Guiding type auxiliary evaluation information processing method and system

Abstract

The invention belongs to the field of auxiliary data processing, and particularly relates to a guiding type auxiliary evaluation information processing method and system.The method includes the steps that a matching decision parameter set of an initial target and a standard knowledge base is constructed, and user portraits and decision parameters are optimized step by step in combination with a dynamic step-by-step guiding question and answer strategy; dynamically adjusting the question and answer guide frequency by adopting a probability sequence iteration updating mechanism; a matching target confidence coefficient sequence is calculated based on a decision judgment model, multi-round feedback optimization is achieved through threshold value comparison, when the confidence coefficient reaches the standard, an optimal strategy is output, and when the confidence coefficient does not reach the standard, step-by-step enhancement is executed circularly till a preset round threshold value is reached, and finally accurate decision making of medical evaluation is achieved. A dynamic discrimination process driven by confidence is adopted; the accuracy and reliability of an evaluation result in a complex target scene are remarkably improved, and meanwhile, an abnormal round control mechanism is provided to guarantee the integrity of a processing flow.

Description

Technical Field

[0001] The present invention belongs to the field of auxiliary data processing, and particularly relates to a guided auxiliary evaluation information processing method and system. Background Art

[0002] Although certain progress has been made in the field of medical intelligent consultation in the prior art, there are still common defects. First of all, existing methods mostly rely on preset Q&A templates (CN111159369B) or text analysis with a fixed process (CN113921149B), lacking the ability to dynamically guide complex targets and being difficult to adapt to the diverse needs of clinical consultations. Secondly, most systems adopt single-modal data processing (CN117423480A, CN118352004A), without integrating multi-dimensional medical data, resulting in limited accuracy of target recognition and comprehensiveness of diagnosis and treatment suggestions. Moreover, the update mechanism of the existing medical knowledge base lags behind (CN119181480A), and it is unable to integrate the latest medical research results and clinical guidelines in real time, affecting the timeliness of diagnosis. In addition, the human-computer interaction mode of traditional methods is fixed (CN111159369B, CN118352004A), lacking personalized guiding strategies, with low patient participation and easy omission of key targets. These defects together lead to significant deficiencies in the clinical applicability, diagnostic accuracy, and user experience of existing intelligent consultation systems, and there is an urgent need to build an intelligent diagnosis and treatment method that can achieve dynamic multi-round guidance, multi-modal data fusion, and has self-learning ability. Summary of the Invention

[0003] Aiming at the deficiencies of the prior art, the present invention proposes a guided auxiliary evaluation information processing method and system. The method constructs a matching decision parameter set for the initial target and the standard knowledge base, combines a dynamic step-by-step guided Q&A strategy to gradually optimize the user profile and decision parameters, and adopts a probability sequence iterative update mechanism to dynamically adjust the Q&A guidance frequency; calculates the confidence sequence of the matching target based on the decision discrimination model, and realizes multi-round feedback optimization through threshold comparison: when the confidence reaches the standard, the optimal strategy is output, and when it does not reach the standard, the step-by-step enhancement is looped until the preset round threshold is reached, finally realizing accurate decision-making for medical evaluation; through the collaborative mechanism of the knowledge base and the Q&A strategy, the present application adopts a confidence-driven dynamic discrimination process, effectively solves the problems of insufficient information collection and single decision-making basis in traditional medical assistance, significantly improves the accuracy and reliability of evaluation results in complex target scenarios, and at the same time has an abnormal round control mechanism to ensure the integrity of the processing flow.

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] A guided auxiliary evaluation information processing method, comprising:

[0006] Based on the initial Q&A information, perform a preliminary match with a preset standard knowledge base to obtain an initial matching target sequence decision parameter set;

[0007] Based on the initial matching target sequence decision parameter set, combine a preset step-by-step guided Q&A assistance strategy and synchronously call the standard knowledge base to obtain a user profile and a step-by-step enhanced decision parameter set for the corresponding matching target sequence;

[0008] The step-by-step guided Q&A assistance strategy constructs the Q&A frequency of the matching target sequence in the next round by step-by-step enhancing the initial discrimination probability sequence corresponding to each matching target in the step-by-step enhanced decision parameter set after each round of Q&A update of the matching target sequence;

[0009] Based on the user profile and the step-by-step enhanced decision parameter set of the corresponding matching target sequence, combine a preset decision discrimination model to obtain a confidence sequence corresponding to the matching target sequence;

[0010] Compare the confidence sequence corresponding to the matching target sequence with a preset confidence threshold. If the confidence of at least one matching target is greater than the preset confidence threshold, output the parameter strategy space corresponding to the matching target with the maximum confidence;

[0011] If the confidence of all matching targets is less than the preset confidence threshold, determine whether the current round of the step-by-step guided Q&A assistance process is less than the preset Q&A round threshold. If it is less, repeat the step-by-step guided Q&A assistance and decision discrimination processes. If until the Q&A round is greater than or equal to the preset Q&A round threshold and all the confidences of the corresponding matching target sequence are still less than or equal to the preset confidence threshold, reply that the Q&A is abnormal.

[0012] Specifically, the process of obtaining the initial matching target sequence decision parameter set includes:

[0013] Based on a preset multi-modal preloading model, combine a preset historical information library and user initial input text information to obtain an initial target triple information sequence;

[0014] The initial target triple information sequence includes the target name, the matching target text parameter, and the initial association matching degree between the target name and the matching target text parameter;

[0015] Based on the initial target triple information sequence, combine the preset standard knowledge base and the preset extended phrase correction model, and obtain the initial matching target sequence decision parameter set in descending order of the initial association matching degree.

[0016] Specifically, the initial matching target sequence decision parameter set includes the initial matching target and the corresponding standard target parameter text feature set, the correction accuracy of the target keyword group corresponding to each matching target, and the initial association matching degree sequence corresponding to the initial matching target sequence;

[0017] The extended phrase correction model is obtained by training the matching confidence algorithm in combination with a preset target extended phrase library, and is used to standardize and correct the non-standard target keyword phrases in the initial target triple information sequence, and obtain the correction accuracy of the corresponding target keyword phrases.

[0018] The target extended phrase library is obtained by combining standard target names, non-standard target keyword phrases containing at least one key character in the standard target names, and associated aliases.

[0019] Specifically, the process of obtaining the user portrait and the corresponding step-by-step enhanced decision parameter set includes:

[0020] Based on the initial matching target sequence decision parameter set, in combination with a preset target parameter-portrait association mapping space and first supplementary parameter information, a user portrait and an initial enhanced decision parameter space are obtained.

[0021] The first supplementary parameter information includes the basic information and location information of the user corresponding to the matching target.

[0022] The parameters included in the user portrait are the same as those included in the first supplementary parameter information.

[0023] The target parameter-portrait association mapping space is obtained by pre-training the user portrait in combination with the standard knowledge base and the historical information base through an association matching algorithm.

[0024] The initial enhanced decision parameter space includes the user portrait, the initial matching target sequence arranged in descending order, and the corresponding initial discrimination probability sequence and matching target text parameters.

[0025] Based on the numerical ratio of the initial discrimination probabilities corresponding to all matching targets in the initial enhanced decision parameter space, the second supplementary information query frequency corresponding to each matching target in the initial enhanced decision parameter space is obtained.

[0026] Specifically, the process of obtaining the user portrait and the corresponding step-by-step enhanced decision parameter set further includes:

[0027] Based on the initial enhanced decision parameter space and the standard knowledge base, through a preset text random generation model, in combination with the second supplementary information query frequency corresponding to each matching target, a second loop question-and-answer text set is generated, and the standard knowledge base and the preset extended phrase correction model are synchronously called to correct the generated second loop question-and-answer text set.

[0028] Based on the corrected second loop question-and-answer text set, the initial enhanced decision parameter space is updated to obtain the second initial discrimination probability corresponding to each matching target after the update.

