analyze the consistency of etiological pathways, the correlation of triggering factors, and the temporal fit between symptom development and etiological transmission; By integrating the consistency of the etiological pathways, the correlation of the triggers, and the temporal fit between the symptom development and the etiological spread, the etiological fit between user symptoms and each standardized disease is determined.

[0008] Preferably, based on the feature matching degree and etiological consistency degree, a preliminary symptom judgment result and a supplementary symptom verification list are generated, including: Based on the feature matching degree and etiological fit degree, the set of source diagnostic directions and the set of candidate optimized diagnostic directions are determined; Determine whether the set of candidate optimized diagnostic directions is an invalid set; If the candidate optimized diagnostic direction set is not an invalid set, then based on the candidate optimized diagnostic direction set, the diagnostic direction that is most closely related to the cause of each source diagnostic direction in the source diagnostic direction set and has the highest feature matching degree is selected, and the optimized diagnostic direction corresponding to each source diagnostic direction is obtained and integrated into the preliminary disease judgment result.

[0009] Preferably, based on the feature matching degree and etiological fit, the set of source diagnostic directions and the set of candidate optimized diagnostic directions are determined, including: Based on the feature matching degree and etiological fit, the diagnostic direction combinations that do not meet the fit criteria are screened out to obtain the source diagnostic direction set. Based on the feature matching degree and etiological fit, determine whether there are diagnostic directions whose fit exceeds the optimization standard; If there are diagnostic directions whose fit exceeds the optimization criteria, then these diagnostic directions that exceed the optimization criteria are combined to obtain a set of candidate optimized diagnostic directions.

[0010] Preferably, after determining whether the candidate optimization diagnostic direction set is an invalid set, the method further includes: If the set of candidate optimized diagnostic directions is invalid, then data from an authoritative medical case database will be retrieved for each source diagnostic direction. The supplementary diagnostic direction formed by combining each source diagnostic direction with the corresponding authoritative case database data is used as the optimized diagnostic direction corresponding to each source diagnostic direction.

[0011] Preferably, based on the preliminary symptom assessment results and the symptom supplement checklist, the user's supplementary symptom details are collected, including: Based on a preset guidance strategy, a layered guidance script is designed for ambiguous symptom entries in the symptom supplementation checklist, and the user's supplementary symptom details are collected through the layered guidance script.

[0012] Preferably, by combining the supplementary symptom details, updating the user-specific symptom atlas and the etiology matching degree, and optimizing the preliminary symptom judgment result, including: Based on a preset conflict verification strategy, the supplementary symptom details are cross-verified with past medical history, conflicting feature nodes in the user-specific symptom map are corrected, and the etiology fit is adjusted based on the corrected symptom map, thereby optimizing the preliminary symptom judgment results.

[0013] Preferably, based on the optimized symptom assessment results, a complete consultation conclusion is generated, including: Based on the preset complication decomposition strategy, the main symptom features and potential complication features in the optimized symptom judgment results are decomposed and copied to the diagnostic analysis module. The boundaries of this decomposition are marked, and the complication association features obtained from subsequent verification are added to the potential complication features. Repeat the above steps of splitting and copying the main symptom features and potential complication features in the optimized symptom judgment results to the diagnostic analysis module, marking the boundaries of this split, and adding the complication association features obtained from subsequent verification to the potential complication features, until only the core main symptom features remain to be analyzed in detail; The remaining core symptoms are imported into the diagnostic analysis module. Redundant feature data generated during the splitting process is removed. The imported symptoms, potential complication features, and related features are logically concatenated. After concatenation, corresponding symptom analysis, targeted examination suggestions, and graded intervention plans are generated and integrated into a complete diagnosis conclusion.

[0014] Compared with the prior art, the beneficial effects of the present invention are: (1) By extracting the core features of symptoms, dissecting the typical symptoms, atypical manifestations and complication elements of the disease, and combining the time sequence and correlation of symptoms, this invention improves the depth and accuracy of disease matching; a comprehensive understanding of symptoms helps to eliminate irrelevant factors and lock in possible causes. (2) By calculating the feature matching degree and the etiology fit degree, the intelligent system can generate preliminary disease judgment results and continuously optimize the judgment results based on the details supplemented by the user; with the help of this iterative optimization process, the accuracy of diagnosis can be significantly improved and misdiagnosis and missed diagnosis can be reduced. (3) By designing layered guidance dialogue and conflict verification strategy, the intelligent consultation system can correct conflict nodes in the symptom map in real time when the user supplements symptom information, thereby updating the etiology fit and ensuring that the diagnosis results always reflect the latest and most complete symptom information. (4) By separating and verifying the main symptoms and complications step by step, the system can identify possible complications in advance and generate targeted examination suggestions and intervention plans, which can effectively help doctors discover potential risks and take preventive measures. Attached Figure Description

