Semantic tagging and multi-axis vector mapping-based unstructured korean medicine clinical data progress management system and method thereof

Abstract

A system and method for managing the progress of unstructured Korean medical clinical data based on semantic tagging and multi-axis vector mapping are disclosed. A system for managing the progress of unstructured Korean medical clinical data based on semantic tagging and multi-axis vector mapping according to one embodiment of the present invention comprises: an input unit for receiving unstructured Korean medical clinical data including at least one of a patient's symptoms, medical history, and treatment history; an analysis unit for analyzing the unstructured Korean medical clinical data into at least one clinical semantic item by natural language processing; a tagging unit for generating tagging data by semantically tagging the clinical semantic item to correspond to a preset symptom axis; a vectorization unit for converting the tagging data generated by the tagging unit into multi-axis vector data including extracted values ​​and weights for each symptom axis; a progress record unit for storing the multi-axis vector data converted by the vectorization unit by treatment stage and generating progress data by treatment stage including changes in the patient's condition after treatment; and a progress analysis unit for generating feedback data regarding at least one of changes in symptoms, treatment effects, and responses to treatment stimuli by comparing and analyzing the progress data by treatment stage generated by the progress record unit.

Description

Technology Field

[0001] The present invention relates to a technology for managing clinical data in Korean medicine, and more specifically, to a semantic tagging and multi-axis vector mapping-based system and method for managing the progress of unstructured clinical data in Korean medicine, which processes unstructured clinical data including a patient's symptoms, medical history, and treatment history using natural language processing, converts it into structured data through semantic tagging and multi-axis vector mapping, and generates feedback data regarding the patient's progress status and treatment response by comparing and analyzing it by treatment stage. Background Technology

[0002] In the clinical practice of Korean medicine, patients' symptoms, medical history, constitutional characteristics, treatment history, and post-treatment responses can be accumulated in various forms, such as medical history interviews, consultations, procedure records, and progress observation records. However, as such clinical data is often stored in the form of descriptive sentences, short notes, abbreviated expressions, or subjective records specific to each practitioner, it mostly exists as unstructured information with an inconsistent data structure.

[0003] In particular, since the manner of expression for the same symptoms may vary among users, and detailed information such as the pattern, intensity, frequency, triggers, alleviating factors, and associated symptoms of pain may be recorded in a mixed manner without a consistent system, there are limitations in making quantitative comparisons or systematic cumulative management for use in subsequent treatment.

[0004] Furthermore, conventional clinical recording methods often limit the recording of the condition at a specific point in time to fragmentary details, making it difficult to track changes in symptoms or trends in treatment response over time as treatment cycles progress.

[0005] For example, even when multiple treatment sessions are performed on the same patient, if records for each treatment session are stored separately in the form of individual memos or independent charts, it may require a significant amount of time and manpower to directly compare the state at the time of the previous treatment with the current state, or to derive repetitive response patterns to specific treatment stimuli.

[0006] As a result, even though clinicians may be able to intuitively grasp the patient's progress, they may struggle to objectively analyze that progress based on structured data or utilize it as reference information for setting the direction of treatment.

[0007] Furthermore, while conventional technologies may utilize electronic medical records or general medical data management systems, these systems are often designed primarily around standardized code input or fixed formats, which presents a problem in that it is difficult to adequately reflect the free-form symptom descriptions that frequently occur in the clinical setting of Korean medicine.

[0008] In particular, since Korean medicine clinical data can include various semantic elements such as patients' subjective expressions, practitioners' judgments, pre- and post-treatment responses, and the simultaneous recording of complex symptoms, a structure is required that allows for multidimensional interpretation by breaking it down into clinical semantic items, rather than remaining at the level of simple text storage.

[0009] Nevertheless, there was a problem in that conventional technology for semantically tagging such unstructured data, vectorizing it according to multiple symptom axes, and linking it with progress data by treatment stage was not sufficiently established.

[0010] Conventional methods for managing clinical data in Korean medicine have problems such as large variations in the representation of unstructured data, difficulty in structurally decomposing clinical significance, difficulty in comparing changes in condition by treatment cycle based on consistent criteria, and difficulty in cumulatively analyzing treatment effects and responses to therapeutic stimuli quantitatively or semi-quantitatively.

[0011] Therefore, there is a need for technology capable of effectively providing feedback on patients' progress and treatment responses by structuring unstructured Korean medical clinical data through natural language processing and semantic tagging, standardizing it into multi-axial vector data, and systematically generating and comparing progress data by treatment cycle. Prior art literature

[0012] Korean Registered Patent Publication No. 10-2936460 Korean Registered Patent Publication No. 10-2339971 The problem to be solved

[0013] The present invention aims to provide a semantic tagging and multi-axis vector mapping-based system and method for managing the progress of unstructured Korean medical clinical data, which can standardize Korean medical clinical information recorded in a free-form narrative into a structured form by processing unstructured Korean medical clinical data, including a patient's symptoms, medical history, and treatment history, into clinical semantic items and performing semantic tagging.

[0014] Furthermore, the purpose is to provide a semantic tagging and multi-axis vector mapping-based system and method for managing the progress of unstructured Korean medical clinical data, which can multidimensionally represent a patient's symptom state and quantitatively compare the state before and after treatment by converting clinical semantic items included in the unstructured Korean medical clinical data into multi-axis vector data by correlating them with multiple pre-established symptom axes.