[0029] Specifically, the process of obtaining the user portrait and the corresponding step-by-step enhanced decision parameter set further includes:

[0030] Based on the updated second initial discrimination probability corresponding to each matching target and the initial association mapping threshold, filter and screen the initial matching target sequence to obtain the second matching target sequence after filtering and updating and the corresponding enhanced decision parameter space; repeat the above process of obtaining the second matching target sequence and the corresponding enhanced decision parameter space until all the initial discrimination probabilities in the updated matching target sequence are greater than the initial association mapping threshold and the matching target corresponding to the maximum initial discrimination probability remains unchanged, and obtain the target sequence with step-by-step enhancement completed and the corresponding enhanced decision parameter space.

[0031] Specifically, the target parameter-portrait association mapping space includes three-level mapping; the three-level mapping includes the BMI index-target mapping layer, the age-target mapping layer, and the gender-target mapping layer;

[0032] The construction process of the BMI index-target mapping layer includes:

[0033] In the state where the BMI index is determined, through the association analysis algorithm, combine the BMI index at each level in different regions with the occurrence frequency and target parameter information of the corresponding different target units of time to obtain the association degree between the BMI index at each level in each region and different targets;

[0034] Based on the BMI index at each level in each region, the target name, and the association degree between the BMI index at each level and different targets, construct the BMI index-target mapping layer.

[0035] Specifically, the construction process of the age-target mapping layer includes:

[0036] Based on the association degree between the BMI index at each level in each region and different targets as the weighted weight and the age grade information in each region, obtain the association degree between each age grade in each region and different targets through association analysis;

[0037] The age grade is one age grade every 5 years;

[0038] Based on the age grade in each region, the target name, and the association degree between each age grade in each region and different targets corresponding to different targets, construct the age-target mapping layer;

[0039] The construction process of the gender-target mapping layer includes:

[0040] Based on the association degree between each age level and different goals within each region as the weighted weight, combined with the gender ratio information at each age level within each region, the association degree between each gender and different goals within each region is obtained, and the gender-goal layer is obtained;

[0041] Based on the BMI index-goal mapping layer, age-goal mapping layer, and gender-goal mapping layer, the goal parameter-portrait association mapping space is obtained through a cascading method.

[0042] Specifically, the preset historical information database is constructed from the user's historical and real-time Q&A result information through a knowledge graph and a graph database;

[0043] The historical information database and the standard knowledge database are connected through the extended phrase correction model and the index information constructed based on the association degree obtained from the Q&A information and the standard target parameter information.

[0044] A guided assisted evaluation information processing system includes: an initial matching module, a step-by-step Q&A module, and a discrimination module;

[0045] The discrimination module includes a confidence unit, a secondary discrimination unit, and an inquiry adjustment unit;

[0046] The initial matching module is used to obtain initial Q&A information and perform a preliminary match based on the initial Q&A information combined with a preset standard knowledge database to obtain an initial matching target sequence decision parameter set;

[0047] The step-by-step Q&A module is used to obtain a step-by-step enhanced decision parameter set of the user portrait and the corresponding matching target sequence based on the initial matching target sequence decision parameter set, combined with a preset step-by-step guided Q&A assistance strategy and synchronously calling the standard knowledge database;

[0048] The confidence unit, based on the user portrait and the step-by-step enhanced decision parameter set of the corresponding matching target sequence, combined with a preset decision discrimination model, obtains a confidence sequence corresponding to the matching target sequence;

[0049] The secondary discrimination unit is used to compare the confidence sequence corresponding to the matching target sequence with a preset confidence threshold. If there is at least one confidence of a matching target greater than the preset confidence threshold, the parameter strategy space corresponding to the matching target with the maximum confidence is output;

[0050] If the confidence levels of all matching targets are less than the preset confidence threshold, the inquiry adjustment unit is used to determine whether the current round of the step-by-step guided Q&A assistance process is less than the preset inquiry round threshold. If it is less, the step-by-step guided Q&A assistance and decision-making discrimination process is repeated. If until the inquiry round is greater than or equal to the preset inquiry round threshold, and all the confidence levels of the corresponding matching target sequence are still all less than or equal to the preset confidence threshold, an abnormal Q&A reply is given.

[0051] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0052] Aiming at the deficiencies of the prior art, the present invention constructs a collaborative mechanism between a multi-modal preloading model and a historical case database, converts the user's initial input text into an initial target triple information sequence including a target name, text parameters, and an associated matching degree, and combines an extended phrase correction model to perform a standardized conversion on non-standard target words, effectively solving the problem of the diversity of target expressions and the matching deviation with the knowledge base; based on the dynamic weight superposition mechanism of the three-level target parameter-portrait association mapping space, through regionalized correlation calculation and hierarchical progressive mapping strategies, the accurate quantification of the user's multi-dimensional portrait features and target correlation degree is realized; the dynamic Q&A frequency allocation method based on the discrimination probability ratio in the step-by-step guided Q&A strategy is adopted, combined with the cyclic optimization mechanism of the text random generation model and the extended phrase correction model, to realize the targeting and semantic standardization of supplementary information collection. Through the confidence threshold-driven multi-round iterative discrimination model, using the probability sequence filtering and matching target sequence dynamic screening mechanism in the parameter space update process, the fault tolerance and convergence efficiency of the decision-making process are enhanced. In addition, based on the index connection architecture of the case database and the standard knowledge base of the knowledge graph, combined with the semantic expansion ability of the target extended phrase library, a two-way mapping channel covering the differences between standard terms and user expressions is formed, enhancing the robustness of target matching and the compatibility of the knowledge base from the data collaboration level. Description of the Drawings

[0053] Figure 1 It is a flowchart of a guided assisted evaluation information processing method according to Embodiment 1 of the present invention;

[0054] Figure 2 It is a module diagram of a guided assisted evaluation information processing system according to Embodiment 2 of the present invention. Detailed Embodiments

[0055] Embodiment 1

[0056] Existing intelligent medical consultation technologies (such as template Q&A trees, BERT target standardization, and large language models) have significant limitations in intelligent medical consultation systems, telemedicine, and chronic disease management: static models cannot dynamically optimize the medical consultation path based on real-time patient feedback (such as ambiguous goals and multimodal data), resulting in missed diagnoses of atypical cases; the interactive interface lacks a dynamic visualization guidance (such as probability graph iteration) and patient interaction mechanism, and incomplete information is collected; the management of historical medical data is rigid and disconnected from real-time medical consultations, making it difficult to support personalized decision-making. These problems restrict the efficiency and accuracy of scenarios such as clinical auxiliary diagnosis and community health management. Therefore, please refer to Figure 1 , an embodiment provided by the present invention: a guided assisted evaluation information processing method, the steps of which include:

[0057] S1. Initialize the user's medical consultation information to obtain initial Q&A information, and based on the initial Q&A information, perform preliminary matching and phrase standardization calibration in combination with a preset standard knowledge base to obtain an initial matching target sequence decision parameter set;

[0058] Furthermore, the matching target in this embodiment is the disease information matched in the pre-diagnosis stage;

[0059] S2. Based on the initial matching target sequence decision parameter set, combine a preset step-by-step guided Q&A assistance strategy and synchronously call the standard knowledge base to obtain a user profile and a step-by-step enhanced decision parameter set corresponding to the matching target sequence, and use the Q&A information after phrase standardization in each step to update the step-by-step enhanced decision parameter set of the matching target sequence in real time;

[0060] The step-by-step guided Q&A assistance strategy constructs the Q&A frequency of the matching target sequence in the next round by matching the initial discrimination probability sequence corresponding to each matching target in the step-by-step enhanced decision parameter set after each round of Q&A update of the matching target sequence;

[0061] S3. Based on the user profile and the step-by-step enhanced decision parameter set corresponding to the matching target sequence, combine a preset decision discrimination model to obtain a confidence sequence corresponding to the matching target sequence;

[0062] Furthermore, the decision discrimination model in this embodiment is constructed by combining a confidence function and a decision tree algorithm.

