Methods and systems for providing digital and intelligent medical services

Abstract

This invention belongs to the field of smart healthcare technology and provides a method and system for providing digital and intelligent medical services. The method includes: receiving basic medical information input by the patient; matching and analyzing the basic medical information with a disease description guidance knowledge base to generate a personalized symptom guidance framework; performing semantic analysis and completeness verification on the symptom description information fed back by the patient through the personalized symptom guidance framework to identify missing core symptom dimensions; generating supplementary guidance instructions for the missing core symptom dimensions to obtain complete symptom description data; standardizing the complete symptom description data based on a symptom description standard terminology library to generate a structured disease summary; and pushing the structured disease summary to the doctor's end. This invention can effectively solve the problems of unprofessional patient descriptions and information omissions, shorten consultation time, reduce information transmission errors, help doctors quickly grasp key symptoms, improve doctor-patient communication efficiency and diagnostic accuracy, and optimize the medical experience.

Description

Technical Field

[0001] This invention relates to the field of smart healthcare technology, and more specifically, to a method and system for providing digitalized medical services. Background Technology

[0002] In the current medical process, patients need to verbally describe their condition to the doctor during the initial or follow-up visit, a process that is generally inefficient and prone to information distortion.

[0003] On the one hand, most patients lack professional medical knowledge and find it difficult to accurately describe the core characteristics of their symptoms. They often get bogged down in redundant descriptions of non-critical details, resulting in a low percentage of effective information transmission and prolonging the consultation time. For example, patients may only vaguely state that they "feel unwell" or "are in unbearable pain," but they cannot clearly specify key information such as the location, nature, time of onset, and triggering factors of the pain, which increases the difficulty for doctors to screen information.

[0004] On the other hand, discrepancies exist between patients' subjective feelings and professional medical descriptions, easily leading to errors in the transmission of medical information. Some patients, influenced by tension and anxiety, may overlook important symptoms or exaggerate their discomfort, resulting in discrepancies between the information obtained by doctors and the actual situation, thus affecting the accuracy of the initial diagnosis. This deficiency in descriptive ability is even more pronounced in primary healthcare settings, potentially increasing the risk of misdiagnosis and missed diagnosis, or requiring multiple examinations to fill information gaps, increasing medical costs and reducing patient satisfaction.

[0005] With the deep penetration of digital technologies such as artificial intelligence and big data into the medical field, building a digital and intelligent medical service system has become an important direction for improving the quality of medical services. However, existing digital and intelligent medical services focus more on doctor-side applications such as assisted diagnosis and intelligent triage, while tools to empower patients to describe their conditions are still lacking.

[0006] Therefore, there is an urgent need for a digital approach that can provide patients with accurate and efficient strategies for describing their condition, helping them to standardize their descriptions, improve doctor-patient communication efficiency, and reduce information transmission errors. Summary of the Invention

[0007] In response, the present invention provides a method, system, electronic device, computer storage medium, and computer program product for providing digital and intelligent medical services, in order to solve at least one of the above-mentioned technical problems.

[0008] In a first aspect, the present invention provides a method for providing digital and intelligent medical services, applied to a server, comprising the following steps:

[0009] A disease description guidance knowledge base is constructed, which includes a core symptom dimension tag library, a symptom description standard terminology library, and a consultation logic rule library for various disease types. The core symptom dimension tag library includes symptom location, nature, onset time, triggering factors, accompanying symptoms, and severity classification tags.

[0010] Receive basic medical information input by the patient, including at least the intended department and preliminary symptom keywords; match and analyze the basic medical information with the disease description guidance knowledge base to generate a personalized symptom guidance framework, including a structured question sequence ordered by priority and terminology selection options;

[0011] The system performs semantic parsing and completeness verification on the symptom description information fed back by patients through the personalized symptom guidance framework, identifies the missing core symptom dimension information, generates supplementary guidance instructions for the missing core symptom dimension information, and then obtains complete symptom description data, including: evaluating the basic medical information to obtain an information completeness assessment value, and generating detailed guidance instructions or concise question guidance instructions based on the relationship between the information completeness assessment value and the assessment threshold.

[0012] The complete symptom description data is standardized based on the symptom description standard terminology library to generate a structured medical condition summary. The structured medical condition summary includes symptom dimension label mapping results and key information weight identifiers. The structured medical condition summary is then pushed to the doctor's end.

[0013] A second aspect of the present invention provides a digital intelligent medical service provision system, applied to a server, the system comprising:

[0014] The module constructs a knowledge base for guiding disease description, which includes a core symptom dimension tag library, a symptom description standard terminology library, and a consultation logic rule library for various disease types. The core symptom dimension tag library includes tags for symptom location, nature, onset time, triggering factors, accompanying symptoms, and severity grading.

[0015] The first guidance module receives basic medical information input by the patient, including at least the intended department and preliminary symptom keywords; it matches and analyzes the basic medical information with the disease description guidance knowledge base to generate a personalized symptom guidance framework, including a structured question sequence ordered by priority and terminology selection options;

[0016] The second guidance module performs semantic parsing and completeness verification on the symptom description information fed back by the patient through the personalized symptom guidance framework, identifies the missing core symptom dimension information, generates supplementary guidance instructions for the missing core symptom dimension information, and then obtains complete symptom description data, including: evaluating the basic medical information to obtain an information completeness assessment value, and generating detailed guidance instructions or concise question guidance instructions based on the relationship between the information completeness assessment value and the assessment threshold.