[0015] Furthermore, the purpose is to provide a semantic tagging and multi-axis vector mapping-based system and method for managing the progress of unstructured Korean medicine clinical data, which can manage the progress of treatment over multiple cycles in a time-series manner and continuously track changes in the patient's condition by generating and accumulating progress data including multi-axis vector data and changes in the patient's condition after treatment for each treatment cycle.

[0016] Furthermore, the purpose is to provide a semantic tagging and multi-axis vector mapping-based atypical Korean medicine clinical data progress management system and method that can objectively identify patient treatment response trends and generate feedback data for subsequent treatment by comparing and analyzing progress data by treatment cycle to determine symptom changes, treatment effects, and responses to treatment stimuli.

[0017] Furthermore, the purpose is to provide a semantic tagging and multi-axis vector mapping-based system and method for managing the progression of unstructured Korean medicine clinical data, which enables the integrated analysis of a patient's condition centered on information meaningful for clinical judgment, going beyond simple record storage, by calculating weights reflecting the clinical importance and pathological associations of clinical significance items for each symptom axis and applying them to multi-axis vector data.

[0018] The problems that the present invention aims to solve are not limited to those mentioned above, and other problems that the present invention aims to solve that are not mentioned herein will be clearly understood by those skilled in the art to which the present invention belongs from the description below. means of solving the problem

[0019] A progress management system for unstructured Korean medical clinical data based on semantic tagging and multi-axis vector mapping according to an embodiment of the present invention comprises: an input unit for receiving unstructured Korean medical clinical data including at least one of a patient's symptoms, medical history, and treatment history; an analysis unit for analyzing the unstructured Korean medical clinical data into at least one clinical semantic item by natural language processing; a tagging unit for generating tagging data by semantically tagging the clinical semantic item to correspond to a preset symptom axis; a vectorization unit for converting the tagging data generated by the tagging unit into multi-axis vector data including extracted values ​​and weights for each symptom axis; a progress record unit for storing the multi-axis vector data converted by the vectorization unit by treatment stage and generating progress data by treatment stage including changes in the patient's condition after treatment; and a progress analysis unit for generating feedback data for at least one of changes in symptoms, treatment effects, and responses to treatment stimuli by comparing and analyzing the progress data by treatment stage generated by the progress record unit.

[0020] In addition, the pre-established symptom axis is characterized by including location, pattern, intensity, frequency, triggering factors, alleviating factors, associated symptoms, and patient history.

[0021] In addition, the analysis unit analyzes unstructured Korean medical clinical data into clinical semantic items having a hierarchical structure including major, medium, and minor categories, and the tagging unit generates tagging data by automatically matching the clinical semantic items to correspond to the symptom axis.

[0022] In addition, the vectorization unit is characterized by including a weight calculation unit that calculates symptom axis weights by reflecting at least one of the clinical importance and pathological association of clinical semantic items included in the tagging data, and a vector generation unit that generates multi-axis vector data based on the symptom axis weights calculated by the weight calculation unit.

[0023] In addition, the progress analysis unit is characterized by including a treatment response judgment unit that generates judgment result data by determining at least one of symptom improvement, deterioration, persistence, and treatment response trend by chronologically comparing and analyzing progress data for each treatment stage corresponding to the treatment stage, and a feedback generation unit that calculates integrated judgment data representing the patient's progress status based on the judgment result data generated by the treatment response judgment unit and generates feedback data based on the integrated judgment data. Effects of the invention

[0024] The semantic tagging and multi-axis vector mapping-based system and method for managing the progress of unstructured Korean medical clinical data according to the present invention has the effect of standardizing Korean medical clinical information, which is recorded in a free-form narrative, into a structured form by processing unstructured Korean medical clinical data, including a patient's symptoms, medical history, and treatment history, into clinical semantic items and performing semantic tagging.

[0025] Furthermore, by correlating clinical semantic items included in atypical Korean medical clinical data with multiple pre-established symptom axes and converting them into multi-axis vector data, it has the effect of representing the patient's symptom state in a multidimensional manner and quantitatively comparing the state before and after treatment.

[0026] In addition, by generating and accumulating progress data including multi-axial vector data and changes in the patient's condition after treatment for each treatment cycle, it has the effect of managing the treatment progress over multiple cycles in a time-series manner and continuously tracking changes in the patient's condition.

[0027] In addition, by comparing and analyzing progress data by treatment cycle to assess changes in symptoms, treatment effects, and responses to therapeutic stimuli, it has the effect of objectively identifying the patient's treatment response trends and generating feedback data for subsequent treatment.

[0028] Furthermore, by calculating weights for each symptom axis that reflect the clinical importance and pathological association of clinical significance items and applying them to multi-axis vector data, it has the effect of enabling an integrated analysis of the patient's progress status centered on information meaningful for clinical judgment, going beyond simple record storage. Brief explanation of the drawing

[0029] Figure 1 is a configuration diagram of an unstructured Korean medicine clinical data progress management system based on semantic tagging and multi-axis vector mapping according to an embodiment of the present invention. FIG. 2 is a diagram illustrating the analysis unit of an unstructured Korean medicine clinical data progress management system based on semantic tagging and multi-axis vector mapping according to an embodiment of the present invention. FIG. 3 is a diagram illustrating the tagging section of an unstructured Korean medicine clinical data progress management system based on semantic tagging and multi-axis vector mapping according to an embodiment of the present invention. FIG. 4 is a diagram illustrating the vectorization section of an unstructured Korean medicine clinical data progress management system based on semantic tagging and multi-axis vector mapping according to an embodiment of the present invention. FIG. 5 is a diagram illustrating the progress analysis section of an unstructured Korean medicine clinical data progress management system based on semantic tagging and multi-axis vector mapping according to an embodiment of the present invention. Specific details for implementing the invention

[0030] Specific details regarding the problem to be solved, the means for solving the problem, and the effects of the invention as described above are included in the embodiments and drawings to be described below. The advantages and features of the present invention, and the methods for achieving them, will become clear by referring to the embodiments described below in detail together with the accompanying drawings.