[0063] S4. Compare the confidence sequence corresponding to the matching target sequence with a preset confidence threshold. If the confidence of at least one matching target is greater than the preset confidence threshold, output the parameter strategy space corresponding to the matching target with the maximum confidence;

[0064] Furthermore, the parameter strategy space in this embodiment includes the pathological parameters corresponding to the matching target (disease) with the maximum confidence and the corresponding diagnosis and treatment methods and rehabilitation suggestions obtained by matching;

[0065] S5. If the confidence levels of all matching targets are less than the preset confidence threshold, determine whether the current round of the step-by-step guided Q&A assistance process is less than the preset round threshold. If it is less, repeat the step-by-step guided Q&A assistance and decision-making discrimination process. If, until the round is greater than or equal to the preset round threshold, all the confidence levels of the corresponding matching target sequence are still less than or equal to the preset confidence threshold, then reply that the Q&A is abnormal.

[0066] Furthermore, to more clearly illustrate the implementation process of the above method, a specific example is given here, which specifically includes the following stages:

[0067] Initialization and target collection:

[0068] User information initialization:

[0069] Initialize to obtain the information of the designated patient of the current user (including the patient's historical medical records, and the reference file information of the hospital / physical examination institution), load the local image recognition and Embedding models on the terminal. High-performance mobile phones or terminals can choose to load the local knowledge base to speed up the diagnosis.

[0070] The user's consultation request and the initially described disease information are preliminarily processed by the knowledge base, and the Agent is called to guide the user to supplement and confirm the key judgmental diseases of the corresponding pathological diseases in descending order of the user's disease matching degree, generating the initial matching disease sequence decision parameter set of the user's currently described disease information feedback.

[0071] According to the content in the initial matching disease sequence decision parameter set, guide the user to supplement basic information in the first step, and determine whether the basic information: height, weight, age, gender, etc. have been completed. If it is completed, generate a diagnosis probability set (from high to low). According to the content in the decision parameter set, focus on asking about the diagnosis pathology with the highest probability, guide the user to gradually supplement disease information in the second step or directly conduct multi-round / combination confirmation of the diseases asked by the AI. If not, return to force supplementation.

[0072] The disease information gradually supplemented in the second step includes: duration, pain location, degree, mental state, questions, stool sample, etc. When it is completed, guide the user to selectively fill in more auxiliary diagnosis information in the third step, including hospital historical diagnosis, hospital test sheets, CT, magnetic resonance imaging, etc. records.

[0073] Furthermore, in the process of multi-round / combination confirmation here, the underlying configured standard knowledge base and the preset extended phrase correction model are synchronously referenced to perform synchronous real-time correction on the user input.

[0074] Knowledge base pre - diagnosis: According to the completed symptom description provided by the user, the knowledge base is called in the form of a parameter set to generate pre - diagnosis information at the knowledge base level;

[0075] Furthermore, the process of using the standardized Q&A information after each step of phrase standardization to update the step - by - step enhanced decision parameter set for the matching target sequence in real - time in this embodiment is as follows:

[0076] Use the standardized Q&A information obtained by standardizing the Q&A phrases by combining the gradually supplemented symptom information in each Q&A process with the standard knowledge base to update and enhance the step - by - step enhanced decision parameter set of the corresponding matching target sequence in the previous step.

[0077] Large - model re - diagnosis: Combine the output of the knowledge base pre - diagnosis and the content of the user's diagnosis decision parameter set for review, and output the current matching disease decision confidence parameter. The backend judges whether the confidence parameter of the preferred pathology decision is greater than the preset confidence threshold. If it is greater, the diagnosis content is output with the preferred pathology, including diagnosis information, disease name, recommended treatment / medication, and at the same time guide the user to choose optional expert consultation, online drug purchase, hospital registration for consultation, etc. Finally, guide the user to score the current diagnosis and then end;

[0078] If it is less, then judge whether the number of inquiry rounds has reached or exceeded the inquiry round threshold. If not, repeat the above - mentioned pre - diagnosis and re - diagnosis stages in a loop. When it is judged that the number of inquiry rounds is greater than or equal to the inquiry round threshold and the confidence corresponding to the confidence parameter of the preferred pathology decision is still less than the preset confidence threshold, an abnormality is output, that is, the diagnosis cannot be made, and it is recommended that the user seek medical diagnosis. At the same time, recommended hospital information is provided. Finally, guide the user to score the current diagnosis and then end.

[0079] Furthermore, the process of obtaining the initial matching target sequence decision parameter set in this embodiment includes:

[0080] Based on the preset multi - modal pre - loading model, combined with the preset historical information library and the user's initial input text information, obtain the initial target triple information sequence; the historical information library in this embodiment is the information library corresponding to the patient's historical diagnosis cases.

[0081] The initial target triple information sequence includes the target name, the matching target text parameter, and the initial correlation matching degree between the target name and the matching target text parameter; the target name is the disease name, and the same applies to subsequent descriptions.

[0082] Based on the initial target triple information sequence, combined with the preset standard knowledge base and the preset extended phrase correction model, obtain the initial matching target sequence decision parameter set in descending order of the initial correlation matching degree.

[0083] Furthermore, the preset historical information library in this embodiment is constructed from the user's historical and real - time diagnosis information through a knowledge graph and a graph database;

[0084] Further, the historical information database in this embodiment is connected to the standard knowledge database through the extended phrase correction model and the index information constructed based on the correlation degree obtained from the query pathological information and the standard pathological information.

[0085] Further, the multi-modal preloading model in this embodiment is constructed by integrating a pre-trained Chinese Bert model, a pre-trained ResNet-18 algorithm, and an LSTM algorithm. Among them, the pre-trained Chinese Bert model is used to process the text data in the preloaded case information database and the user's initial input inquiry information, the pre-trained ResNet-18 algorithm is used to process the image data in the preloaded case information database, and the LSTM algorithm is used to process the time series numerical data in the preloaded case information database;

[0086] Further, the exemplary process of constructing and processing data of the multi-modal preloading model in this embodiment includes:

[0087] Input modality:

[0088] Text data: User symptom description (such as "headache for three days, accompanied by nausea"), case text records.

[0089] Image data: Medical images (CT, X-ray films), scanned copies of laboratory test reports.

[0090] Time series data: Historical medical treatment timeline (such as blood pressure monitoring records, medication cycles).

[0091] Model integration strategy:

[0092] Chinese BERT (text processing):

[0093] Use the pre-trained BERT-Base-Chinese model (12-layer Transformer, hidden layer 768 dimensions).

[0094] After inputting the disease text, extract the 768-dimensional vector marked by CLS as the disease semantic feature.

[0095] Example: User inputs "throbbing pain in the temple" → output feature vector;

[0096] ResNet-18 (image processing):

[0097] Load the ResNet-18 model pre-trained on ImageNet, remove the fully connected layer, and retain the output of the last convolutional layer (512-dimensional feature).

[0098] Image preprocessing: Normalize to a resolution of 224×224 and standardize the RGB channels.

[0099] Example: CT image → output feature vector;

[0100] LSTM (for time series processing):

[0101] A two - layer LSTM network (with a 128 - dimensional hidden layer), and the input is the standardized time series data (such as a blood pressure value sequence).

[0102] Time alignment: Pad zeros to the maximum length for unequal - length sequences and extract the hidden state at the last time step.

[0103] Example: Blood pressure record [120 / 80, 130 / 85,...] → output feature vector;

[0104] Fuse the above - mentioned disease semantic features, image features, and numerical sequence features, and input them into the graph database (Neo4j) of the case information library, perform similarity retrieval, and output: triple <disease name, matching pathology, associated matching degree>.

[0105] Exemplarily, the initial matching process includes:

[0106]

[0107] Furthermore, the initial matching target sequence decision parameter set in this embodiment includes the initial matching target and the corresponding standard pathology text feature set, the correction accuracy of the target keyword group corresponding to each matching target, and the initial associated matching degree sequence corresponding to the initial matching target sequence;

[0108] The extended phrase correction model is obtained by training the matching confidence algorithm in combination with a preset target extended phrase library, and is used to perform standardization correction on the non - standard target keyword groups in the initial target triple information sequence and obtain the correction accuracy of the corresponding target keyword groups;

[0109] The target extended phrase library is composed of a standard target name, non - standard target keyword groups containing at least one key character in the standard target name, and associated aliases. Further, the target extended phrase library in this embodiment is the disease extended phrase library, and other targets in the embodiment also represent the meaning of diseases.