[0017] The generation module standardizes the complete symptom description data based on the symptom description standard terminology library to generate a structured medical condition summary. The structured medical condition summary includes symptom dimension label mapping results and key information weight identifiers, and pushes the structured medical condition summary to the doctor's terminal.

[0018] A third aspect of the present invention provides an electronic device comprising: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the computer program, when executed by the processor, implements the method as described in any of the preceding claims.

[0019] A fourth aspect of the present invention provides a computer storage medium storing a computer program that can be executed by a processor to implement the method as described in any of the preceding claims.

[0020] A fifth aspect of the present invention provides a computer program product comprising a computer program executable by a processor to implement the method as described in any of the preceding claims.

[0021] This invention provides a standardized dimensional framework by constructing a professional knowledge base, and accurately collects symptom information using a personalized guidance framework. Complete data is obtained through semantic parsing, integrity verification, and supplementary guidance, and then standardized processing generates a structured summary. This effectively solves the problems of unprofessional patient descriptions and missing information, shortens consultation time, reduces information transmission errors, helps doctors quickly grasp key symptoms, improves doctor-patient communication efficiency and diagnostic accuracy, and optimizes the medical experience. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating a method for providing digital and intelligent medical services according to an embodiment of the present invention;

[0024] Figure 2 This is a schematic diagram of the system architecture disclosed in an embodiment of the present invention;

[0025] Figure 3 This is a schematic diagram of the structure of a digital intelligent medical service provision system disclosed in an embodiment of the present invention. Detailed Implementation

[0026] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0028] To facilitate understanding of the embodiments of the present invention, further explanations and descriptions will be provided below with reference to the accompanying drawings and specific embodiments. These embodiments do not constitute a limitation on the embodiments of the present invention.

[0029] like Figure 1 As shown in the figure, an embodiment of the present invention discloses a method for providing digital and intelligent medical services, applied to a server, including the following steps:

[0030] 100. Construct a disease description-guided knowledge base, which includes a core symptom dimension tag library, a symptom description standard terminology library, and a consultation logic rule library for various disease types. The core symptom dimension tag library includes symptom location, nature, onset time, triggering factors, accompanying symptoms, and severity grading tags.

[0031] like Figure 2 As shown, the system architecture of this invention includes a server, a patient terminal, and a doctor terminal, with the server connected to both the patient terminal and the doctor terminal. The server generates a structured medical condition summary by interacting with the patient terminal, and then forwards the structured medical condition summary to the doctor terminal, thereby achieving precise information delivery for digital and intelligent medical services.

[0032] The server integrates authoritative resources such as medical guidelines, clinical case data, and disease diagnosis and treatment standards to pre-build a structured knowledge base covering various disease types. Among these, the core symptom dimension tag library identifies key dimensions for symptom description from a medical perspective, including symptom location (e.g., "upper abdomen," "knee"), nature (e.g., "stabbing pain," "distending pain"), onset time (e.g., "30 minutes after meals," "worsening at night"), triggering factors (e.g., "after exposure to cold," "distress"), accompanying symptoms (e.g., "fever with cough," "dizziness with nausea"), and severity level (e.g., "mild pain," "severe fatigue"), providing a standardized dimensional framework for symptom description. The symptom description standardization terminology library includes the correspondence between medical professional terms and colloquial expressions, resolving the discrepancy between patient descriptions and professional terminology. The consultation logic rule library, based on disease diagnosis and treatment logic, establishes association rules between symptom dimensions and a question priority algorithm to ensure the scientific validity of the guidance framework.

[0033] 200, receives basic medical information input by the patient, including at least the intended department and preliminary symptom keywords; matches and analyzes the basic medical information with the disease description guidance knowledge base to generate a personalized symptom guidance framework, including a structured question sequence and terminology selection options sorted by priority.

[0034] The server first receives basic medical information input by the patient through the patient's terminal, including the intended medical department such as internal medicine or orthopedics, as well as preliminary symptom keywords such as headache and diarrhea.

[0035] Next, the server performs multi-dimensional matching analysis between the basic medical information obtained and the disease description guidance knowledge base. Specifically, based on the intended medical department, it locates the corresponding disease category database, matches relevant core symptom dimension tags through preliminary symptom keywords, and combines the priority algorithm in the consultation logic rule base to generate a personalized symptom guidance framework adapted to the patient.

[0036] Among them, the structured question sequence in the personalized symptom guidance framework is ordered according to the importance of medical diagnosis. For example, ask about the location of the core symptoms first, and then ask about the time of onset, so as to avoid the disorder of information collection. The term selection options (such as providing options such as "dull pain / sharp pain / throbbing pain" for the nature of pain) can reduce the difficulty of patients' expression, reduce vague descriptions, and achieve accurate guidance for patients' description of their condition.

[0037] 300. Semantic parsing and completeness verification are performed on the symptom description information fed back by the patient through the personalized symptom guidance framework to identify missing core symptom dimension information; supplementary guidance instructions are generated for the missing core symptom dimension information to obtain complete symptom description data, including: evaluating the basic medical information to obtain an information completeness assessment value, and generating detailed guidance instructions or concise question guidance instructions based on the relationship between the information completeness assessment value and the assessment threshold.