[0032] Hereinafter, the present invention will be described in more detail with reference to the attached drawings.

[0033] FIG. 1 is a configuration diagram of an unstructured Korean medicine clinical data progress management system based on semantic tagging and multi-axis vector mapping according to an embodiment of the present invention, FIG. 2 is a diagram explaining the analysis unit of an unstructured Korean medicine clinical data progress management system based on semantic tagging and multi-axis vector mapping according to an embodiment of the present invention, FIG. 3 is a diagram explaining the tagging unit of an unstructured Korean medicine clinical data progress management system based on semantic tagging and multi-axis vector mapping according to an embodiment of the present invention, FIG. 4 is a diagram explaining the vectorization unit of an unstructured Korean medicine clinical data progress management system based on semantic tagging and multi-axis vector mapping according to an embodiment of the present invention, and FIG. 5 is a diagram explaining the progress analysis unit of an unstructured Korean medicine clinical data progress management system based on semantic tagging and multi-axis vector mapping according to an embodiment of the present invention.

[0035] Referring to FIG. 1, the semantic tagging and multi-axis vector mapping-based unstructured Korean medicine clinical data progress management system (100) of the present invention may include an input unit (110), an analysis unit (120), a tagging unit (130), a vectorization unit (140), a progress record unit (150), and a progress analysis unit (160).

[0036] The input unit (110) receives unstructured clinical data including at least one of a patient's symptoms, medical history, and treatment history; the analysis unit (120) processes the unstructured clinical data into at least one clinical semantic item; the tagging unit (130) generates tagging data by semantically tagging the clinical semantic item to correspond to a preset symptom axis; the vectorization unit (140) converts the tagging data generated by the tagging unit (130) into multi-axis vector data including extracted values ​​and weights for each symptom axis; the progress record unit (150) stores the multi-axis vector data converted by the vectorization unit (140) by treatment stage and generates progress data by treatment stage including changes in the patient's condition after treatment; and the progress analysis unit (160) can generate feedback data regarding at least one of changes in symptoms, treatment effects, and reactions to treatment stimuli by comparing and analyzing the progress data by treatment stage generated by the progress record unit (150).

[0037] More specifically, the input unit (110) can receive unstructured clinical data including a patient's natural language statement and a doctor's medical text input from a designated user terminal.

[0038] Here, the aforementioned atypical clinical data may include the chief complaint, major symptoms, primary symptoms, and other symptoms directly described by the patient, and may also include subjective judgment information additionally recorded by the Korean medicine practitioner during the process of questioning, observation, palpation, or treatment.

[0039] For example, the input unit (110) may be configured to set the patient's address symptoms, important symptoms, major symptoms, and other symptoms as checkpoints through a progress record UI provided to the user terminal, and to receive the response before and after acupuncture stimulation, whether to apply tonification or sedation methods, and the amount of stimulation as real-time numerical or selectable data.

[0040] In addition, the input unit (110) can receive additional clinical information such as the patient's gender, age, body type, emotions, appetite, sleep state, bowel and bladder state, abdominal examination results, tongue examination results, pulse examination results, and past treatment history, thereby enabling more precise clinical meaning analysis in subsequent steps.

[0042] For example, if a patient states to medical staff, "My neck has been stiff and feels like it is being stabbed with a knife for the past three days, and I also feel a little dizzy, and it gets a little better when I rest," the input unit (110) can receive this free-form sentence in its original form.

[0043] In addition, the input unit (110) can separately receive additional auxiliary input values ​​such as VAS 7, persistence, posterior neck, stinging pain, accompanying dizziness, and relief during rest, along with patient statements in the form of free-form sentences.

[0044] In this case, the input unit (110) receives free-form descriptive data and structured auxiliary data in a merged form, and can secure clinical information of various expression methods for the same patient without omission.

[0046] The above analysis unit (120) can process the unstructured clinical data of Korean medicine received through the above input unit (110) into natural language and analyze it into one or more clinical semantic items.

[0047] The above natural language processing may include at least one of morphological analysis, syntactic analysis, named entity recognition, semantic similarity analysis, matching with a pre-established clinical term dictionary, complex symptom separation logic, or synonym normalization logic.

[0048] In particular, as illustrated in FIG. 2, a hierarchical input structure including major, medium, and minor categories is shown, so the analysis unit (120) can hierarchically derive clinical semantic items by first classifying the patient's free-form sentence into symptom groups at the major category level, and then subdividing them into medium and minor categories.

[0049] For example, under the major category of "Digestive System," a medium category of "Digestion / Abdomen" can be established, and further subcategories such as postprandial fullness, epigastric distension, nausea, belching, dysphagia, abdominal pain, and diarrhea can be derived. By enabling the analysis of clinical semantic items with such a hierarchical structure, the semantic scope of the same expression can be subdivided stepwise.