[0110] Furthermore, the preset historical information library in this embodiment is constructed by the user's history and real - time Q&A result information through a knowledge graph and a graph database;

[0111] Furthermore, the historical information library in this embodiment is connected to the standard knowledge library through the index information constructed by the extended phrase correction model and the association degree obtained based on the Q&A information and the standard target parameter information.

[0112] Furthermore, the core implementation principle of the extended phrase correction model in this embodiment includes:

[0113] Model architecture:

[0114] Input: The user's original disease phrase (such as "head swelling", "difficulty breathing").

[0115] Processing flow:

[0116] Keyword splitting: Extract the core disease words ("head", "breathing") through a medical word segmentation tool (such as LTP).

[0117] Expansion library retrieval: Search for non-standard words containing the core word and the corresponding standard words in the disease expansion phrase library.

[0118] Example library entry:

[0119]

[0120] Confidence calculation: Use the weighted calculation of the edit distance (Levenshtein Distance) and word vector similarity (Word2Vec) to calculate the matching score of the candidate standard word. For example: Confidence = 0.6×(1 - edit distance / maximum length) + 0.4×cosine similarity of word vectors;

[0121] Threshold filtering: Retain the candidate words with a confidence > 0.7, and sort them by score to output the final correction result.

[0122] Training data construction:

[0123] Extract the doctor-patient conversation records from the electronic medical records and manually label the mapping relationship between non-standard diseases and standard terms.

[0124] Expand the non-standard word library through data augmentation techniques (such as replacing synonyms, adding noise).

[0125] Real-time processing steps and examples:

[0126] Step 1: Input processing and initial triple generation;

[0127] User input: "Head swelling for three days, occasionally difficulty breathing";

[0128] Multi-modal preloading model processing:

[0129] Text modality: Chinese BERT extracts the disease semantic vector (CLS vector).

[0130] Image modality: If there is no image input, skip this step.

[0131] Time series data: Parse the historical heart rate records (such as [72, 85, 78] bpm), and LSTM generates time series features.

[0132] Output initial triples:

[0133]

[0134]

[0135] Step 2: Expansion phrase correction and parameter set construction;

[0136] Non-standard word correction:

[0137] The input word "head swelling" matches the standard word "headache" in the expansion library, with a confidence of 0.85;

[0138] The input word "difficulty breathing" is corrected to "dyspnea", with a confidence of 0.92.

[0139] Update triples:

[0140]

[0141] Generate decision parameter set:

[0142]

[0143]

[0144] Key instance illustration:

[0145] Instance 1: Multi-level mapping of non-standard words

[0146] User input: "chest pain with a constricting feeling";

[0147] Expansion library matching:

[0148] First-level matching: Core word "chest" → Standard word "thorax" (confidence 0.8);

[0149] Second-level matching: Core word "constricting pain" → Standard word "colic" (confidence 0.9);

[0150] Combined correction: "thoracic colic" (comprehensive confidence 0.85);

[0151] Matched pathology: Coronary heart disease (similarity 0.82), Gastroesophageal reflux (similarity 0.45).

[0152] Instance 2: Correction accuracy affects decision-making

[0153] Scenario: User input "dizziness and vertigo" is corrected to "vertigo" (accuracy 0.88);

[0154] Matching without correction:

[0155] Anemia (similarity 0.70), Benign paroxysmal positional vertigo (similarity 0.65);

[0156] After correction and matching:

[0157] Benign paroxysmal positional vertigo (similarity 0.65 × 0.88 = 0.572);

[0158] Meniere's disease (newly added match, similarity 0.75 × 0.88 = 0.66);

[0159] Result: Meniere's disease becomes the primary candidate target.

[0160] Furthermore, the process of obtaining the user portrait and the corresponding step-by-step enhanced decision parameter set in this embodiment includes:

[0161] Based on the initial matching target sequence decision parameter set, combined with the preset target parameter-portrait association mapping space and the first supplementary parameter information, obtain the user portrait and the initial enhanced decision parameter space;

[0162] Furthermore, the target parameter in this embodiment is the standard pathological parameter corresponding to each disease, which is obtained from the professional medical information database and sorted out by professionals in this field to obtain the standard knowledge base (i.e., the standard medical knowledge base);

[0163] The first supplementary parameter information includes the basic information and location information of the user corresponding to the matching target; among them, the basic information includes height, weight, age, gender and location information;

[0164] The user portrait is the same as the parameters included in the first supplementary parameter information;

[0165] The target parameter-portrait association mapping space is obtained by pre-training through an association matching algorithm by combining the user portrait with the standard knowledge base and the historical information base;

[0166] The initial enhanced decision parameter space includes the user portrait, the initial matching target sequence arranged in descending order, and the corresponding initial discrimination probability sequence and the matching target text parameters;

[0167] Based on the numerical ratio of the initial discrimination probabilities corresponding to all matching targets in the initial enhanced decision parameter space, obtain the second supplementary information query frequency corresponding to each matching target in the initial enhanced decision parameter space.

[0168] Based on the initial enhanced decision parameter space and the standard knowledge base, through the preset text random generation model, combined with the second supplementary information query frequency corresponding to each matching target, generate the second loop Q&A text set, and synchronously call the standard knowledge base and the preset extended phrase correction model to correct the generated second loop Q&A text set;

[0169] Update the initial enhanced decision parameter space based on the corrected second-round Q&A text set, and obtain the second initial discrimination probability corresponding to each matching target after the update.

[0170] Based on the second initial discrimination probability corresponding to each matching target after the update and in combination with the initial association mapping threshold, screen and filter the initial matching target sequence to obtain the second matching target sequence after filtering and update and the corresponding enhanced decision parameter space;

[0171] Repeat the process of obtaining the above-mentioned second matching target sequence and the corresponding enhanced decision parameter space until all the initial discrimination probabilities in the updated matching target sequence are greater than the initial association mapping threshold and the matching target corresponding to the maximum initial discrimination probability remains unchanged, and obtain the target sequence after step-by-step enhancement is completed and the corresponding enhanced decision parameter space.

[0172] Furthermore, in this embodiment, in order to better illustrate the process of obtaining the above-mentioned step-by-step enhanced decision parameter set, the following specific example implementation process is given, which specifically includes:

[0173] Example background:

[0174] A 35-year-old female patient, 160 cm tall and weighing 72 kg (BMI is 28, belonging to the obese category), lives in Shanghai and complains of "pulsating headache accompanied by nausea". The candidate pathological targets initially matched by the system are migraine, hypertensive encephalopathy, and increased intracranial pressure, and the initial probabilities are 55%, 2%, and 20% respectively.

[0175] Step 1: User portrait and initial parameter space construction;

[0176] The system dynamically corrects in combination with the patient's basic information (age, gender, BMI, geographical location) and the preset target parameter-portrait association mapping space. This mapping space is trained based on the historical case data in Shanghai and reflects the association rules between different population characteristics and diseases.

[0177] Obesity association correction: The association degree between obesity (BMI = 28) and migraine increases by 15%, and the association degree with hypertensive encephalopathy increases by 30%.

[0178] Age correction: 35 years old belongs to the high-incidence age group of migraine, and the probability increases by 10%; while the probability of hypertensive encephalopathy decreases by 5% in young patients.

[0179] Gender correction: The risk coefficient of migraine in women increases by an additional 20%.

[0180] After multi-dimensional correction, the probability of migraine increased from 55% to 79.8%, hypertensive encephalopathy was fine-tuned to 31.3%, and the increased intracranial pressure remained unchanged at 20%. The system generates an initial enhanced parameter space containing the corrected probability, user profile, and target description, providing a benchmark for subsequent Q&A guidance.