[0038] In this step, the server provides the personalized symptom guidance framework to the patient and, upon receiving the symptom description information from the patient, initiates the following dual processing mechanism:

[0039] Natural Language Processing (NLP) technology is used to semantically parse the symptom descriptions input by patients, transforming text or options into structured data and identifying the symptom dimension labels contained within. Simultaneously, the semantic parsing results are validated for completeness. Specifically, based on a core symptom dimension label library, the acquired symptom dimension labels are compared with the necessary dimensions required for the disease type, marking missing key information (e.g., the patient only describes abdominal pain but does not specify the onset time). For missing information, the server generates targeted supplementary guidance instructions based on a consultation logic rule library, such as "Please specify the exact time the abdominal pain started and its duration," and pushes this to the patient's end. Through interactive questioning, the information is gradually completed, ultimately obtaining complete symptom description data covering all core dimensions, resolving information omissions caused by the patient's lack of expertise or oversight.

[0040] In addition, the process of generating supplementary guidance instructions also includes: evaluating the basic medical information to obtain an information completeness assessment value, and generating detailed guidance instructions or concise question-based guidance instructions based on the relationship between the information completeness assessment value and an assessment threshold. This enables personalized guidance methods to improve guidance efficiency. Further details will be provided later.

[0041] 400. Based on the symptom description standard terminology library, the complete symptom description data is standardized to generate a structured medical condition summary. The structured medical condition summary includes symptom dimension label mapping results and key information weight identifiers. The structured medical condition summary is then pushed to the doctor's end.

[0042] In this step, the server standardizes the complete symptom description data. Specifically, through symptom dimension label mapping, the patient's description is precisely matched with medical terms in the symptom description standard terminology library; for example, "twisting abdominal pain" is mapped to "abdominal spasmodic pain." Key information weighting is then prioritized based on the symptom's contribution to disease diagnosis, assigning priority to important information (such as "persistent high fever with chest pain"). Thus, the resulting structured medical summary is characterized by a unified format, standardized terminology, and a clear focus, conforming to doctors' reading habits while condensing core diagnostic information.

[0043] The server pushes the structured medical condition summary to the doctor, enabling the doctor to quickly grasp key medical elements, reduce information screening time, solve the problems of information distortion and inefficiency in traditional verbal description, achieve accurate pre-diagnosis information transmission between doctors and patients, and ultimately improve the quality and efficiency of medical services.

[0044] Understandably, patients can interact with the server as described above while waiting for their appointment, or before going to a medical institution, thereby providing a structured summary of their condition to the doctor they are about to see or who has made an appointment with, in order to help the doctor understand the patient's condition more accurately.

[0045] This invention provides a standardized dimensional framework by constructing a professional knowledge base, and accurately collects symptom information using a personalized guidance framework. Complete data is obtained through semantic parsing, integrity verification, and supplementary guidance, and then standardized processing generates a structured summary. This effectively solves the problems of unprofessional patient descriptions and missing information, shortens consultation time, reduces information transmission errors, helps doctors quickly grasp key symptoms, improves doctor-patient communication efficiency and diagnostic accuracy, and optimizes the medical experience.

[0046] As an example, the basic medical information is matched and analyzed with the disease description guidance knowledge base to generate a personalized symptom guidance framework, including:

[0047] 201. Semantically expand the keywords of the preliminary symptoms to generate a set of synonyms and related symptom association words. Based on the set of synonyms and association words, search the core symptom dimension tag library to obtain a set of candidate core symptom dimension tags.

[0048] 202. Based on the weight value of the disease type corresponding to the intended medical department, calculate the basic correlation score of each label in the candidate core symptom dimension label set, sort them from high to low according to the basic correlation score, and screen out candidate core symptom dimension labels with basic correlation scores lower than the correlation threshold.

[0049] 203. Based on the disease-symptom correlation algorithm preset in the consultation logic rule base, calculate the diagnostic contribution weight of each core symptom dimension label in the candidate core symptom dimension label set after screening, determine the question priority of each core symptom dimension according to the diagnostic contribution weight, and complete the accurate sorting.

[0050] 204. For each core symptom dimension tag after precise sorting, appropriate terminology options are retrieved from the symptom description standard terminology library and combined to form a personalized symptom guidance framework.

[0051] In this embodiment, natural language processing technology is used to semantically expand the initial symptom keywords (such as "stomach ache") input by the patient, generating a set of synonyms (such as "abdominal pain," "abdominal discomfort") and related symptom association words (such as "bloating," "nausea"), thus broadening the search scope. Based on the expanded vocabulary, a matching search is performed in the core symptom dimension tag library to filter out all relevant core symptom dimension tags (such as "symptom location - abdomen," "nature - colic," etc.), forming a candidate core symptom dimension tag set, thus solving the search bias problem caused by patients' non-standard descriptions.

[0052] Based on the patient's intended department (e.g., internal medicine) and the corresponding disease type weight (e.g., digestive system diseases have a higher weight than respiratory system diseases within internal medicine), a basic correlation score is calculated for each tag in the candidate tag set (correlation score = symptom-disease matching degree × preset weight). After sorting the basic correlation scores from high to low, tags below the correlation threshold are filtered out (e.g., orthopedic symptom tags with extremely low correlation to internal medicine are excluded), achieving initial simplification of candidate tags and ensuring that the retained tags are highly correlated with the disease type of the department selected by the patient.

[0053] Based on the disease-symptom correlation algorithm in the consultation logic rule base, a diagnostic contribution weight is calculated for the filtered tags (diagnostic contribution weight = symptom specificity to disease diagnosis × clinical decision value). The priority of questions is determined by sorting according to the diagnostic contribution weight, so that questions on symptom dimensions that are more critical to disease diagnosis (such as onset time and triggering factors) are asked first, thus achieving accurate sorting.