[0051] In addition, the analysis unit (120) can also extract time information, intensity information, duration information, triggering event information, and progress information included in the patient's expression.

[0052] For example, the expression "started 3 days ago, suddenly, continuously, worsens with movement, and improves with rest" can be broken down into semantic items corresponding to the time of onset, pattern of occurrence, frequency or persistence, aggravating factors, and alleviating factors, respectively.

[0053] Accordingly, multiple clinical semantic items can be extracted simultaneously from a single short sentence, and in the subsequent tagging step, each item can be processed to correspond to a different symptom axis.

[0055] The above tagging unit (130) can generate tagging data by semantically tagging the clinical meaning item analyzed by the above analysis unit (120) so that it corresponds to a preset symptom axis.

[0056] At this time, the aforementioned pre-established symptom axis may include location, character, degree, factors causing and alleviating symptoms, symptoms, and patient history, and, if necessary, information regarding onset and progression may also be reflected.

[0057] More specifically, the above-mentioned area may include meridians, skeletal muscle locations, anatomical sites, or three-dimensional location information, and the above-mentioned aspect may include characteristics, duration, patterns, and radiation of pain or discomfort.

[0058] In addition, the above intensity and frequency may include the intensity of pain, the degree of perception, and the frequency of occurrence, and the above triggering factors and alleviating factors may include aggravating factors and alleviating factors according to specific posture, movement, time of day, external stimulation, rest, or thermal stimulation.

[0059] The above-mentioned accompanying symptoms may include associated symptoms and concurrent symptoms, and the above-mentioned patient history may include social history, occupational history, family history, genetic background, diagnostic history, treatment history, medication history, and food intake history.

[0060] Accordingly, the tagging unit (130) can convert unstructured Korean medical clinical data entered in a free-form manner into structured tagging data by classifying it into multiple symptom axes according to clinical meaning.

[0062] For example, if a patient states that the back of their neck hurts, that description can be tagged as Location, and if they describe it as feeling like being stabbed with a knife, that description can be tagged as Character.

[0063] In addition, expressions such as VAS 7 can be tagged as Degree, and expressions of persistent pain can be reflected as information corresponding to frequency or Progress.

[0064] Additionally, the expression that it worsens with movement can be tagged as a trigger factor, the expression that it improves with rest as a relief factor, and expressions such as dizziness or nausea as associated symptoms.

[0065] Furthermore, information that a patient performs computer work for long hours as an office worker can be tagged as occupational history in the patient history, and if there is a history of treatment or diagnosis of the same body part in the past, it can be tagged as diagnosis history or treatment history in the patient history.

[0066] In this way, the tagging unit (130) can standardize free-form clinical data into a multidimensional semantic structure by tagging multiple semantic units included in a single sentence by corresponding them to different symptom axes.

[0068] Meanwhile, the vectorization unit (140) can convert the tagging data generated by the tagging unit (130) into multi-axis vector data including symptom axis extraction values ​​and weights.

[0069] In this case, the multi-axial vector data may include extracted values ​​for each symptom axis corresponding to Location, Character, Intensity, Frequency, Precipitating Factors, Alleviating Factors, Associated Symptoms, and Patient History, and each extracted value may be assigned a weight that reflects the importance, pathological relevance, or contribution to treatment judgment of the corresponding clinical significance item.

[0070] Accordingly, the vectorization unit (140) can convert unstructured clinical data of Korean medicine, entered in a free-form style, into a comparable numerical or semi-quantitative data structure rather than a simple text record.

[0071] More specifically, the vectorization unit (140) can identify clinical semantic items corresponding to each symptom axis and convert the extracted value corresponding to the item into a preset form of a code value, category value, numeric value, or combination value.

[0072] For example, information on body parts such as the posterior neck, lumbar region, shoulder region, epigastric region, or right lower abdomen can be converted into extracted values ​​of the region axis, and expressions such as stinging pain, dull pain, numbness, heaviness, radiating pain, or pulling can be converted into extracted values ​​of the modality axis.

[0073] In addition, numerical information regarding pain intensity can be converted into extracted values ​​on the intensity axis, and expressions such as continuous, intermittent, severe in the morning, or worsening at night can be reflected as values ​​representing frequency or temporal variation characteristics.

[0074] Expressions such as worsening upon movement, worsening after meals, or worsening upon cold stimulation can be reflected as extracted values ​​for triggering factors, while expressions such as relief during rest, relief upon heat stimulation, or reduction in specific postures can be reflected as extracted values ​​for alleviating factors.

[0075] Expressions such as dizziness, nausea, sleep disorders, fatigue, or indigestion can be reflected as extracted values ​​on the associated symptoms axis, and information regarding occupational history, family history, past diagnosis history, treatment history, and medication history can be reflected as extracted values ​​on the patient history axis.

[0077] Additionally, the vectorization unit (140) may assign weights to the extracted values ​​for each symptom axis. The weights may be calculated by reflecting the clinical importance, pathological association, presence of the main symptom, presence of the important symptom, frequency of recurrence, or contribution to the judgment of the course of treatment of the corresponding clinical significance item.

[0078] For example, a relatively high weight may be assigned to items corresponding to the primary complaint of the patient, and a high weight may also be assigned to pathology-specific expressions that directly influence the decision on the direction of treatment.