[0181] Exemplarily, the initial enhanced parameter space includes:

[0182]

[0183] Step 2: Dynamic Q&A generation and semantic correction;

[0184] Based on the corrected probability ratio, the system automatically allocates the questioning frequency for each pathology in each round of Q&A. Migraine, with the highest probability (79.8%), occupies the main part of the 6 questions in this round, and hypertensive encephalopathy and increased intracranial pressure are each allocated 2 questions.

[0185] Migraine-related questions (6):

[0186] Focus on typical features, such as "Do visual auras such as flashes or dark spots appear before the headache?" "Is the pain concentrated on one side and accompanied by photophobia?";

[0187] Hypertensive encephalopathy questions (2):

[0188] Pay attention to acute conditions, such as "Has the blood pressure suddenly risen above 180 / 120 mmHg recently?";

[0189] Increased intracranial pressure questions (2):

[0190] Ask about specific manifestations, such as "Is there blurred vision or projectile vomiting?";

[0191] The above questioning frequencies are just hypothetical values, and in actual applications, key questions are asked for those with a large proportion according to the frequency ratio.

[0192] After the user answers, the system converts colloquial expressions into standard medical terms through an extended phrase correction model. For example, correcting the patient's description of "flashes in front of the eyes" to "visual aura" and standardizing "a bit high blood pressure" to "Grade 2 hypertension (160 / 100 mmHg)".

[0193] Step 3: Parameter update and target screening;

[0194] The system adjusts the confidence probability of each pathology according to the user's answer content:

[0195] Supportive answer (such as confirming visual aura): The probability of migraine increases by 18% and reaches 97.8%.

[0196] Negative answer (such as denying a sudden increase in blood pressure): The probability of hypertensive encephalopathy decreases by 15% to 16.3%; the probability of increased intracranial pressure decreases to 5% due to denying blurred vision.

[0197] According to the preset probability threshold (0.5), the system screens out hypertensive encephalopathy and increased intracranial pressure, and only retains migraine as the only candidate target. The updated parameter space focuses on the in-depth verification of migraine, including new disease parameters (such as visual aura confirmation, blood pressure recording) and adjusted high confidence (97.8%).

[0198] Step 4: Iteration termination and result output;

[0199] Since only migraine remains in the current matching target sequence and its probability far exceeds the threshold, and the probability of this target remains unchanged as the maximum value in two consecutive iterations, the system determines that the convergence condition is reached and terminates the Q&A process. Finally, an enhanced decision parameter set is output.

[0200] Furthermore, the target parameter-portrait association mapping space in this embodiment includes three-level mapping; the three-level mapping includes a BMI index-target mapping layer, an age-target mapping layer, and a gender-target mapping layer;

[0201] Furthermore, the construction process of the BMI index-target mapping layer in this embodiment includes:

[0202] In the state where the BMI index is determined, through the association analysis algorithm, combined with the frequency of occurrence (disease) of each level of BMI index and the corresponding different targets in different regions per unit time and the target parameter information, the association degree between each level of BMI index and different targets in each region is obtained;

[0203] Based on each level of BMI index, target name in each region and the association degree between each level of BMI index and different targets, the BMI index-target mapping layer is constructed.

[0204] Furthermore, to more clearly illustrate the three-level mapping of the target parameter-portrait association mapping space, this embodiment gives some corresponding examples, specifically including:

[0205] BMI index-target mapping layer construction example:

[0206] Scenario: Construct the BMI index-target mapping layer for the East China region;

[0207] Data source: 100,000 electronic medical records from the top three hospitals in the East China region in the past 5 years, counting the disease distribution of patients with different BMI levels.

[0208] BMI classification and associated diseases:

[0209] BMI ≥ 28 (obesity):

[0210] Hypertension: the prevalence frequency accounts for 32% (correlation degree 0.82);

[0211] Type 2 diabetes: the prevalence frequency accounts for 18% (correlation degree 0.75);

[0212] Osteoarthritis: the prevalence frequency accounts for 12% (correlation degree 0.68);

[0213] BMI 24 - 27.9 (overweight):

[0214] Fatty liver: the prevalence frequency accounts for 25% (correlation degree 0.70);

[0215] Sleep apnea: the prevalence frequency accounts for 10% (correlation degree 0.60);

[0216] BMI 18.5 - 23.9 (normal):

[0217] Functional gastrointestinal disease: the prevalence frequency accounts for 15% (correlation degree 0.55);

[0218] BMI < 18.5 (underweight):

[0219] Anemia: the prevalence frequency accounts for 20% (correlation degree 0.65);

[0220] The specific mapping result is:

[0221]

[0222]

[0223] Furthermore, the construction process of the age - target mapping layer includes:

[0224] Based on the correlation degree between each level of BMI index and different targets in each region as the weighted weight, combined with the age - level information in each region, the correlation degree between each age level and different targets within each region is obtained through correlation analysis;

[0225] Furthermore, in this embodiment, the age level is one age level every 5 years;

[0226] The age - target mapping layer is constructed based on the age level, target name in each region, and the correlation degree between different targets corresponding to each age level within each region;

[0227] Example of constructing the age - target mapping layer:

[0228] Scenario: Based on the BMI - target layer in the East China region, construct the age - target mapping layer for the age group of 35 - 39 years old;

[0229] Data weighting rule: Using the BMI layer correlation as the weight and combining the age distribution to correct the disease risk.

[0230] Calculation example:

[0231] Correction of the correlation of hypertension in the age group of 35 - 39 years old:

[0232] For the group with BMI ≥ 28: Correlation 0.82 × obesity proportion in this age group 30% = 0.246;

[0233] For the group with BMI 24 - 27.9: Correlation 0.70 × overweight proportion 40% = 0.280;

[0234] Comprehensive correlation: 0.246 + 0.280 = 0.526 (after standardization is 0.75);

[0235] Mapping layer structure:

[0236]

[0237] Furthermore, the construction process of the gender - target mapping layer in this embodiment includes:

[0238] Based on the correlation between each age level and different targets within each region as the weighted weight and combining the gender ratio information at each age level within each region, obtain the correlation between each gender and different targets within each region, and obtain the gender - target layer;

[0239] Furthermore, the construction example of the gender - target mapping layer in this embodiment includes:

[0240] Scenario: Based on the age - target layer of 35 - 39 years old in East China, construct a gender - specific mapping;

[0241] Data source: The proportion of male and female patients in this age group (45% male, 55% female) and the gender distribution of diseases.

[0242] Calculation example:

[0243] Correction of the gender correlation of hypertension:

[0244] Proportion of male patients 60% → Correlation enhancement: 0.75 × (1 + 0.6) = 1.20;

[0245] Proportion of female patients 40% → Correlation attenuation: 0.75 × (1 - 0.2) = 0.60;

[0246] Correction of the gender correlation of anxiety disorder:

[0247] Proportion of female patients 70% → Correlation enhancement: 0.58 × (1 + 0.7) = 0.986;

[0248] The mapping layer structure includes:

[0249]

[0250]

[0251] Based on the BMI index - target mapping layer, age - target mapping layer, and gender - target mapping layer, a target parameter - portrait association mapping space is obtained through a cascading method.

[0252] Furthermore, dynamic diagnostic decision examples for the above - mentioned target parameter - portrait association mapping space include:

[0253] System behavior:

[0254] Initial matching: The user's main complaint of "headache" is initially matched with migraine (probability 0.65) and anxiety disorder (probability 0.50).

[0255] Portrait correction:

[0256] BMI = 29 → probability of hypertension + 0.82×regional weight 0.3 → increased to 0.65 + 0.25 = 0.90;

[0257] Female + 38 years old → probability of anxiety disorder × 1.7 → 0.50×1.7 = 0.85;

[0258] Question - and - answer guidance:

[0259] For anxiety disorder: Ask "Do you have insomnia or palpitations?" (frequency proportion 40%)

[0260] For migraine: Ask "Is the headache unilateral? Do you have visual aura?" (frequency proportion 50%)

[0261] User's answer: Confirmed insomnia, denied unilateral headache → confidence level of anxiety disorder increased to 0.95, migraine decreased to 0.60.