[0054] Taking "acute appendicitis" as an example, the calculation logic of the diagnostic contribution weight is explained:

[0055] Assuming the patient complains of "abdominal pain" and chooses to seek surgical treatment, after initial screening, the three core symptom dimensions of "right lower quadrant tenderness", "fever", and "nausea" are retained as labels.

[0056] "Right lower quadrant tenderness": It occurs in 92% of appendicitis patients (specificity 0.92) and is a key sign for differentiating appendicitis from other abdominal pain diseases (clinical decision value 0.9). Diagnostic contribution weight = 0.92 × 0.9 = 0.828.

[0057] "Fever": It occurs in 65% of patients with appendicitis (specificity 0.65), but it can also occur in a variety of infectious diseases (clinical decision value 0.5), and its diagnostic contribution weight is 0.65 × 0.5 = 0.325.

[0058] "Nausea": Occurs in 58% of appendicitis patients (specificity 0.58), with low differentiation for diagnosis (clinical decision value 0.3), and a diagnostic contribution weight of 0.58 × 0.3 = 0.174.

[0059] After sorting by weight, the priority of the questions is "right lower quadrant tenderness" > "fever" > "nausea", which is consistent with the logic of focusing on characteristic signs first in clinical diagnosis.

[0060] Finally, for each core symptom dimension label after precise sorting, appropriate terminology options are retrieved from the symptom description standard terminology library (such as matching the "nature" dimension with the "dull pain / colic / distending pain" option). The sorted symptom dimensions are combined with the corresponding terminology options to form a structured question sequence that prioritizes key issues and then addresses secondary ones. This ultimately generates a personalized symptom guidance framework that is both professional and targeted, ensuring that the guidance logic is consistent with the clinical diagnostic approach.

[0061] As an example, semantic parsing and completeness verification are performed on the symptom descriptions provided by patients through a personalized symptom guidance framework to identify missing core symptom dimensions, including:

[0062] 301. Receive symptom description information from patients through a personalized symptom guidance framework, including text description information and option selection results;

[0063] 302. Natural language processing is used to segment, identify entities and perform semantic mapping on the text description information, and semantic mapping is performed on the option selection results based on preset rules to associate the symptom description information with the corresponding core symptom dimension tags in the core symptom dimension tag library.

[0064] 303. Based on the precisely sorted core symptom dimension label sequence, construct an integrity verification matrix, in which each verification item corresponds to a core symptom dimension label; compare each associated label obtained through semantic mapping with the integrity verification matrix item by item, and mark the verification items that do not match the associated label.

[0065] 304. Based on the core symptom dimension necessity level classification preset in the consultation logic rule base, the marked unmatched verification items are filtered to determine the missing core symptom dimension information that needs to be supplemented.

[0066] In this embodiment, two types of information are received from the patient through a personalized symptom guidance framework: one is the text description information entered by the patient, such as "My stomach has been hurting intermittently since last night"; the other is the selection result of the option selected from the guidance framework (such as selecting "colic" in the "Pain nature" option).

[0067] For text descriptions, natural language processing techniques are used to sequentially perform word segmentation (breaking down words like "last night / start / stomach / batch / pain"), entity recognition (identifying "stomach" as "abdomen" and "pain" as "ache"), and semantic mapping (associating with tags like "symptom location - abdomen" and "onset time - last night"). For option selection results, semantic mapping is directly performed based on preset rules (e.g., selecting the "colic" option directly associates it with the "nature - colic" tag). Through these two mapping methods, diverse symptom descriptions are uniformly associated with a core symptom dimension tag library, achieving the transformation of unstructured information into structured tags.

[0068] Based on the core symptom dimension label sequence (e.g., "right lower quadrant tenderness → fever → nausea") precisely sorted in step 203, a completeness verification matrix is ​​constructed. Each verification item in the matrix corresponds to a core symptom dimension label that must be collected. Each associated label obtained from semantic mapping is compared with the matrix verification item by item. For example, if the patient does not describe information related to "fever", the verification item corresponding to "fever" is marked as unmatched, thus clarifying the specific dimension of missing information.

[0069] By combining the pre-defined necessity levels (such as "core necessity" and "minor supplement") in the consultation logic rule base, marked unmatched verification items are filtered. For example, "right lower abdominal tenderness" is classified as "core necessity," and must be supplemented if it does not match; "nausea," if classified as "minor supplement," can be simplified and supplemented as appropriate. By using level filtering, the missing core symptom dimensions that need to be supplemented are accurately determined, avoiding inefficiency caused by indiscriminate questioning.

[0070] As an example, supplementary guidance instructions are generated for missing core symptom dimensions to obtain complete symptom description data, including:

[0071] 305. Extract the departmental urgency and symptom keyword risk level features from the basic medical information, and extract semantic completeness, terminology standardization, and description accuracy features from the symptom description information; wherein, the description accuracy feature is calculated by the matching deviation rate between the patient's description and the standardized terminology database.

[0072] 306. Based on preset weights, the extracted multi-dimensional features are fused and analyzed to obtain the information completeness evaluation value;

[0073] 307. If the information completeness assessment value is lower than the assessment threshold, generate detailed guidance instructions with example descriptions; if the information completeness assessment value is higher than the assessment threshold, generate concise question guidance instructions.

[0074] 308. Generate supplementary guidance instructions based on detailed guidance instructions or concise question guidance instructions, push the supplementary guidance instructions to the patient's end, receive supplementary symptom information from the patient, until complete symptom description data is obtained.