[0015] Figure 1 This is a schematic flowchart of the overall method in one embodiment of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Example 1, please refer to Figure 1 This invention provides a technical solution: a large-scale intelligent medical consultation model based on AI algorithms, comprising: S1. Obtain basic information about user consultations and professional knowledge base data in the medical field. The basic information about user consultations includes symptoms, lifestyle habits and medical history. The professional knowledge base data in the medical field includes disease characteristics, treatment guidelines and drug compatibility information. S2. Based on the user's basic consultation information and medical professional knowledge base data, extract the core features of the user's symptoms and build a user-specific symptom map and a standardized disease feature library; S3. Based on the user-specific symptom map and the standardized disease feature library, calculate the feature matching degree and etiological fit between the user's symptoms and each standardized disease. S4. Based on feature matching degree and etiology fit degree, generate preliminary disease judgment results and symptom supplementary verification list; S5. Based on the preliminary symptom diagnosis results and the symptom supplement checklist, collect details of the user's supplementary symptoms; combine the supplementary symptom details to update the user-specific symptom atlas and etiology matching, and optimize the preliminary symptom diagnosis results; S6. Based on the optimized symptom diagnosis results, generate a complete consultation conclusion, which includes symptom analysis, examination suggestions, and intervention plan.

[0018] It's important to note that, firstly, the system needs to collect user symptoms such as headache, fever, and nausea; lifestyle habits such as diet, exercise, and sleep patterns; and medical history such as any similar medical history and medications used. This information helps understand the user's health status and potential risks. The medical expertise database includes disease characteristics, treatment guidelines, and medication compatibility information. Disease characteristics refer to common symptoms of each disease, such as polyuria and polydipsia in diabetes. Treatment guidelines refer to diagnostic criteria and treatment methods for different diseases, while medication compatibility information refers to the medications that should be used for different diseases. Based on the symptoms and information provided by the user, the system will extract the core features of the symptoms. For example, if the user mentions persistent headaches and blurred vision, the system will mark these... The system compares user-specific symptoms with the core features of known diseases. A user-specific symptom atlas is a personalized atlas generated based on the user's specific symptoms, lifestyle habits, medical history, and other information, helping to record and visualize the user's health status. A standardized disease feature library is a set of standardized descriptions of known diseases, including symptoms, possible causes, and common treatments, used by the system for comparison. The system compares the user's personalized symptom atlas with the standardized disease feature library. By calculating symptom matching degree and etiology fit degree, the system can determine which diseases best match the user's symptom presentation. Feature matching degree measures the similarity between the user's symptoms and the typical symptoms of a disease; etiology fit degree measures the degree of fit between the user's lifestyle habits, medical history, and other background information and the likelihood of the disease occurring. Based on feature matching and etiological fit, the system provides a preliminary diagnosis, identifying the most likely diseases and providing a symptom supplement checklist. This checklist is for further verification and confirmation, requiring users to provide more detailed information, such as whether there are other related symptoms. Users provide more symptom details based on the checklist, and the system reassesses the symptom profile and etiological fit. This step helps to further refine the diagnosis, ruling out irrelevant diseases or improving the match rate for a particular disease. Based on the optimized symptom profile and etiological fit, a detailed symptom analysis is generated, explaining the relationship between the cause and symptoms. Based on the symptom analysis, the system suggests necessary examinations, such as blood tests and imaging studies, to further confirm the diagnosis. Finally, based on the final diagnosis and examination recommendations, the system generates a targeted intervention plan, including medication and lifestyle modifications.

[0019] In one optional embodiment, based on user consultation information and medical professional knowledge base data, core features of user symptoms are extracted to construct a user-specific symptom atlas and a standardized disease feature library, including: Based on the user's basic information from the consultation, the symptom occurrence characteristics, symptom association characteristics, and symptom persistence characteristics are extracted to obtain the core characteristics of the user's symptoms. Based on the medical professional knowledge base data, typical symptom elements, atypical manifestation elements and complication-related elements of various diseases are decomposed to obtain a standardized disease feature set; Integrate core features of user symptoms to build a user-specific symptom atlas, and integrate standardized disease feature sets to construct a standardized disease feature library.