[0262] Output result: Anxiety disorder is the preferred diagnosis, and psychological assessment and anti - anxiety treatment are recommended.

[0263] In summary, in this embodiment, a multi-modal preloading model integrating BERT, ResNet-18, and LSTM is first used to achieve the mapping of heterogeneous feature spaces for text-based disease descriptions, medical image features, and physiological time-series data. In the text modality, the BERT model captures the contextual semantic associations of disease descriptions through a deep attention mechanism (such as the temporal feature matching between "throbbing pain in the temple" and "pulsating headache"). In the image modality, the local pathological features of medical images are extracted through the convolutional kernels of ResNet-18 (such as the spatial distribution pattern of abnormal temporal lobe morphology in CT images). In the time-series modality, the dynamic evolution laws of physiological parameters such as blood pressure and heart rate are analyzed using the recurrent network structure of LSTM. The three are concatenated at the feature level to form a high-dimensional joint representation vector, breaking through the limitations of traditional single-modal analysis, enabling the disease matching process to cover semantic, spatial, and temporal dimension evidence chains simultaneously. Furthermore, by expanding the hybrid similarity calculation (edit distance + cosine similarity of word vectors) of the phrase correction model and the hierarchical mapping mechanism of the disease expansion thesaurus, the patient's colloquial expressions (such as "head swelling") are corrected to standard medical terms (such as "headache"), eliminating the semantic gap caused by differences in expression and achieving precise alignment between unstructured input and the standardized knowledge base.

[0264] Secondly, based on the three-level cascaded correction mechanism (BMI-target layer, age-target layer, gender-target layer) of the target parameter-portrait association mapping space, the system dynamically adjusts the probability weights of the initial matching targets. For example, in the BMI-target layer, through regional statistical modeling (such as increasing the weight coefficient of the migraine incidence rate in obese populations in Shanghai), combined with the risk attenuation factor at the age level (such as reducing the baseline probability of meningitis in 35-year-old patients) and the gender-specific weighting term (such as the migraine risk multiplier for women), a dynamic quantification model of the association between multi-dimensional user portraits and diseases is formed. This mechanism enables the discrimination probability of the initial matching target sequence to reflect individual risk characteristics rather than relying on a static probability distribution. On this basis, the step-by-step guidance strategy uses a Markov decision process to generate a dynamic question frequency distribution: adaptively allocate questioning resources according to the probability ratios of each target in the current round (such as 60.9% of migraine accounting for 79.8% of the total probability of 1.311), and preferentially target the key discriminant diseases of high-probability pathologies (such as the visual aura of migraine) for targeted questioning. At the same time, the text random generation model combines the constraints of the standard knowledge base to ensure the medical standardization and logical coherence of the generated questions (such as "Is the pain unilateral?" conforms to the ICD-11 diagnostic standard entries), avoiding interference from irrelevant information caused by open-ended questions. This probability-driven question generation strategy significantly shortens the acquisition path of effective evidence and reduces the number of redundant interactions.

[0265] Third, a dynamic decision-making closed loop is constructed through confidence threshold comparison and multiple rounds of reinforcement loops. After each round of question-answering, the decision-discrimination model uses an attention-weighted multi-source evidence fusion algorithm to calculate an updated confidence score based on text matching, imaging evidence strength, and temporal pattern consistency (e.g., the progressive exacerbation of migraine). For supportive answers (e.g., confirmation of visual aura), a linear gain strategy is used to increase the confidence of the corresponding pathology; for negative answers (e.g., denial of sudden blood pressure rise), a probability decay penalty is applied. After the confidence sequence is filtered by a threshold, the system automatically removes low-probability targets (e.g., targets with a probability of increased intracranial pressure falling to 5%), narrowing the range of matching targets for the next round and forming a progressive decision-focusing mechanism. Furthermore, the design of the iteration termination condition (the maximum probability target is stable and all targets exceed the threshold) prevents the risk of misjudgment caused by premature convergence and avoids the waste of resources caused by infinite loops. This mechanism, based on evidence accumulation and dynamic contraction of the parameter space, enables the system to maintain stable decision direction even when faced with ambiguous symptoms or multiple competing pathologies.

[0266] Fourth, the historical information repository and the standard knowledge base are bidirectionally indexed via a knowledge graph and a graph database (Neo4j). The symptom-pathology mapping relationship constructed by the extended phrase correction model serves as a semantic bridge, supporting retrospective queries from non-standard terms to ICD codes (e.g., "heart pain" is mapped to "chest pain" and associated with the ICD-10 code I20 for coronary heart disease). During the diagnostic process, the system writes the corrected symptom parameters, user profiles, and question-and-answer records to the graph database nodes in real time, establishing a symptom-pathology-profile association network using an attribute graph model. This structured data storage not only provides a traceable chain of evidence for the generation of decision parameter sets (e.g., demonstrating the derivational path from "visual aura → migraine") but also optimizes the statistical model of the historical case database by continuously accumulating real-time consultation data, fostering data-driven self-evolution. For example, when new epidemiological characteristics are added to a region, the BMI indicator-target mapping layer can automatically update the correlation coefficient, ensuring the timeliness of the profile correction factor.

[0267] Fifth, a preset query round threshold and confidence anomaly determination mechanism provide dual safeguards: if the confidence threshold cannot be reached after multiple iterations, the system proactively terminates the process and recommends referral, avoiding the risk of delays due to algorithmic limitations. Furthermore, the design of localized image recognition and embedding models on high-performance terminals balances response speed and energy consumption through dynamic allocation of computing resources (for example, loading the local knowledge base only on high-end devices), ensuring smooth mobile application operation. This layered processing strategy maintains the accuracy of the core algorithm while expanding the system's applicability across diverse hardware environments.

[0268] Example 2

[0269] See also Figure 2, Another embodiment provided by the present invention: A guided assisted evaluation information processing system, comprising: an initial matching module, a step-by-step question and answer module, and a discrimination module;

[0270] The discrimination module includes a confidence unit, a secondary discrimination unit, and an inquiry adjustment unit;

[0271] The initial matching module is used to obtain initial question and answer information, and based on the initial question and answer information, perform preliminary matching in combination with a preset standard knowledge base to obtain an initial matching target sequence decision parameter set;

[0272] The step-by-step question and answer module is used to obtain a user profile and a step-by-step enhanced decision parameter set corresponding to the matching target sequence based on the initial matching target sequence decision parameter set in combination with a preset step-by-step guided question and answer assistance strategy and synchronously call the standard knowledge base;

[0273] The discrimination module is used to perform a follow-up diagnosis discrimination and adjust the diagnosis result based on the user profile and the corresponding step-by-step enhanced decision parameter set in combination with a preset decision discrimination model;

[0274] The confidence unit, based on the user profile and the step-by-step enhanced decision parameter set corresponding to the matching target sequence, in combination with a preset decision discrimination model, obtains a confidence sequence corresponding to the matching target sequence;

[0275] The inquiry adjustment unit is used to count the number of pre-diagnosis inquiry rounds, and in the state where there is no result in the follow-up diagnosis, adjust the follow-up diagnosis output result according to the counted number of pre-diagnosis inquiry rounds and the inquiry round threshold.

[0276] The secondary discrimination unit is used to compare the confidence sequence corresponding to the matching target sequence with a preset confidence threshold. If the confidence of at least one matching target is greater than the preset confidence threshold, the parameter strategy space corresponding to the matching target with the maximum confidence is output;

[0277] If the confidence of all matching targets is less than the preset confidence threshold, it is determined by the inquiry adjustment unit whether the number of rounds of the current step-by-step guided question and answer assistance process is less than the preset inquiry round threshold. If it is less, the step-by-step guided question and answer assistance and decision discrimination processes are repeated. If until the number of inquiry rounds is greater than or equal to the preset inquiry round threshold, and the confidence of all corresponding to the matching target sequence is still all less than or equal to the preset confidence threshold, an abnormal question and answer is replied.