[0075] In real-world medical settings, patients exhibit significant differences in their medical knowledge. Some patients struggle to accurately describe symptom details (e.g., describing "colic" as "a twisting pain in the abdomen") or omit crucial information (e.g., forgetting to specify the duration of pain). Conversely, some patients possess the ability to accurately describe their symptoms. This results in inconsistent quality of raw information obtained by doctors, impacting diagnostic efficiency. To address this, this embodiment incorporates a multi-dimensional feature extraction process to comprehensively assess the quality of the information provided by patients and dynamically adjust subsequent guidance strategies. Specifically:

[0076] Two key features are extracted from basic medical information: the urgency level of the department (e.g., "high urgency" for the emergency department and "normal urgency" for the general internal medicine outpatient clinic) and the risk level of symptom keywords (e.g., "chest pain" and "high fever" are marked as "high risk" and "mild headache" is marked as "low risk").

[0077] Simultaneously, three core features are extracted from the symptom description information: semantic completeness, terminology standardization, and descriptive accuracy. Semantic completeness measures the proportion of provided symptom dimensions, with a score of (number of provided core symptom dimension labels / total number of core symptom dimension labels after precise ranking) × 100. Terminology standardization characterizes the degree of matching between the patient's description and medical terminology, with a score of (number of words matching medical terminology in the patient's description / total number of effective words in the patient's description) × 100. Descriptive accuracy is calculated by the matching deviation rate between the patient's description and the standardized terminology database for symptom descriptions; a lower deviation rate indicates higher accuracy, with a score of [1 - (number of words deviating from the standardized terminology database in the patient's description / total number of effective words in the patient's description)] × 100 = (number of words matching the standardized terminology database in the patient's description / total number of effective words in the patient's description) × 100.

[0078] Based on the pre-defined feature weights in the consultation logic rule base (such as risk level weight being higher than normativity weight), the extracted multi-dimensional features are subjected to weighted fusion analysis. For example, a high risk level corresponds to a high base score, and semantic completeness is scored according to the completion ratio. The information completeness assessment value is obtained by weighted summation, which quantifies the quality and completeness of the current symptom information.

[0079] Examples are given below:

[0080] The patient complained of "chest tightness and pain for 2 hours" and selected "emergency department" as the department of preference. The symptom description information provided by the personalized symptom guidance framework was: "I started feeling discomfort in my chest last night, with intermittent pain and some shortness of breath." The option selection result was "Pain nature: dull pain".

[0081] Features were extracted from basic medical information: (1) Departmental urgency: The patient selected “Emergency Department”, which corresponds to “high urgency”, with a quantitative score of 80 points (out of 100). (2) Symptom keyword risk level: The core keywords “chest tightness and pain” and “shortness of breath” are high-risk symptoms of the cardiovascular system, marked as “high risk”, with a quantitative score of 90 points.

[0082] Features were extracted from the symptom description information: (1) Semantic completeness: The core symptom dimension label sequence after precise sorting is “pain location → pain nature → onset time → accompanying symptoms → duration”. The patient provided 4 items, accounting for 80%, and the quantitative score was 80 points. (2) Terminology standardization: The patient’s expression “chest discomfort” corresponds to the medical term “chest discomfort” and “pain in waves” corresponds to “paroxysmal pain”. The matching degree is 70%, and the quantitative score is 70 points. (3) Description accuracy feature: By comparing with the standard terminology database for symptom description, “chest” deviates from the standard term “lower sternum”, while “dull pain” matches the standard term. The deviation rate is calculated to be 20% (deviation rate = 1 - accuracy). Therefore, the accuracy quantitative score is 80 points (100-20).

[0083] Preset feature weights (from the consultation logic rule base) are shown in Table 1 below as an example:

[0084] Table 1

[0085]

[0086] Information completeness assessment score = 80×0.2+90×0.3+80×0.25+70×0.1+80×0.15=16+27+20+7+12=82 points.

[0087] The information completeness assessment score (82 points) quantifies the quality of the current symptom information: because the patient chose the emergency department (high urgency) and the core symptom risk level is high, even with a small number of description deviations, the overall information completeness is still at a high level, providing data basis for generating supplementary guidance instructions (such as a concise question on "duration").

[0088] If the information completeness assessment score is below the assessment threshold, it indicates that the information is severely lacking or the description is confusing. A detailed guidance instruction with examples is generated, such as "Please describe the location of the pain (example: upper abdomen / lower right abdomen)." If the information completeness assessment score is above the assessment threshold, it indicates that the information is basically usable but has a few missing parts. A concise question guidance instruction is generated, such as "Please provide the time of onset of the pain." This differentiated strategy can balance guidance efficiency and information quality.

[0089] Supplementary guidance instructions are generated based on detailed or concise question-based guidance instructions. These instructions are then pushed to the patient's terminal. After receiving supplementary symptom information from the patient, the semantic parsing and completeness verification process is repeated. If key deficiencies still exist, supplementary instructions are generated again based on the updated assessment values ​​until complete symptom description data covering all core symptom dimensions is obtained, ensuring the accuracy of subsequent disease summaries.

[0090] As an example, the evaluation threshold is determined in the following manner:

[0091] Obtain historical diagnostic data from doctors in the target department and extract diagnostic efficiency indicators for doctors' descriptions of different symptoms, including average consultation time, accuracy of key symptom identification, and consistency of diagnostic conclusions.