[0020] It should be noted that the system needs to extract key features from the symptom information provided by the user to help understand the nature of the symptoms. These features can be categorized as follows: Symptom occurrence characteristics, referring to the time, frequency, and conditions of symptom onset; for example, whether the headache occurs in the morning, whether it is related to diet or exercise, and whether it occurs suddenly; symptom association characteristics, referring to the interrelationships or coexistence of symptoms; for example, whether the headache occurs together with symptoms such as blurred vision, nausea, and vomiting, which helps reveal the type and extent of the illness; symptom persistence characteristics, referring to the duration, changes, and progression of symptoms; for example, whether the headache is intermittent or continuous, and whether the symptoms gradually worsen, this information helps determine the acuteness or chronicity or severity of the illness; and core user symptoms. The features are obtained through analysis of the above dimensions; these core features will serve as the basis for further analysis and diagnosis. Detailed symptom information of various diseases is extracted from the medical professional knowledge base, and the symptom elements are broken down into several dimensions to form a standardized set of disease feature characteristics. Specifically, they can be divided into the following categories: Typical symptom elements are the most common and iconic symptoms of the disease, which can usually directly help doctors make a preliminary diagnosis; for example, the typical symptoms of influenza are fever, fatigue, muscle pain, and headache; Atypical manifestation elements refer to the symptoms that some diseases may present differently from common symptoms, or the symptoms may be atypical in some individuals; for example, some diabetic patients may not have obvious polyuria or polydipsia, or even obvious weight loss. Some illnesses may be accompanied by complications, and the system needs to extract relevant information about these complications to help comprehensively assess the patient's health status. For example, complications of heart disease may include hypertension, heart failure, and arrhythmia. By breaking down and organizing these different symptom elements, a standardized set of symptom features is formed, which provides a standardized reference framework for subsequent symptom comparison and etiology inference. The system combines the user's core symptom features with the standardized set of symptom features to form a user-specific symptom atlas. Specifically, the symptom atlas is a diagram composed of multiple symptom features, showing various aspects of the user's health status. It can reflect the user's symptom occurrence characteristics, correlation characteristics, and persistence characteristics, and infer the user's potential health problems based on this information. This symptom atlas is personalized, tailored specifically for each user, reflecting the user's specific symptom manifestations. The system will automatically... The system dynamically updates users' symptom profiles, continuously optimizing as more symptom information is added and changes occur. It integrates extensive medical expertise, constructing a standardized symptom feature database based on typical, atypical, and complication symptoms. This database contains complete symptom information for various diseases, serving as a core reference for medical diagnosis. It has several key features: each symptom element in the database is standardized, ensuring uniform formatting for different diseases, facilitating subsequent comparison and analysis; it comprehensively includes symptom elements for both common and rare diseases, with detailed classifications of typical symptoms, atypical manifestations, and complications for each disease; this database helps the system accurately match diseases by comparing users' symptom profiles; and the standardized symptom feature database is continuously updated as medical research progresses and new cases are discovered, ensuring the scientific rigor and forward-looking nature of diagnosis and treatment.

[0021] In an optional embodiment, based on a user-specific symptom atlas and a standardized symptom feature library, the feature matching degree and etiological fit between the user's symptoms and each standardized symptom are calculated, including: Map the symptom nodes in the user-specific symptom graph to the typical symptom attributes in the standardized disease feature library; Calculate the symptom entity overlap ratio, symptom association edge strength similarity, and time series matching degree; The feature matching degree between user symptoms and each standardized disease is determined by integrating the overlap ratio of symptom entities, the strength similarity of symptom association edges, and the time series matching degree. Extract external influencing factors, symptom time series, and associated edge directions from the user's personalized symptom atlas; Match the etiological transmission pathways, causative types, and pathological mechanisms in the standardized symptom feature database; analyze the consistency of etiological pathways, the correlation of causative factors, and the temporal fit between symptom development and etiological transmission; By integrating the consistency of etiological pathways, the correlation of triggers, and the temporal fit between symptom development and etiological transmission, the etiological fit between user symptoms and various standardized diseases can be determined.

[0022] It's important to note that the system maps user-provided symptom information to typical symptom attributes in a standardized disease feature library—that is, it matches them. Specifically, the user-specific symptom atlas is the user's personalized symptom dataset, where each symptom and manifestation is called a symptom node. The standardized disease feature library contains typical symptom attributes for various diseases, such as the standard manifestations of specific diseases; for example, the typical symptoms of influenza are fever, cough, and fatigue. The system needs to compare each symptom node with the typical symptoms in the disease feature library to find a match. For example, if a user mentions a headache, the system will compare it with the typical symptom "headache" in the standard disease feature library to determine if it matches. Matching exists; Symptom entity overlap ratio refers to how many symptoms in a user's symptom profile overlap with symptoms in a standardized disease feature library; for example, if a user's symptom profile lists fever, headache, and muscle pain, and influenza in the standardized disease library also has these symptoms, then their overlap ratio is high; Symptom association edge strength similarity: Symptoms in the symptom profile are usually related to each other, for example, fever and headache may occur together; "Association edge" refers to the relationship between these symptoms; by calculating the strength of the association between user symptoms, such as whether they occur simultaneously or are causally related, it can be assessed whether these associations are similar to the symptoms of standard diseases; Time-series matching: Symptoms typically exhibit temporal regularity and sequence. Time-series matching assesses whether the temporal development trend of symptoms matches the temporal pattern of a specific disease. For example, a cold may initially present as a sore throat, followed by a cough and other symptoms within hours or days. Using these three indicators, the system can quantify the matching degree between user symptoms and standardized diseases. Integrating the results of the three similarity calculations mentioned above yields the overall matching degree between user symptoms and standardized diseases. By weighting the overlap ratio, the strength of the association edge, and the time series, the system generates a "matching score" for each standardized disease. This score reflects the similarity between the user's symptoms and the disease, thus assisting doctors. The system may determine the most probable diagnosis; analyze the external influencing factors of user symptoms and their temporal development; specific operations include: user symptoms may be influenced by various external factors, such as lifestyle habits, environment, or other health problems; these external factors can help the system analyze the causes of symptom triggers; further analyze the temporal progression of symptoms, including the process from symptom onset to aggravation and relief; for example, if a user reports that their symptoms are gradually worsening, this may indicate an acute onset of some disease; the association edges in the symptom graph not only represent the strength of the relationship between symptoms but also their directionality; some symptoms may be causally related to other symptoms, such as headaches being caused by fever; analyzing these directional relationships can help determine the development path of symptoms; The system compares and matches the time series of user symptoms, external influencing factors, and the direction of related edges with the etiology transmission paths, trigger types, and pathological mechanisms in a standardized disease feature database. Specifically: the etiology transmission path refers to the causal chain from the initial stage of a disease to its development; for example, some infectious diseases may spread from a local infection to the whole body. The system needs to analyze the symptom development path in the user's symptom atlas to match this causal transmission path. Trigger types: each disease usually has different triggers, such as infection, genetics, and lifestyle. By analyzing external influencing factors such as the user's lifestyle habits and environmental changes, the system can identify possible triggers and... The system matches the symptoms with triggers in a standard symptom feature database; pathological mechanisms refer to the physiological and biological mechanisms behind the symptoms; for example, some diseases may be caused by an inflammatory response triggered by an abnormal immune system; the system needs to analyze the user's symptoms and external factors to infer possible pathological mechanisms; analyze the temporal fit of the etiology transmission path, trigger type, and symptom time series; that is, the system needs to check whether the user's symptom development is consistent with the etiology transmission path, whether the trigger is related to the symptom changes, and whether the symptom development conforms to the temporal sequence of known etiologies; determine whether the development of the user's symptoms conforms to the causal chain of known diseases, and confirm whether the evolution of symptoms conforms to the logic of etiology transmission; The system examines whether the triggers are related to the symptoms to help determine the source of the symptoms; whether the occurrence, aggravation, or relief of the symptoms matches the time sequence of the spread of known causes; if the symptoms highly match the development pattern of known causes, it indicates that the cause may match the user's symptoms; finally, the system integrates these analysis results to determine the degree of causal fit between the user's symptoms and various standardized diseases; if the causal path, trigger type, and symptom development time sequence match well, the system can infer that the user's symptoms match the cause of a specific disease, thereby improving the accuracy of diagnosis.