[0079] On the other hand, low weights may be assigned to items with low clarity of information or relatively low clinical significance. According to the examples presented in the invention data, the site may be reflected at a level of 1.0, the pattern at 0.9, the intensity at 0.7, the frequency at 0.8, the triggering factor at 0.1, the alleviating factor at 0.5, the associated symptoms at 0.6, and the history or time-related information at 0.9 to 1.0. This is merely an example and can be set differently depending on the patient's condition, type of disease, and purpose of analysis.

[0081] For example, if a patient states that the back of the neck has been stiff and stabbing with a knife for 3 days, accompanied by dizziness, and that it is slightly relieved when resting, the vectorization unit (140) can correspond extraction values ​​such as posterior neck on the region axis, stabbing pain on the pattern axis, VAS 7 on the intensity axis, persistence on the frequency axis, unknown or none on the triggering factor axis, rest on the relieving factor axis, vertigo on the accompanying symptom axis, and 3 days after onset on the patient history axis.

[0082] And the above vectorization unit (140) can be configured into multi-axis vector data in the form of, for example, location=Cervical_Post, weight=1.0, character=Stabbing, weight=0.9, degree=VAS7, weight=0.7, frequency=Continuous, weight=0.8, trigger=null, weight=0.1, relief=Rest, weight=0.5, associated_sx=Dizziness, weight=0.6, history_days=3, weight=0.9 by combining weights corresponding to the extracted values ​​for each axis.

[0083] Multi-axis vector data configured in this way can be converted and stored in a tabular format, JSON format, or vector database storage format.

[0085] As another example, the vectorization unit (140) can not only list the patient's symptoms as individual axis values ​​but also normalize them into mutually comparable vector profiles.

[0086] For example, by generating multi-axial vector data at the time of initial examination and multi-axial vector data after second treatment for the same patient with the same structure, it is possible to quantitatively or semi-quantitatively compare which axis decreased, which axis was maintained, and which axis new adverse reactions occurred between each treatment cycle.

[0087] Accordingly, the vectorization unit (140) can provide base data for calculating the vector change amount for each treatment stage, deriving a time-series topographic map change pattern, or generating a unique progress profile for each patient in a subsequent step.

[0089] As another example, the vectorization unit (140) may include a weight calculation unit and a vector generation unit. The weight calculation unit may calculate weights for each symptom axis by reflecting at least one of the clinical importance and pathological association of the clinical semantic item included in the tagging data, and the vector generation unit may generate the multi-axis vector data based on the weights for each symptom axis calculated by the weight calculation unit.

[0090] That is, the vectorization unit (140) can adjust the vector structure so that clinically more important items are reflected relatively more significantly even with the same text input, and accordingly, can provide a data structure that is more suitable for analyzing treatment response and determining progress.

[0092] As another example, the vectorization unit (140) may also provide indexing information to enable the multi-axis vector data to be sorted by patient, disease group, treatment stage, or treatment stimulus type.

[0093] Accordingly, subsequent analyses may be possible, such as directly comparing the results of first-line and second-line treatments within the same patient, or comparing the response patterns of multiple patient groups to which the same treatment stimulus was applied.

[0094] Additionally, the vectorization unit (140) can convert the multi-axis vector data into an input format for generating a three-dimensional curved terrain map, in which case the extracted values ​​and weights for each axis can be used as height, distribution density, or feature quantities for pattern recognition.

[0096] The above progress record unit (150) can store the multi-axis vector data converted by the above vectorization unit (140) by treatment stage and generate treatment stage progress data including changes in the patient's condition after treatment.

[0097] At this time, the above-mentioned progress record (150) can store multi-axis vector data collected at each treatment point from the initial consultation to the nth treatment for the same patient in chronological order, and can manage treatment stimulation information, pre-treatment state information, post-treatment state information, and change amount information corresponding to each treatment stage together.

[0098] Accordingly, the above progress record (150) can go beyond simply storing accumulated medical notes and generate structured progress data showing how the patient's symptom state changed according to the treatment process.

[0100] More specifically, the above progress record (150) is linked with a digital-based progress record UI, and medical staff can receive and store information regarding the response before stimulation, response after stimulation, whether tonification or sedation is applied, amount of stimulation, treatment area, applied acupoints, and time of treatment for each treatment stage, while the patient's chief symptom, important symptom, major symptom, and other symptoms are set as checkpoints.

[0101] In addition, the above progress record (150) can record the response to a specific acupoint within the same patient, the difference in effect by treatment day, the response curve according to the change in stimulation amount, and the response trend of repeated treatment for the same symptom.

[0102] In this way, the above-mentioned progress record section (150) can enable quantitative comparison between stages in the above-mentioned progress analysis section (160) by reconstructing the treatment flow for the same symptom group in units of treatment stages.

[0104] For example, if the patient complains of posterior neck pain, dizziness, and sleep disorders at the time of the initial examination, the above progress record (150) can store multi-axis vector data at the time of the initial examination as first treatment stage data.

[0105] Subsequently, when a specific combination of acupoints and a preset amount of stimulation are applied in the first treatment, the progress recorder (150) can record changes in pain intensity immediately after treatment, whether dizziness is reduced, whether sleep is improved, etc., along with the stimulation information, and store them as progress data after the first treatment.

[0106] Subsequently, when the same combination of acupoints is applied during the second treatment but the amount of stimulation is increased, the above progress record (150) can record the amount of symptom change along with the information on the increased amount of stimulation and store it as data for the second treatment stage.