[0092] Based on the diagnostic efficiency index, calculate the doctor's diagnostic efficacy score corresponding to different information completeness levels;

[0093] The minimum information completeness corresponding to the diagnostic efficacy score reaching the preset benchmark value was selected as the initial evaluation threshold;

[0094] A correction coefficient is generated based on the physician characteristic parameters of the target department. The initial assessment threshold is dynamically adjusted using the correction coefficient to obtain the assessment threshold. The physician characteristic parameters include at least the total number of physicians in the department, the proportion of physicians with the title of attending physician or above, and the average number of years of clinical experience.

[0095] First, retrieve all historical medical records of doctors in the target department (such as cardiology or gastroenterology) from the hospital information system, and extract the following diagnostic efficiency indicators: average consultation time (the time taken from information collection to preliminary diagnosis in a single consultation), key symptom identification accuracy (the proportion of doctors who accurately capture core symptom dimensions from patient descriptions), and consistency of diagnostic conclusions (the degree of matching between diagnostic results for the same case among different doctors). For example, for cardiology, focus on extracting medical data related to symptoms such as chest pain and palpitations.

[0096] Next, the extracted diagnostic efficiency indicators were standardized (converted to a score of 0-100) and weighted according to departmental diagnostic priorities (e.g., the accuracy of key symptom identification is weighted higher than the average consultation time). A weighted summation was used to calculate the physician diagnostic efficacy score corresponding to different information completeness levels (e.g., 60, 70, and 80 points) to quantify the impact of information quality on diagnostic effectiveness. For example, when the information completeness is 75 points, the cardiologist's diagnostic efficacy score reaches 85 points, indicating that the symptom description data at this information completeness level can support basic diagnostic needs.

[0097] The lowest information completeness corresponding to the first time the diagnostic efficacy score reaches the preset benchmark value is selected from the calculation results and determined as the preliminary assessment threshold. For example, if the diagnostic efficacy score reaches 80 points for the first time when the information completeness is 70 points, then 70 points is used as the preliminary assessment threshold.

[0098] Next, collect the characteristic parameters of doctors in the target department, including the total number of doctors in the department (reflecting the scale of medical resources), the proportion of doctors with the title of attending physician or above (reflecting the level of doctors' qualifications), and the average number of years of clinical experience (reflecting the accumulation of clinical experience). Based on these parameters, a correction coefficient is generated in the following way: if the proportion of doctors with high professional titles in the department is high and the average number of years of clinical experience is long, the correction coefficient can be appropriately reduced (e.g., 0.9), allowing for a slightly lower information completeness while still ensuring diagnostic quality; conversely, the correction coefficient is increased (e.g., 1.1), strengthening the information collection requirements by raising the threshold. See the example in the following comparison table 2 for details:

[0099] Table 2

[0100]

[0101] Note: In Table 2, "Percentage of attending physicians and above" and "Average years of clinical experience" are the core influencing factors, which are given priority in determining the baseline range of the correction coefficient; "Total number of doctors in the department" is used as an auxiliary factor.

[0102] The range of correction coefficient values ​​is set based on doctors' ability to process incomplete information in clinical practice. Highly qualified and experienced teams can accept a slightly lower level of information completeness (correction coefficient < 1), while less qualified and experienced teams require a higher level of information completeness (correction coefficient > 1). In practical applications, the threshold range (such as the proportion of professional titles and the segmentation of clinical years) can be dynamically adjusted according to the specific circumstances of the hospital department to ensure that the correction coefficient matches the department's diagnostic and treatment capabilities.

[0103] The correction coefficient is multiplied by the initial assessment threshold to obtain the final assessment threshold, which is then stored in the consultation logic rule base as the judgment benchmark for subsequent guidance strategy switching.

[0104] like Figure 3 As shown, this embodiment of the invention also provides a digital intelligent medical service provision system, applied to a server, the system comprising:

[0105] Module 11 is used to build a disease description guidance knowledge base, which includes a core symptom dimension tag library, a symptom description standard terminology library, and a consultation logic rule library for various disease types. The core symptom dimension tag library includes symptom location, nature, onset time, triggering factors, accompanying symptoms, and severity grading tags.

[0106] The first guidance module 12 receives basic medical information input by the patient, including at least the intended medical department and preliminary symptom keywords; it matches and analyzes the basic medical information with the disease description guidance knowledge base to generate a personalized symptom guidance framework, including a structured question sequence ordered by priority and terminology selection options.

[0107] The second guidance module 13 performs semantic parsing and completeness verification on the symptom description information fed back by the patient through the personalized symptom guidance framework, identifies the missing core symptom dimension information, generates supplementary guidance instructions for the missing core symptom dimension information, and then obtains complete symptom description data, including: evaluating the basic medical information to obtain an information completeness assessment value, and generating detailed guidance instructions or concise question guidance instructions based on the relationship between the information completeness assessment value and the assessment threshold.

[0108] The generation module 14 standardizes the complete symptom description data based on the symptom description standard terminology library to generate a structured medical condition summary. The structured medical condition summary includes symptom dimension label mapping results and key information weight identifiers. The structured medical condition summary is then pushed to the doctor's end.

[0109] As an example, the first boot module 12 specifically:

[0110] Semantic expansion is performed on the initial symptom keywords to generate a set of synonyms and related symptom association words. Based on the set of synonyms and association words, the core symptom dimension tag library is retrieved to obtain a set of candidate core symptom dimension tags.

[0111] Based on the weight value of the disease type corresponding to the intended medical department, the basic correlation score of each label in the candidate core symptom dimension label set is calculated. The labels are initially sorted from high to low according to the basic correlation score, and candidate core symptom dimension labels with basic correlation scores lower than the correlation threshold are screened out.