[0023] In an optional embodiment, based on feature matching degree and etiological fit, a preliminary symptom diagnosis result and a supplementary symptom checklist are generated, including: Based on feature matching degree and etiological fit degree, determine the set of source diagnostic directions and the set of candidate optimized diagnostic directions; Determine whether the set of candidate optimization diagnostic directions is an invalid set; If the candidate optimized diagnostic direction set is not an invalid set, then based on the candidate optimized diagnostic direction set, the diagnostic direction that is most closely related to the cause of each source diagnostic direction in the source diagnostic direction set and has the highest feature matching degree is selected, and the optimized diagnostic direction corresponding to each source diagnostic direction is obtained and integrated into the preliminary disease judgment result.

[0024] It should be noted that the system determines two sets of diagnostic directions based on the feature matching degree and etiology fit calculated in previous steps: the source diagnostic direction set refers to the set of disease diagnostic directions initially determined based on user symptoms, etiology paths, and other data; the source diagnostic directions refer to the possible symptom directions that the system considers after preliminary analysis of user symptoms; these diagnostic directions may be the few disease directions that best match the symptom matching degree and etiology fit; the candidate optimized diagnostic direction set is a set of potential candidates further filtered from the source diagnostic direction set; these candidate directions are based on factors such as the etiology's transmission path, triggering type, and time series, taking into account possible optimization directions; for example, if new data supports the development of certain etiologies or symptoms, the candidate optimized directions may point to more precise or more consistent symptom developments; After obtaining the set of candidate optimized diagnostic directions, the system needs to verify their validity. An invalid set is typically defined as one where the system has not found sufficient evidence to support these candidate directions, or where all candidate directions do not closely match the user's symptoms or causes, failing to provide effective diagnostic clues. For example, these candidate directions may not be effective diagnostic paths due to low symptom feature matching, inconsistent etiological pathways, or extremely rare conditions. If the set of candidate optimized diagnostic directions is empty, or if analysis reveals that all candidate directions do not meet the criteria, the system will determine it as an invalid set, requiring a return to the previous analysis steps or the search for other data support. If the set of candidate optimized diagnostic directions is valid, the system will optimize and filter it based on... The system considers the following factors: It examines the causal correlation between each source diagnostic direction and candidate optimized diagnostic directions. Causal correlation measures the similarity or matching degree between the causal factors of the source diagnostic direction and those of the candidate optimized directions. The system prioritizes candidate optimized directions whose causal transmission paths and triggering types are highly consistent with the source diagnostic direction. In addition to causal matching, the system also considers feature matching, i.e., the similarity of symptom presentation. If the symptom features of a candidate optimized diagnostic direction highly match the user's symptoms, then this direction is considered a more promising optimized diagnostic direction. Through these two criteria, the system selects the optimized diagnostic direction corresponding to each source diagnostic direction—that is, the diagnostic direction that the system considers most likely to be consistent with the source diagnostic direction and has more accurate symptom features. The system integrates each source diagnostic direction with its optimized diagnostic direction to generate preliminary symptom judgment results. The optimized diagnostic direction corresponding to each source diagnostic direction means that the system no longer relies solely on the preliminary diagnosis of the source diagnostic direction, but further improves accuracy through optimized candidate diagnostic directions. These optimized diagnostic directions may be directions that are more consistent with the current symptom presentation, etiological path, and triggering type. After integrating all optimized diagnostic directions, the system obtains a preliminary set of diagnostic results. These results are not the final diagnosis, but rather potential disease directions optimized based on multi-dimensional information such as symptoms, etiology, and time series. Doctors or the diagnostic system will further confirm or adjust the diagnostic directions based on these results.