[0278] Embodiment 3

[0279] Another embodiment provided by the present invention: A guided assisted evaluation information processing platform, which is used to implement a guided assisted evaluation information processing method, including: a login interface, an output interface, an initial matching module, a step-by-step question and answer module, and a discrimination module; the discrimination module includes a confidence unit, a secondary discrimination unit, and an inquiry adjustment unit;

[0280] First, enter the login interface and detect whether the user is logged in. If not logged in, perform the login process. If already logged in, load the historical information database, standard knowledge database, and extended phrase correction model through the initial matching module, and generate a greeting message;

[0281] Second, during the above login process, confirm whether the information of the attending person is bound. If it is confirmed that the binding is successful, determine the bound attending person. If not bound, perform the binding process. When the binding is completed, generate an attending person information card, and based on the attending person information card, confirm the selection of the attending person;

[0282] Third, based on the confirmed selection of the attending person, through the step-by-step question and answer module, combined with the loaded historical information database, standard knowledge database, and extended phrase correction model, conduct multi-round guided pre-diagnosis inquiries to obtain a user profile and the corresponding step-by-step enhanced decision parameter set;

[0283] Further, the specific exemplary process of the multi-round guided pre-diagnosis inquiry in this embodiment includes:

[0284] The attending person starts to ask a question, such as "I have a cough". Based on the question asked, combined with the historical information database, standard knowledge database, and extended phrase correction model, confirm the initial matching target sequence decision parameter set;

[0285] The initial matching target sequence decision parameter set includes the corresponding height, weight, age, mental state, body temperature, cough time and duration, whether there is phlegm, and pain in other parts for a coughing attending person;

[0286] Based on the confirmed initial matching target sequence decision parameter set, complete the parameters through the step-by-step question and answer module, and conduct a missing test on each completed parameter. After M rounds of completion and passing the missing test, obtain a user profile and the corresponding step-by-step enhanced decision parameter set;

[0287] Fourth, based on the user profile and the corresponding step-by-step enhanced decision parameter set, obtain the matching target confidence through the confidence unit;

[0288] Fifth, based on the confidence level of the matching target, the secondary discrimination unit determines whether there is a matching target with a confidence level greater than the preset confidence level threshold. If so, the diagnostic parameters corresponding to the target with the maximum confidence level and the corresponding diagnostic strategy are output on the output interface, and a diagnostic score is given to the corresponding diagnostic process. If not, the inquiry adjustment unit determines whether the current step-by-step guided pre-diagnosis inquiry round is less than the preset inquiry round threshold. If it is less than, the above-mentioned distributed guided pre-diagnosis and decision-making discrimination process is repeated. If until the inquiry round is greater than or equal to the preset inquiry threshold, and the confidence levels of all corresponding matching targets are still less than or equal to the preset confidence level threshold, it is replied that the diagnosis cannot be made and the optimal medical treatment information is recommended, and at the same time, a diagnostic score is given to the corresponding diagnostic process.

[0289] Further, in this embodiment, in order to more clearly illustrate the multi-round inquiry process of the above-mentioned guided assisted evaluation information processing platform, the following specific example is given:

[0290] Agent Medical Assistant: Hello! To better assist you, first, please tell me some of your basic information, such as: gender, age, height, weight, etc.

[0291] Patient: I am male, 45 years old, 175 cm tall, and 80 kg in weight.

[0292] Agent Medical Assistant: Thank you for your information! Next, could you tell me what the main symptoms you are currently experiencing are? Such as: fever, cough, headache, etc.

[0293] Patient: I have been having headaches recently, and it seems to be getting worse, especially at night.

[0294] Agent Medical Assistant: I see. Please further describe your headache symptoms: How long has it lasted? Are there any other accompanying symptoms, such as nausea, vomiting, or dizziness?

[0295] Patient: It has lasted for about three days. The headache is intermittent, and I occasionally feel a little nauseous, but no vomiting, and my eyes are also a bit uncomfortable.

[0296] Agent Medical Assistant: Okay, thank you for your detailed description. Next, please tell me if there is any auxiliary diagnostic information, such as: Do you have a past medical history, or have you had any examinations done? Such as blood pressure, blood sugar, etc.

[0297] Patient: I have a history of high blood pressure and have been taking medicine to control it for almost five years. I also had a blood pressure check recently, and the doctor said everything is normal.

[0298] Agent: Thank you for sharing. Based on your symptom description and medical history, I will continue to guide you to provide more symptom information to help us better judge your condition. Please tell me whether your symptoms include fever, chills, fatigue, chest tightness, etc.

[0299] Patient: No fever, no chills, no fatigue, and no chest discomfort.

[0300] Medical Agent: Okay, thank you for your patience. Next, I'll need more detailed diagnostic information. Please tell me when and how often your symptoms occur: For example, are they most severe in the morning, at noon, or at night?

[0301] Patient: The symptoms are mostly more severe at night and milder when getting up in the morning.

[0302] Medical Agent: OK, I understand. Next, I'll need you to provide some additional medical information, such as whether you've had any key tests like temperature and blood pressure in your medical history.

[0303] Patient: As I said, I have a history of hypertension and my recent blood pressure tests were all within normal range.

[0304] Medical Agent: Thank you. Based on the information we've collected so far, I'll provide a preliminary diagnosis. Considering your symptoms and medical history, it seems likely that your headache is related to hypertension. We recommend that you seek further testing, particularly a CT or MRI scan of the brain, to rule out other possible causes.

[0305] Medical Assistant Agent: If you need, I can help you make an appointment with the nearest specialist or recommend a medical institution. Do you need it?

[0306] Patient: Yes, can you make an appointment for me with a neurologist?

[0307] Medical Assistant Agent: Okay, I will make an appointment for you to see a neurologist. Please wait a moment.

[0308] Medical Assistant Agent: The appointment has been completed. Please confirm your information. Your appointment is tomorrow at 2:00 PM at the Neurology Department of XX Hospital. Is there anything else I can help you with?

[0309] Patient: No more, thank you!

[0310] Medical Assistant Agent: You're welcome. I wish you a speedy recovery! If you need any help, please feel free to contact me.

[0311] Example 4

[0312] An electronic device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, a guided assisted evaluation information processing method is implemented.

[0313] A computer-readable storage medium stores computer instructions, and when the computer instructions run, a guided assisted evaluation information processing method is executed.

[0314] The embodiments of the present invention have been described above in conjunction with the accompanying drawings. However, the present invention is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Under the inspiration of the present invention, those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments without departing from the spirit of the present invention and the scope protected by the claims. All of these fall within the protection scope of the present invention.

[0315] If the technical solution of the present disclosure involves personal information, before the product applying the technical solution of the present disclosure processes personal information, the rules for processing personal information have been clearly informed, and the individual's independent consent has been obtained. If the technical solution of the present disclosure involves sensitive personal information, before the product applying the technical solution of the present disclosure processes sensitive personal information, the individual's separate consent has been obtained, and at the same time, the requirement of "express consent" is met. For example, at a personal information collection device such as a camera, a clear and prominent sign is set to inform that the personal information collection range has been entered and personal information will be collected. If an individual voluntarily enters the collection range, it is considered consent to the collection of their personal information; or on the device for processing personal information, when the rules for processing personal information are informed by obvious signs / information, personal authorization is obtained through pop-up information or by asking the individual to upload their personal information themselves; among them, the rules for processing personal information may include information such as the personal information processor, the purpose of processing personal information, the processing method, and the types of personal information processed.