[0112] Based on the disease-symptom correlation algorithm preset in the consultation logic rule base, the diagnostic contribution weight of each core symptom dimension label in the candidate core symptom dimension label set after screening is calculated, and the question priority of each core symptom dimension is determined according to the diagnostic contribution weight to complete the accurate sorting.

[0113] For each core symptom dimension tag after precise sorting, appropriate terminology options are retrieved from the symptom description standard terminology library and combined to form a personalized symptom guidance framework.

[0114] As an example, the second boot module 13 specifically:

[0115] Receive symptom descriptions from patients through a personalized symptom guidance framework, including text descriptions and option selection results;

[0116] Natural language processing is used to segment, identify entities, and perform semantic mapping on text description information, and semantic mapping is performed on option selection results based on preset rules to associate the symptom description information with the corresponding core symptom dimension tags in the core symptom dimension tag library.

[0117] Based on the precisely sorted core symptom dimension label sequence, an integrity verification matrix is ​​constructed, in which each verification item corresponds to a core symptom dimension label; each associated label obtained through semantic mapping is compared with the integrity verification matrix item by item, and verification items that do not match an associated label are marked.

[0118] By combining the necessity level classification of core symptom dimensions in the pre-set core symptom dimension rule base, the marked unmatched verification items are filtered to determine the missing core symptom dimension information that needs to be supplemented.

[0119] As an example, the second boot module 13 specifically:

[0120] The basic medical information is used to extract the urgency of departmental intent and the risk level of symptom keywords. The symptom description information is used to extract the semantic completeness, terminology standardization, and description accuracy features. The description accuracy feature is calculated by the matching deviation rate between the patient's description and the standardized terminology database.

[0121] Based on preset weights, the extracted multi-dimensional features are fused and analyzed to obtain an information completeness assessment value;

[0122] If the information completeness assessment value is lower than the assessment threshold, a detailed guidance instruction with example descriptions is generated; if the information completeness assessment value is higher than the assessment threshold, a concise question guidance instruction is generated.

[0123] Supplementary guidance instructions are generated based on detailed or concise question-based guidance instructions. These supplementary instructions are then pushed to the patient's device, which receives additional symptom information from the patient until complete symptom description data is obtained.

[0124] As an example, the evaluation threshold is determined in the following manner:

[0125] Obtain historical diagnostic data from doctors in the target department and extract diagnostic efficiency indicators for doctors' descriptions of different symptoms, including average consultation time, accuracy of key symptom identification, and consistency of diagnostic conclusions.

[0126] Based on the diagnostic efficiency index, calculate the doctor's diagnostic efficacy score corresponding to different information completeness levels;

[0127] The minimum information completeness corresponding to the diagnostic efficacy score reaching the preset benchmark value was selected as the initial evaluation threshold;

[0128] A correction coefficient is generated based on the physician characteristic parameters of the target department. The initial assessment threshold is dynamically adjusted using the correction coefficient to obtain the assessment threshold. The physician characteristic parameters include at least the total number of physicians in the department, the proportion of physicians with the title of attending physician or above, and the average number of years of clinical experience.

[0129] This invention also provides an electronic device comprising: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the computer program, when executed by the processor, implements the method as described in any of the foregoing embodiments.

[0130] This invention also provides a computer storage medium storing a computer program that can be executed by a processor to implement the methods described in any of the foregoing claims.

[0131] This invention also provides a computer program product comprising a computer program that can be executed by a processor to implement the method as described in any of the foregoing embodiments.

[0132] Although the invention has been specifically shown and described with reference to preferred embodiments, those skilled in the art will understand that various modifications in form and detail may be made without departing from the spirit and scope of the invention. Accordingly, the disclosed invention should be considered merely illustrative and limited only by the scope specified in the appended claims.

Claims

1. A digital medical service providing method, applied to a server, and characterized in that, The method comprises the following steps: constructing a disease description guide knowledge base, including a core symptom dimension label library corresponding to various disease types, a symptom description standard term library, and an inquiry logic rule library, wherein the core symptom dimension label library includes symptom site, nature, onset time, inducing factor, accompanying symptom, and degree classification label; receiving basic medical information input by the patient, including at least medical department intention and preliminary symptom keywords; matching and analyzing the basic medical information with the disease description guide knowledge base to generate an individualized symptom guide framework, including a structured question sequence and term selection items sorted by priority; performing semantic analysis and integrity check on the symptom description information fed back by the patient through the individualized symptom guide framework to identify missing core symptom dimension information; generating supplementary guide instructions for the missing core symptom dimension information to obtain complete symptom description data, including: evaluating the basic medical information to obtain an information completeness evaluation value, and generating detailed guide instructions or concise question guide instructions based on the relationship between the information completeness evaluation value and the evaluation threshold; standardizing the complete symptom description data based on the symptom description standard term library to generate a structured disease summary, including symptom dimension label mapping results and key information weight identification, and pushing the structured disease summary to the doctor end; matching and analyzing the basic medical information with the disease description guide knowledge base to generate an individualized symptom guide framework, including: performing semantic expansion on the preliminary symptom keywords to generate a synonym set and related symptom association words, retrieving the core symptom dimension label library based on the synonym set and the association words to obtain a candidate core symptom dimension label set; combining the disease type weight value corresponding to the medical department intention to calculate the basic association score of each label in the candidate core symptom dimension label set, preliminarily sorting the labels from high to low, and screening out the candidate core symptom dimension labels with a basic association score lower than an association threshold; calculating the diagnosis contribution weight of each core symptom dimension label in the screened candidate core symptom dimension label set according to the preset disease-symptom correlation degree algorithm in the inquiry logic rule library, determining the questioning priority of each core symptom dimension according to the diagnosis contribution weight, and completing accurate sorting; for each core symptom dimension label after accurate sorting, retrieving the adaptive term selection items from the symptom description standard term library to form an individualized symptom guide framework.