[0025] In an optional embodiment, determining the source diagnostic direction set and the candidate optimized diagnostic direction set based on feature matching degree and etiological fit degree includes: Based on feature matching degree and etiological fit, the combination of diagnostic directions that does not meet the fit standard is selected to obtain the source diagnostic direction set. Based on feature matching degree and etiological fit, determine whether there are diagnostic directions whose fit exceeds the optimization standard; If there are diagnostic directions whose fit exceeds the optimization criteria, then these diagnostic directions are combined to obtain a set of candidate optimized diagnostic directions.

[0026] It should be noted that diagnostic directions that do not adequately match the patient's symptoms and background information are filtered out. For example, if the system analysis finds that the feature matching degree or etiological fit of certain potential diseases is far below the set standard value, these diagnostic directions will be considered unsuitable for the patient and cannot serve as effective diagnostic clues. Therefore, these directions that do not meet the standard will be excluded, and the final set of source diagnostic directions will be retained, which are those preliminary diagnostic directions that are relatively suitable for the patient's situation and have been filtered through the matching criteria. After obtaining the set of source diagnostic directions, the system needs to further evaluate the fit of each diagnostic direction to see if the fit of some directions has exceeded the optimization criteria. The system sets a matching standard, usually a threshold, representing an ideal or optimal value for feature matching and etiological matching. For example, if the feature matching of a certain diagnostic direction has reached over 90% and the etiological matching is also very high, then it may exceed the optimization standard. The process of determining whether there are diagnostic directions with matching exceeding the optimization standard aims to identify those with excessively high matching, meaning that these diagnostic directions closely match the patient's symptoms, etiology, and other information, and are more accurate than ordinary source diagnostic directions; their features and etiology match the patient with a degree exceeding the preset optimization standard. If, in the second step of the assessment, it is found that the fit of certain diagnostic directions exceeds the set optimization criteria, these directions will be selected to form a new set of diagnostic directions. The candidate optimized diagnostic direction set refers to the combination of diagnostic directions whose fit exceeds the optimization criteria. They are more consistent with the patient's characteristics than other directions in the source diagnostic direction set and may represent a more accurate diagnostic result. The system will combine these diagnostic directions to obtain a candidate optimized diagnostic direction set. These optimized directions will serve as potential "final diagnostic directions" to provide a basis for subsequent clinical decision-making.

[0027] In an optional embodiment, after determining whether the set of candidate optimization diagnostic directions is an invalid set, the method further includes: If the set of candidate optimized diagnostic directions is invalid, then retrieve authoritative case database data in the medical field for each source diagnostic direction; The supplementary diagnostic directions formed by combining each source diagnostic direction with the corresponding authoritative case database data are used as the optimized diagnostic directions for each source diagnostic direction.

[0028] It should be noted that in the previous steps, the diagnostic directions selected through feature matching and etiological fit did not meet the optimization criteria, or the system did not find directions with a fit exceeding the optimization criteria. This may be due to unclear patient symptoms, or the thresholds for feature matching and etiological fit being set too high, resulting in no suitable optimization direction. The presence of invalid sets indicates that the system did not find suitable candidate optimization directions, and therefore cannot obtain a clear optimal diagnostic direction from the existing data. Source diagnostic directions refer to those that meet certain criteria through preliminary screening but do not meet the optimization criteria. These directions still have some relevance and value, but may not be directly used as optimized diagnostic directions due to insufficient fit. Authoritative case database data in the medical field refers to real cases, research reports, clinical trial results, etc., provided by medical experts, clinicians, or authoritative medical institutions. This data has high credibility and can provide more disease diagnosis information, treatment experience, related symptoms, and treatment plans. By calling this data, the system can provide supplementary information for source diagnostic directions. By combining data from an authoritative case database, the system will conduct supplementary analysis on each source diagnostic direction. This supplementary analysis may include: medical history information: finding historical records of similar patients through the case database to understand their symptoms and diagnostic processes; treatment effects: referencing treatment plans and effects of similar conditions to provide possible treatment pathways for the source diagnostic direction; clinical practice experience: integrating expert opinions and experience based on authoritative data to verify the effectiveness of the source diagnostic direction; disease progression and complications: understanding the disease progression of similar patients through the case database to help determine the rationality and potential risks of the source diagnostic direction. After supplementing information with data from the authoritative case database, each source diagnostic direction will form an optimized diagnostic direction. This optimized direction is more referential and accurate than the original source diagnostic direction because it combines real case data and expert practice experience. An optimized diagnostic direction refers to a diagnostic direction that has been further confirmed and optimized based on the original source diagnostic direction by introducing data from the authoritative case database. The optimized diagnostic direction not only considers the patient's symptoms but also incorporates broader medical knowledge, making the diagnostic results more reliable.