[0107] In addition, if the acupoint combination is changed during the third treatment, the changed treatment stimulation information and the results of symptom changes can be stored together with separate treatment information.

[0108] Accordingly, the conditions after the initial examination, the first treatment, the second treatment, and the third treatment can be accumulated in a form that can be compared within the same data structure.

[0110] In addition, the above progress record unit (150) can generate progress data for each treatment stage by reflecting the difference between multi-axis vector data corresponding to each treatment stage.

[0111] For example, if after the first treatment the pain intensity is reduced but the frequency is maintained, after the second treatment the radiating pain is relieved but fatigue is increased, and after the third treatment sleep disorder is relieved but temporary dizziness remains, the above progress record (150) can generate progress data including the amount of change in the intensity axis, the amount of change in the pattern axis, the amount of change in the associated symptoms axis, and time-lapse information linked to the patient history.

[0112] In this case, since the above progress data may include change values ​​for each axis, direction of change, timing of change, and applied treatment stimulus information, it can be reconstructed into a change-centered structured treatment flow rather than a simple listing of symptoms.

[0114] As another example, the above progress record (150) can store data to enable comparison of responses according to multiple acupoint combinations or multiple treatment stimulation conditions within the same patient.

[0115] For example, the amount of vector change after treatment when the first acupoint combination and the second acupoint combination are applied to the same symptom can be stored separately, and accordingly, in a subsequent analysis step, it is possible to compare which acupoint combination shows a greater improvement effect.

[0116] In addition, when the stimulus amount is varied in stages, the magnitude of symptom reduction, duration, and presence or absence of adverse reactions under each stimulus amount condition can be stored in conjunction, so this can be utilized as basic data to derive response curves for each stimulus amount.

[0118] In addition, the above progress record section (150) can store progress data for each treatment stage in conjunction with a data structure for visualization.

[0119] For example, by converting multi-axis vector data generated at each treatment cycle into a 3D Surface Plot generation format and saving them together, the symptom distribution at the initial visit and the symptom distribution after the nth treatment cycle can be visualized as topographic maps at different points in time.

[0120] Accordingly, it is possible to visually track how much the patient's condition has been flattened along a specific axis, how much the elevation pattern of a specific symptom group has decreased after treatment, or whether specific side effects have newly emerged.

[0121] For example, the symptom distribution at the time of initial examination and the symptom distribution after nth treatment are distinguished and shown, and the topographic pattern can be updated according to the progression of treatment, and the above progress record section (150) can be configured to enable updating of the time-series topographic map for each patient by accumulating and reflecting vector data for each stage.

[0123] As another example, the above progress record (150) can reclassify and store progress data by treatment stage by patient, disease group, treatment method, or treatment stimulus.

[0124] For example, progress data of multiple patients belonging to the same disease group can be reorganized into a single disease group dataset, response patterns of patient groups to which the same therapeutic stimulus was applied can be extracted separately, and progress data corresponding to specific acupoint combinations or specific stimulus amount ranges can be managed as separate groups.

[0125] Accordingly, the above progress record (150) can be used not only for tracking the progress of individual patients but also as a data storage structure for analyzing response patterns at the group level or building a subsequent learning dataset.

[0127] As another example, the above progress record unit (150) can store standardized multi-axis vector data through the above 8 symptom axes and digital progress record information as the ninth axis in conjunction.

[0128] Here, the above 8 symptom axes can function as reference information that multidimensionally expresses the patient's clinical condition, and the above 9th axis can function as reference information that cumulatively records the amount of change by treatment cycle, the response to treatment stimulation, and the time-series trend of the condition.

[0129] Accordingly, the above progress record unit (150) may not only store clinical classification results at a single point in time, but may also implement a dynamic progress record structure that continuously tracks and updates changes in the patient's condition from the initial examination to the nth treatment.

[0131] As another example, the above progress record (150) can provide accumulated progress data by treatment stage as learning raw data for subsequent reinforcement learning, digital twin synchronization, or clinical decision support.

[0132] That is, the data accumulated in the above progress record (150) may be used as a basic dataset for future automatic updates of algorithm weights, prediction of treatment effects, stimulation response simulation, or recommendation of treatment strategies.

[0134] The above progress analysis unit (160) can generate feedback data for at least one of symptom change, treatment effect, and response to treatment stimulus by comparing and analyzing the progress data for each treatment stage generated in the above progress record unit (150).

[0135] At this time, the above-mentioned progress analysis unit (160) can determine whether the patient's condition improves, worsens, fluctuates temporarily, or is maintained at a certain level by comparing the amount of change, direction of change, and duration of change of multi-axis vector data for each treatment stage with multiple treatment stage data for the same patient arranged in a time series.

[0136] Accordingly, the above-mentioned progress analysis unit (160) is not limited to simply interpreting the state of individual stages, but can structurally derive the reaction trend that appears throughout the entire treatment process.

[0138] More specifically, the above-mentioned progress analysis unit (160) compares the changes in the axis of the area, pattern, intensity, frequency, causative factor, alleviating factor, accompanying symptom, and patient history included in the progress data for each treatment stage corresponding to each treatment stage, and determines that if a specific symptom axis decreases by more than a preset standard, it is judged as improvement, if it increases by more than a preset standard, it is judged as deterioration, and if it is maintained similarly within a predetermined range, it is judged as a continuous or maintained state.