2. The digital medical service providing method of claim 1, wherein: performing semantic analysis and integrity check on the symptom description information fed back by the patient through the individualized symptom guide framework to identify missing core symptom dimension information, including: receiving symptom description information fed back by the patient through the individualized symptom guide framework, including text description information and option selection results; performing word segmentation, entity recognition, and semantic mapping on the text description information using natural language processing, and performing semantic mapping on the option selection results based on preset rules to associate the symptom description information to the corresponding core symptom dimension label in the core symptom dimension label library; Based on the accurate sorting of the core symptom dimension label sequence, a completeness verification matrix is constructed, and each verification item in the matrix corresponds to a core symptom dimension label; each associated label obtained through semantic mapping is compared with the completeness verification matrix item by item, and the verification items that do not match the associated label are marked; Combined with the preset core symptom dimension necessity level division in the inquiry logic rule library, the marked unmatched verification items are screened to determine the missing core symptom dimension information that needs to be supplemented.

3. The digital medical service providing method of claim 2, wherein: For the missing core symptom dimension information, a supplementary guidance instruction is generated to obtain complete symptom description data, including: Extracting department intention urgency and symptom keyword risk level features from the basic medical information, and extracting semantic completeness, term standardization and description accuracy features from the symptom description information; wherein the description accuracy feature is calculated by the matching deviation rate of patient expression and standard term library; Based on the preset weight, the multi-dimensional features extracted are fused and analyzed to obtain an information completeness evaluation value; If the information completeness evaluation value is lower than the evaluation threshold, a detailed guidance instruction containing an example is generated; if the information completeness evaluation value is higher than the evaluation threshold, a concise question guidance instruction is generated; Based on the detailed guidance instruction or the concise question guidance instruction, a supplementary guidance instruction is generated, which is pushed to the patient end to receive the supplementary symptom information fed back by the patient until complete symptom description data is obtained.

4. The digital medical service providing method of claim 3, wherein: The evaluation threshold is determined by the following method: Obtain the historical diagnosis data of the target department doctor, extract the doctor's diagnosis efficiency index for different symptom description information, including average inquiry time, key symptom recognition accuracy and diagnosis conclusion consistency; Based on the diagnosis efficiency index, the doctor's diagnosis efficiency score corresponding to different information completeness levels is calculated, and the lowest information completeness corresponding to the diagnosis efficiency score reaching the preset benchmark value is selected as the preliminary evaluation threshold; Generate a correction coefficient according to the doctor's characteristic parameters of the target department, and use the correction coefficient to dynamically adjust the preliminary evaluation threshold to obtain the evaluation threshold; wherein the doctor's characteristic parameters at least include: total number of department doctors, proportion of attending and above titles, and average clinical years. 5.A digital medical service providing system applied to a server, characterized in that, The system comprises: A construction module constructs a disease description guidance knowledge base, including a core symptom dimension label library corresponding to various disease types, a symptom description standard term library and an inquiry logic rule library, wherein the core symptom dimension label library includes symptom site, nature, onset time, inducing factor, accompanying symptom and degree classification label; A first guidance module receives the basic medical information input by the patient, including at least the department intention and the preliminary symptom keywords; the basic medical information is matched and analyzed with the disease description guidance knowledge base to generate a personalized symptom guidance framework, including a priority-ordered structured question sequence and a term selection item; The second guiding module performs semantic analysis and integrity check on symptom description information fed back by the patient through the personalized symptom guiding framework, and identifies missing core symptom dimension information; generates supplementary guiding instructions for the missing core symptom dimension information, and then obtains complete symptom description data, including: evaluating the basic medical information to obtain an information completeness evaluation value, and generating detailed guiding instructions or concise question guiding instructions based on the relationship between the information completeness evaluation value and the evaluation threshold; The generating module standardizes the complete symptom description data based on a symptom description specification term library, generates a structured disease summary, and pushes the structured disease summary to a doctor terminal, wherein the structured disease summary includes symptom dimension label mapping results and key information weight identifiers; The first guiding module specifically: Performs semantic expansion on the preliminary symptom keywords, generates a synonym set and related symptom association words, retrieves a core symptom dimension label library based on the synonym set and the association words, and obtains a candidate core symptom dimension label set; Combines disease type weight values corresponding to the medical department intention to calculate the basic association scores of each label in the candidate core symptom dimension label set, preliminarily sorts the labels from high to low, and screens out candidate core symptom dimension labels with a basic association score lower than an association threshold; According to a preset disease-symptom association degree algorithm in an inquiry logic rule library, calculates the diagnosis contribution weight of each core symptom dimension label in the screened candidate core symptom dimension label set, determines the questioning priority of each core symptom dimension according to the diagnosis contribution weight, and completes accurate sorting; For each core symptom dimension label after accurate sorting, an adaptive term selection item is retrieved from a symptom description specification term library to form a personalized symptom guiding framework.

6. An electronic device, the electronic device comprising: At least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the computer program is executed by the processor to implement the method of any one of claims 1-4.

7. A computer storage medium, characterized in that: The computer storage medium stores a computer program executable by the processor to implement the method of any one of claims 1-4.

8. A computer program product, characterised in that: The computer program product includes a computer program executable by the processor to implement the method of any one of claims 1-4.

Images