[0029] In an optional embodiment, based on the preliminary symptom assessment results and the symptom supplement checklist, details of the user's supplementary symptoms are collected, including: Based on a pre-defined guidance strategy, a layered guidance script is designed for ambiguous symptom items in the symptom supplementation checklist, and the user is asked to provide supplementary symptom details through the layered guidance script.

[0030] It's important to note that pre-defined guidance strategies refer to a series of guidance principles and methods established by developers during the system or application design phase. These strategies aim to help the system interact with users more effectively, ensuring accurate and efficient information collection. Guidance strategies can include: how to guide users through dialogue, under what circumstances to ask questions, how to adjust the way questions are asked (e.g., open-ended or closed-ended questions), and how to handle ambiguous answers from users. Ambiguous symptom items refer to symptoms that lack sufficient specific information and cannot be clearly described; for example, a user might say, "I feel a little unwell," but without further details such as where the discomfort is, what it feels like, or when it started. These items require further clarification and the collection of more information through further questioning to better assist doctors or the system in making an accurate diagnosis. Layered guidance dialogue refers to designing a progressively deeper, hierarchical set of questions or dialogue to guide users to describe their symptoms more clearly and in detail. Layering means collecting more information by progressively refining the approach based on the ambiguity of the symptoms; for example, for the ambiguous symptom of "headache," one could first ask the user, "Where does the headache occur?" Then, one could continue by asking, "Is it a persistent or intermittent headache?" and further, "Can you describe the intensity of the headache?" etc. When designing tiered guidance scripts, the following principles are typically followed: From broad to specific: Start by asking broad questions to allow users to freely express their symptoms, then gradually guide them to more specific details; Concise and easy to understand: Ensure the guidance scripts are simple and easy to comprehend, avoiding confusion or pressure for users; Respect user feelings: Avoid making users feel uncomfortable or unwilling to continue describing their symptoms, and strive to make the questions seem natural and empathetic; Flexibility: Adjust the questioning style based on the user's answers; some symptom items may require special explanations or hints; The purpose of collecting additional symptom details from users is to use guidance scripts to get them to provide more specific information. This helps the system or doctor understand the user's condition more accurately and avoids potential information gaps during the diagnosis process.

[0031] In one optional embodiment, by incorporating supplementary symptom details, the user-specific symptom atlas and etiology matching are updated to optimize the preliminary symptom diagnosis results, including: Based on a preset conflict verification strategy, supplementary symptom details are cross-verified with past medical history, conflicting feature nodes in the user's exclusive symptom map are corrected, and the etiology fit is adjusted based on the corrected symptom map, thereby optimizing the preliminary symptom judgment results.

[0032] It should be noted that a pre-defined conflict verification strategy refers to a set of verification rules established in advance during the diagnostic system design phase. This set is used to identify and handle potential contradictions or conflicts between information provided by the user. The goal of this strategy is to detect and resolve inconsistencies or conflicts in user symptom descriptions, medical histories, or other medical information, thereby ensuring the accuracy and consistency of the generated symptom atlas. Conflict verification strategies may include: symptom and medical history matching rules, symptom models of common diseases, and known medical common sense, used to identify and mark potentially contradictory symptom or medical history entries. Supplementary symptom details are additional information provided by the user during the diagnostic process, which may be a more detailed description of the initial symptoms. Detailed or more specific supplementary information; past medical history refers to the user's historical medical records, including past illnesses, surgeries, allergies, family history, etc.; the cross-checking process compares the user's new symptom details with their past medical history to check for potential conflicts; for example, if a user provides "headache" as a new symptom, but has a record of brain disease in their past medical history, the system will automatically check whether the two pieces of information are consistent and whether there is a conflict; through cross-checking, some inconsistent or contradictory symptom information can be found, for example: if a user without a relevant medical history reports symptoms that are inconsistent with that medical history, the system will mark the possible conflict for further confirmation or adjustment; A user-specific symptom atlas is a personalized disease feature map built based on a user's symptoms, medical history, and other information. This atlas is typically generated by mapping various symptoms, past medical history, and physical examination results onto a multi-dimensional medical model. Conflicting feature nodes are points in the symptom atlas where certain symptoms or medical history information is inconsistent with or conflicts with other nodes in the user's atlas (such as existing medical history and symptoms). These inconsistent points are marked as "conflicting nodes." The system uses a pre-defined conflict verification strategy to identify and correct these conflicting nodes, ensuring that the information recorded in the atlas is not contradictory. For example, if a symptom does not match the disease type recorded in the past medical history, the system will automatically correct or mark it to avoid misdiagnosis. After the symptom atlas is corrected, the system will reassess the degree of match between each possible cause and the current symptom. Fit refers to the degree to which a disease matches a user's current symptoms. After conflict verification and correction, the feature nodes in the symptom map become more accurate and consistent, allowing the system to more precisely calculate the fit between each possible cause and the user's symptoms. For example, the system might find that a certain symptom actually better matches the presentation of a certain disease, while the fit of other possible causes decreases. Therefore, after adjustment, the system can arrive at a more appropriate diagnosis. The preliminary symptom judgment is based on the user's input symptoms and historical information, and is a preliminary inference of possible diseases or health problems derived from the system's initial analysis. By optimizing the fit of causes, the system can improve the accuracy of the preliminary symptom judgment. For example, the symptom map after conflict verification and correction increases the fit of some potential causes and decreases the fit of other irrelevant causes, ultimately allowing the system to provide a more accurate preliminary symptom judgment.