[0139] For example, if both pain intensity and frequency decrease over consecutive treatment cycles, the treatment effect can be judged as positive; if the location remains the same but the pattern changes from stabbing pain to dull pain, it can be judged as partial improvement; and if associated symptoms increase or new adverse reactions appear, it can be judged as caution or a negative change.

[0140] In this way, the above-mentioned progress analysis unit (160) can determine the patient's progress more precisely than a simple numerical comparison by comprehensively interpreting the simultaneous change of multiple axes.

[0142] For example, if a patient initially showed posterior neck pain, stinging pain, VAS 7, persistent pain, and dizziness, and after the first treatment the VAS decreased to 5 with some dizziness remaining, and after the second treatment the VAS decreased to 3 with the stinging pain being relieved to dull pain, and after the third treatment the pain frequency changed to intermittent, the above-mentioned progress analysis unit (160) can determine that there is partial improvement after the first treatment, clear improvement after the second treatment, and entry into a stabilization phase after the third treatment.

[0143] Conversely, if pain intensity decreases after a specific treatment cycle but associated symptoms such as nausea, fatigue, or sleep disorders newly increase, the overall course is not judged as simple improvement; instead, a complex assessment result may be generated that includes the possibility of additional responses to the treatment stimulus or side effects.

[0144] Accordingly, the above-mentioned progress analysis unit (160) can evaluate the overall progress status in an integrated manner while interpreting the treatment effect and adverse reaction separately.

[0146] In addition, the above-mentioned progress analysis unit (160) can determine the patient's response tendency to a specific treatment stimulus or a specific acupoint combination by linking the treatment stimulus information and symptom change information included in the progress data for each treatment stage.

[0147] For example, when the first acupoint combination is applied to the same patient, the pain intensity decreases rapidly but persistence remains, and when the second acupoint combination is applied, the accompanying symptoms improve significantly even though the reduction in pain intensity is small, the above-mentioned progress analysis unit (160) can determine that each treatment stimulus has different effect characteristics.

[0148] In addition, even if the same combination of acupoints is applied, the response curve may appear differently depending on the change in the amount of stimulation. Therefore, the above-mentioned progress analysis unit (160) can compare the response curves for each amount of stimulation to determine whether the most stable improvement trend appears at a specific range of stimulation amounts.

[0150] As another example, as illustrated in FIG. 4, the vectorization unit (140) may include a weight calculation unit (141) that calculates symptom axis weights by reflecting at least one of the clinical importance and pathological association of the clinical semantic item included in the tagging data, and a vector generation unit (142) that generates the multi-axis vector data based on the symptom axis weights calculated by the weight calculation unit (141).

[0151] Through this, the aforementioned multi-axial vector data can be generated in a form that goes beyond a simple listing of items and reflects clinically more important semantic elements relatively more significantly.

[0153] As another example, referring to FIG. 5, the above-mentioned progress analysis unit (160) may include a treatment response judgment unit (161) and a feedback generation unit (162).

[0154] The above treatment response judgment unit (161) can generate judgment result data by comparing and analyzing the progress data for each treatment stage corresponding to the treatment stage in a time series to determine at least one of the degree of symptom improvement, degree of deterioration, persistence, and treatment response trend.

[0155] The above judgment result data may include, for example, the amount of change by axis, the direction of change, the duration of change, response sensitivity, and response stability to repeated stimuli.

[0156] And the feedback generation unit (162) can calculate an integrated judgment value representing the patient's progress status based on the judgment result data generated by the treatment response judgment unit, and generate feedback data based on the integrated judgment value.

[0157] Accordingly, the above-mentioned progress analysis unit (160) can aggregate individual judgment results for each stage into one integrated state information.

[0159] The above integrated judgment value can be calculated by comprehensively reflecting the amount of change in each symptom axis, the weight of each symptom axis, the presence of the main symptom, the presence of a significant symptom, responsiveness to therapeutic stimuli, and the degree of increase or decrease in associated symptoms.

[0160] For example, a high improvement score may be calculated if both the intensity and frequency of the main symptoms decrease, the pattern changes in a direction of alleviation, and associated symptoms decrease.

[0161] On the other hand, if the main symptom is partially alleviated but associated symptoms increase, or if dizziness, nausea, or fatigue increases after a specific stimulus, an intermediate judgment value indicating a state of attention may be produced.

[0162] In addition, if a worsening response is repeatedly observed to the same stimulus after a certain number of cycles, a negative judgment value may be calculated.

[0163] As such, the aforementioned integrated judgment value does not merely reflect a change on a single axis, but can function as a comprehensive indicator representing the patient's overall course.

[0165] In addition, the feedback generation unit (162) can generate feedback data provided to medical staff based on the integrated judgment value.

[0166] The above feedback data may be provided in at least one of the following forms: analysis results in text form, numeric status values, graphs, or visual warning displays.

[0167] For example, the above feedback data may include analytical phrases such as recommending the maintenance of the current treatment direction, the need to adjust the stimulation amount, the need to review changes in acupoint combinations, or concerns about side effects.

[0169] As another example, the above-mentioned progress analysis unit (160) can perform a visualization-based change judgment by comparing the symptom distribution at the time of the initial visit and the symptom distribution after the nth treatment.

[0170] For example, after converting the multi-axis vector data of each treatment course accumulated and stored in the progress record unit (150) into a time-series topographic map based on a 3D Surface Plot, the progress analysis unit (160) can determine whether the patient's condition has improved by comparing the change in height of the terrain, the change in distribution density, and the degree of uplift or flattening of a specific area.