Claims

1. A guided assisted evaluation information processing method, characterized in that Including: Based on the initial Q&A information, perform preliminary matching by combining with a preset standard knowledge base to obtain an initial matching target sequence decision parameter set; Based on the initial matching target sequence decision parameter set, combine with a preset step-by-step guided Q&A assistance strategy and synchronously call the standard knowledge base to obtain a user profile and a step-by-step enhanced decision parameter set corresponding to the matching target sequence; The step-by-step guided Q&A assistance strategy constructs the Q&A frequency of the matching target sequence in the next round by step-by-step enhancing the initial discrimination probability sequence corresponding to each matching target in the step-by-step enhanced decision parameter set after each round of Q&A update of the matching target sequence; Based on the user profile and the step-by-step enhanced decision parameter set corresponding to the matching target sequence, combine with a preset decision discrimination model to obtain a confidence sequence corresponding to the matching target sequence; Compare the confidence sequence corresponding to the matching target sequence with a preset confidence threshold. If the confidence of at least one matching target is greater than the preset confidence threshold, output the parameter strategy space corresponding to the matching target with the highest confidence; If the confidence of all matching targets is less than the preset confidence threshold, determine whether the current round of the step-by-step guided Q&A assistance process is less than the preset Q&A round threshold. If it is less, repeat the step-by-step guided Q&A assistance and decision discrimination process. If until the Q&A round is greater than or equal to the preset Q&A round threshold, and the confidence of all matching targets corresponding to the matching target sequence is still all less than or equal to the preset confidence threshold, then reply that the Q&A is abnormal.

2. The guided assisted evaluation information processing method according to claim 1, wherein The obtaining process of the initial matching target sequence decision parameter set includes: Based on a preset multi-modal preloading model, combine with a preset historical information library and user initial input text information to obtain an initial target triple information sequence; The initial target triple information sequence includes a target name, matching target text parameters, and an initial association matching degree between the target name and the matching target text parameters; Based on the initial target triple information sequence, combine with the preset standard knowledge base and a preset extended phrase correction model, and obtain the initial matching target sequence decision parameter set in descending order according to the size of the initial association matching degree.

3. The guided assisted evaluation information processing method according to claim 2, wherein The initial matching target sequence decision parameter set includes an initial matching target and a corresponding standard target parameter text feature set, the correction accuracy of the target keyword group corresponding to each matching target, and an initial association matching degree sequence corresponding to the initial matching target sequence; The extended phrase correction model is obtained by training with a matching confidence algorithm combined with a preset target extended phrase library, and is used to perform standardization correction on non-standard target keyword groups in the initial target triple information sequence and obtain the correction accuracy of the corresponding target keyword group; The target extended phrase library is composed of a standard target name, non-standard target keyword groups containing at least one key character in the standard target name, and associated aliases.

4. The guided assisted evaluation information processing method according to claim 3, wherein The obtaining process of the user profile and the corresponding step-by-step enhanced decision parameter set includes: Based on the initial matching target sequence decision parameter set, combine with a preset target parameter-profile association mapping space and first supplementary parameter information to obtain a user profile and an initial enhanced decision parameter space; The first supplementary parameter information includes the basic information and location information of the user corresponding to the matching target; The user profile is the same as the parameters included in the first supplementary parameter information; The target parameter-profile association mapping space is pre-trained and obtained through an association matching algorithm by combining the user profile with the standard knowledge base and the historical information base; The initial enhanced decision parameter space includes the user profile, an initial matching target sequence arranged in descending order, and corresponding initial discrimination probability sequences and matching target text parameters; Based on the numerical ratio of the initial discrimination probabilities corresponding to all matching targets in the initial enhanced decision parameter space, the second supplementary information query frequency corresponding to each matching target in the initial enhanced decision parameter space is obtained.

5. The guided assisted evaluation information processing method according to claim 4, wherein The process of obtaining the user profile and the corresponding step-by-step enhanced decision parameter set further includes: Based on the initial enhanced decision parameter space and the standard knowledge base, a second loop Q&A text set is generated through a preset text random generation model in combination with the second supplementary information query frequency corresponding to each matching target, and the standard knowledge base and a preset extended phrase correction model are synchronously called to correct the generated second loop Q&A text set; The initial enhanced decision parameter space is updated based on the corrected second loop Q&A text set, and the second initial discrimination probability corresponding to each matching target after the update is obtained.

6. The guided assisted evaluation information processing method according to claim 5, characterized in that, The process of obtaining the user profile and the corresponding step-by-step enhanced decision parameter set further includes: Based on the second initial discrimination probability corresponding to each matching target after the update in combination with the initial association mapping threshold, the initial matching target sequence is screened and filtered to obtain a filtered and updated second matching target sequence and the corresponding enhanced decision parameter space; Repeat the process of obtaining the second matching target sequence and the corresponding enhanced decision parameter space until all the initial discrimination probabilities in the updated matching target sequence are greater than the initial association mapping threshold and the matching target corresponding to the maximum initial discrimination probability remains unchanged, and obtain the target sequence with step-by-step enhancement completed and the corresponding enhanced decision parameter space.

7. The guided-assisted evaluation information processing method according to claim 6, characterized in that The target parameter-profile association mapping space includes three-level mapping; the three-level mapping includes a BMI index-target mapping layer, an age-target mapping layer, and a gender-target mapping layer; The construction process of the BMI index-target mapping layer includes: Under the condition that the BMI index is determined, through an association analysis algorithm in combination with the occurrence frequency and target parameter information of each level of BMI index in different regions within a corresponding different target unit time, the association degree between each level of BMI index in each region and different targets is obtained; The BMI index-target mapping layer is constructed based on each level of BMI index, target name, and the association degree between each level of BMI index and different targets in each region.

8. The guided-assisted evaluation information processing method according to claim 7, characterized in that The construction process of the age-target mapping layer includes: Based on the association degree between each level of BMI index in each region and different targets as a weighted weight in combination with the age grade information in each region, the association degree between each age grade in each region and different targets is obtained through association analysis; The age grade is one age grade every 5 years; The age - target mapping layer is constructed based on the age level, target name in each region, and the correlation degree between different targets corresponding to each age level in each region. The construction process of the gender - target mapping layer includes: Based on the correlation degree between each age level and different targets in each region as the weighted weight, combined with the gender ratio information of each age level in each region, the correlation degree between each gender and different targets in each region is obtained, and the gender - target layer is obtained. Based on the BMI index - target mapping layer, age - target mapping layer, and gender - target mapping layer, the target parameter - portrait association mapping space is obtained through a cascading method.

9. The guided assisted evaluation information processing method according to claim 8, wherein The preset historical information database is constructed from user historical and real - time Q&A result information through a knowledge graph and a graph database. The historical information database and the standard knowledge database are connected through the extended phrase correction model and the index information constructed based on the correlation degree obtained from Q&A information and standard target parameter information.

10. A guided assisted evaluation information processing system for implementing a guided assisted evaluation information processing method according to any one of claims 1-9, characterized in that It includes: An initial matching module, a step - by - step Q&A module, and a discrimination module; the discrimination module includes a confidence unit, a secondary discrimination unit, and an inquiry adjustment unit. The initial matching module is used to obtain initial Q&A information, and based on the initial Q&A information, combined with the preset standard knowledge database, perform a preliminary match to obtain an initial matching target sequence decision parameter set. The step - by - step Q&A module is used to, based on the initial matching target sequence decision parameter set, combined with the preset step - by - step guided Q&A assistance strategy and synchronously call the standard knowledge database, obtain the user portrait and the step - by - step enhanced decision parameter set of the corresponding matching target sequence. The confidence unit, based on the user portrait and the step - by - step enhanced decision parameter set of the corresponding matching target sequence, combined with the preset decision discrimination model, obtains the confidence sequence corresponding to the matching target sequence. The secondary discrimination unit is used to compare the confidence sequence corresponding to the matching target sequence with the preset confidence threshold. If the confidence of at least one matching target is greater than the preset confidence threshold, the parameter strategy space corresponding to the matching target with the maximum confidence is output. If the confidence of all matching targets is less than the preset confidence threshold, the inquiry adjustment unit determines whether the current round of the step - by - step guided Q&A assistance process is less than the preset inquiry round threshold. If it is less, the step - by - step guided Q&A assistance and decision discrimination process are repeated. If until the inquiry round is greater than or equal to the preset inquiry round threshold, and the confidence of all corresponding matching target sequences is still all less than or equal to the preset confidence threshold, an abnormal Q&A reply is given.

Images