[0033] In an optional embodiment, a complete medical history is generated based on the optimized symptom assessment results, including: Based on the preset complication decomposition strategy, the main symptom features and potential complication features in the optimized symptom judgment results are decomposed and copied to the diagnostic analysis module. The boundaries of this decomposition are marked, and the complication association features obtained from subsequent verification are added to the potential complication features. Repeat the above steps of splitting and copying the main symptom features and potential complication features in the optimized symptom judgment results to the diagnostic analysis module, marking the boundaries of this split, and adding the complication association features obtained from subsequent verification to the potential complication features, until only the core main symptom features remain to be analyzed in detail; The remaining core symptoms are imported into the diagnostic analysis module. Redundant feature data generated during the splitting process is removed. The imported symptoms, potential complication features, and related features are logically concatenated. After concatenation, corresponding symptom analysis, targeted examination suggestions, and graded intervention plans are generated and integrated into a complete diagnosis conclusion.

[0034] It should be noted that the pre-defined complication decomposition strategy refers to a set of rules or methods pre-set in the diagnostic system to identify and decompose complication features that may be related to the primary symptom. These rules help to progressively analyze the complex information in the symptom, enabling a more detailed and accurate diagnosis. The primary symptom features refer to the patient's current major health problems or symptoms, such as fever or headache. Potential complication features refer to other diseases or symptoms that may be related to the primary symptom; these complications may not be fully manifested but may become crucial in subsequent diagnosis. At this stage, the system decomposes the primary symptom features and potential complication features and copies them to the diagnostic analysis module as the basis for further analysis and verification. During the segmentation process, the system marks and labels each feature node to clearly identify which are primary symptom features and which are complication features, ensuring clear differentiation during subsequent processing. In the subsequent diagnosis process, the system confirms complication features through further data verification (such as new symptoms, test results, etc.). If symptoms or data related to potential complications are found, this information is added to the potential complication features to ensure that complications are updated in a timely and complete manner. Once the system has decomposed and initially labeled the primary symptom features and potential complication features, the system will continuously repeat this process, further decomposing the features of each complication until all potential complications related to the primary symptom have been analyzed in detail. The system continues to break down the features of potential complications until all complication-related information has been processed. At this point, what remains are the core primary symptom features, which are usually the most critical and direct symptoms and have not yet been analyzed in detail. After breaking down the complication features, what remains are the core features directly related to the primary symptom. These features are then imported into the diagnostic analysis module for final detailed analysis. During the breakdown process, some redundant feature data may be generated, which is not needed for the final diagnosis. The system automatically cleans up this redundant data to avoid confusion and improve computational efficiency. The system integrates the primary symptom features, potential complication features, and other relevant features to form a logical chain. This means that the system will combine relevant information based on the correlation between features to derive a more accurate symptom model; the system will analyze all symptoms and features, providing explanations or descriptions of the symptom and explaining the possible causes behind the symptoms; based on the analysis results, the system will generate specific examination suggestions, such as recommending a certain medical test or further medical history investigation; based on the severity of the symptom and risk assessment, the system will propose appropriate intervention plans, which may include drug treatment, surgery, health management suggestions, etc.; finally, all this information will be integrated into a complete consultation conclusion, provided to the doctor or patient; these conclusions include diagnosis, recommended examinations, intervention plans, and possible follow-up steps to help patients or doctors make decisions.

[0035] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. A large-scale intelligent medical consultation model technology method based on AI algorithms, characterized in that, include: Acquire basic information about user consultations and medical professional knowledge base data, wherein the basic information about user consultations includes symptom presentation, lifestyle habits and medical history, and the medical professional knowledge base data includes disease characteristics, treatment guidelines and drug compatibility information; Based on the user's basic consultation information and medical professional knowledge base data, the core features of the user's symptoms are extracted to construct a user-specific symptom atlas and a standardized disease feature library. Based on the user-specific symptom atlas and standardized disease feature library, the feature matching degree and etiological consistency between user symptoms and each standardized disease are calculated. Based on the feature matching degree and etiology fit degree, a preliminary symptom judgment result and a supplementary symptom verification list are generated. Based on the preliminary symptom diagnosis results and the symptom supplement checklist, collect the user's supplementary symptom details; combine the supplementary symptom details to update the user-specific symptom atlas and the etiology matching degree, and optimize the preliminary symptom diagnosis results; Based on the optimized symptom diagnosis, a complete consultation conclusion is generated, which includes symptom analysis, examination recommendations, and intervention plan.

2. The intelligent medical consultation model technology method based on AI algorithm according to claim 1, characterized in that, Based on the user's basic consultation information and medical professional knowledge base data, core features of user symptoms are extracted to construct a user-specific symptom atlas and a standardized disease feature library, including: Based on the user's basic consultation information, symptom occurrence features, symptom association features, and symptom persistence features are extracted to obtain the core features of the user's symptoms. Based on the medical professional knowledge base data, typical symptom elements, atypical manifestation elements, and complication-related elements of various diseases are decomposed to obtain a standardized disease feature set; By integrating the core features of the user symptoms, a user-specific symptom atlas is constructed, and by integrating the standardized disease feature set, a standardized disease feature library is built.

3. The intelligent medical consultation model technology method based on AI algorithm according to claim 2, characterized in that, Based on the user-specific symptom atlas and standardized disease feature library, the feature matching degree and etiological consistency between user symptoms and each standardized disease are calculated, including: Map the symptom nodes in the user-specific symptom atlas to the typical symptom attributes in the standardized disease feature library; Calculate the symptom entity overlap ratio, symptom association edge strength similarity, and time series matching degree; The feature matching degree between user symptoms and each standardized disease is determined by integrating the overlap ratio of symptom entities, the similarity of the strength of symptom association edges, and the time series matching degree. Extract external influencing factors, symptom time series, and associated edge directions from the user-specific symptom atlas; Match the etiological transmission pathways, triggering factors, and pathological mechanisms in the standardized symptom feature database; analyze the consistency of etiological pathways, the correlation of triggering factors, and the temporal fit between symptom development and etiological transmission; By integrating the consistency of the etiological pathways, the correlation of the triggers, and the temporal fit between the symptom development and the etiological spread, the etiological fit between user symptoms and each standardized disease is determined.

4. The intelligent medical consultation model technology method based on AI algorithm according to claim 3, characterized in that, Based on the feature matching degree and etiological fit, a preliminary symptom diagnosis result and a supplementary symptom checklist are generated, including: Based on the feature matching degree and etiological fit degree, the set of source diagnostic directions and the set of candidate optimized diagnostic directions are determined; Determine whether the set of candidate optimized diagnostic directions is an invalid set; If the candidate optimized diagnostic direction set is not an invalid set, then based on the candidate optimized diagnostic direction set, the diagnostic direction that is most closely related to the cause of each source diagnostic direction in the source diagnostic direction set and has the highest feature matching degree is selected, and the optimized diagnostic direction corresponding to each source diagnostic direction is obtained and integrated into the preliminary disease judgment result.

5. The intelligent medical consultation model technology method based on AI algorithm according to claim 4, characterized in that, Based on the feature matching degree and etiological fit, a set of source diagnostic directions and a set of candidate optimized diagnostic directions are determined, including: Based on the feature matching degree and etiological fit, the diagnostic direction combinations that do not meet the fit criteria are screened out to obtain the source diagnostic direction set. Based on the feature matching degree and etiological fit, determine whether there are diagnostic directions whose fit exceeds the optimization standard; If there are diagnostic directions whose fit exceeds the optimization criteria, then these diagnostic directions that exceed the optimization criteria are combined to obtain a set of candidate optimized diagnostic directions.

6. The intelligent medical consultation model technology method based on AI algorithm according to claim 5, characterized in that, After determining whether the candidate optimization diagnostic direction set is an invalid set, the method further includes: If the set of candidate optimized diagnostic directions is invalid, then data from an authoritative medical case database will be retrieved for each source diagnostic direction. The supplementary diagnostic direction formed by combining each source diagnostic direction with the corresponding authoritative case database data is used as the optimized diagnostic direction corresponding to each source diagnostic direction.

7. The intelligent medical consultation model technology method based on AI algorithm according to claim 6, characterized in that, Based on the preliminary symptom assessment results and the symptom supplement checklist, details of the user's supplementary symptoms are collected, including: Based on a preset guidance strategy, a layered guidance script is designed for ambiguous symptom entries in the symptom supplementation checklist, and the user's supplementary symptom details are collected through the layered guidance script.

8. The intelligent medical consultation model technology method based on AI algorithm according to claim 7, characterized in that, Based on the supplementary symptom details, the user-specific symptom atlas and the etiology matching degree are updated to optimize the preliminary symptom diagnosis results, including: Based on a preset conflict verification strategy, the supplementary symptom details are cross-verified with past medical history, conflicting feature nodes in the user-specific symptom map are corrected, and the etiology fit is adjusted based on the corrected symptom map, thereby optimizing the preliminary symptom judgment results.

9. The intelligent medical consultation model technology method based on AI algorithm according to claim 8, characterized in that, Based on the optimized symptom assessment results, a complete medical history is generated, including: Based on the preset complication decomposition strategy, the main symptom features and potential complication features in the optimized symptom judgment results are decomposed and copied to the diagnostic analysis module. The boundaries of this decomposition are marked, and the complication association features obtained from subsequent verification are added to the potential complication features. Repeat the above steps of splitting and copying the main symptom features and potential complication features in the optimized symptom judgment results to the diagnostic analysis module, marking the boundaries of this split, and adding the complication association features obtained from subsequent verification to the potential complication features, until only the core main symptom features remain to be analyzed in detail; The remaining core symptoms are imported into the diagnostic analysis module. Redundant feature data generated during the splitting process is removed. The imported symptoms, potential complication features, and related features are logically concatenated. After concatenation, corresponding symptom analysis, targeted examination suggestions, and graded intervention plans are generated and integrated into a complete diagnosis conclusion.

Images