[0171] In this case, if the elevation of an area where a specific group of symptoms is concentrated continuously decreases, it can be determined that the pathological burden is being alleviated, and if a new protruding area occurs, the possibility of a new adverse reaction or residual pathology can be assessed.

[0172] This visualization-based change judgment can be used as an auxiliary analysis means of the above-mentioned progress analysis unit (160).

[0174] As another example, the above-mentioned progress analysis unit (160) can cross-compare patient-specific progress data based on disease group or treatment method.

[0175] For example, the relative level of response can be determined by comparing the course patterns of multiple patients belonging to the same disease group with the current course pattern of a specific patient, and the presence of an exceptional response can be determined by comparing the response trend of a specific patient with the average response trend observed in a group of patients who received the same therapeutic stimulus.

[0176] Accordingly, the above-mentioned progress analysis unit (160) can perform not only individual patient-centered progress judgment but also correction judgment through comparison with collective standards.

[0177] This cross-comparison function can be utilized as a foundation for subsequent clinical decision support or the expansion of self-evolving algorithms.

[0179] As another example, the above-mentioned progress analysis unit (160) can output the results of analyzing the progress data for each treatment stage in a form that can be linked with a digital twin, a reinforcement learning engine, or a clinical decision support system.

[0180] For example, the above feedback data can be converted into the form of label data, reward value data, or state transition data that a subsequent algorithm can learn.

[0181] Accordingly, the progress analysis unit (160) of the present invention may function as a core interface that transmits real-time progress judgment results to a subsequent self-evolving system, going beyond a simple result display module.

[0183] Meanwhile, the above-mentioned progress analysis unit (160) can calculate an integrated progress judgment value indicating the patient's progress status in order to quantitatively compare progress data by treatment stage.

[0184] For example, the above integrated progress judgment value G t It can be defined as shown in [Mathematical Formula 1] below.

[0186] [Mathematical Formula 1]

[0187]

[0189] Here, G t is the integrated course judgment value, N is the number of symptom axes, and w i represents a weight indicating the clinical importance of the i-th symptom axis, and ΔS i,t represents the amount of change in symptoms at the treatment cycle, and ΔR i,t represents the change in response before and after treatment within the same treatment cycle, and ΔT i,t represents the change in time series trend over multiple treatment cycles, and α, β, and γ represent the weights of each term.

[0191] At this time, the above change in symptoms ΔS i,t is the vector value V of the i-th symptom axis at the t-1th treatment order. i,t-1 and the vector value V of the i-th symptom axis at the t-th treatment order i,t It can be calculated based on the difference. That is, if symptom intensity or burden decreases, ΔS i,t As it increases, it can be reflected in a direction of improvement.

[0192] In addition, the change in the above treatment response ΔR i,t is the i-th symptom axis value P after treatment of the t-th treatment order. i,t and the pre-treatment i-th symptom axis value B of the t-th treatment order i,tIt can be calculated based on the difference. That is, by examining the difference in response immediately before and after treatment within the same course, it is possible to reflect whether there was immediate improvement immediately after treatment on the same day, or whether there was an immediate response to the stimulus.

[0193] In addition, the change in the above time series trend ΔT i,t It can be calculated based on the following [Mathematical Formula 3].

[0195] [Mathematical Formula 3]

[0196]

[0197] Here, ΔT i,t represents the change in time series trend, k represents the number of previous treatment cycles being compared, and V i,t Is It refers to the vector value of the i-th symptom axis in the t-th treatment order, and V i,t-m can mean the i-th symptom axis vector value of the previous orders.

[0198] In other words, the change in time series trend ΔT i,t It can reflect whether it is a continuous improvement or a temporary fluctuation, rather than a change in vocalization.

[0199] For example, it is possible to distinguish between cases where a patient improves only once and then worsens again, and conversely, cases where they gradually improve three consecutive times.

[0201] Accordingly, the above-mentioned progress analysis unit (160) can calculate an integrated judgment value that reflects not only changes in a single symptom axis but also changes in multiple symptom axes, the response to treatment stimulation, and persistence over time.

[0203] In addition, the above-mentioned progress analysis unit (160) has an integrated progress judgment value G normalized by a weighted sum as shown in [Equation 2] below, in order to reduce distortion caused by differences in the number of active intermediate axes per patient. t norm It can produce.

[0205] [Mathematical Formula 2]

[0206]

[0208] In addition, the feedback generation unit (162) may determine the feedback stage based on the normalized integrated progress judgment value based on [Equation 1] and [Equation 2].

[0209] For example, if the normalized integrated progress judgment value is greater than or equal to a preset first reference value θ1, the feedback generating unit (162) determines the feedback step as 'recommended', and if the normalized integrated progress judgment value is less than the first reference value θ1 and greater than or equal to the second reference value θ2, the feedback generating unit (162) determines the feedback step as 'caution', and if the normalized integrated progress judgment value is less than the second reference value θ2, the feedback generating unit (162) may determine the feedback step as 'concern about side effects'.

[0210] More specifically, when the above feedback step is determined to be 'recommended', the feedback generating unit (162) may determine that the currently applied treatment stimulus, acupoint combination, stimulus amount, or treatment direction is inducing a positive response regarding the patient's progress condition.

[0211] For example, if the vector values ​​of the intensity and frequency axes corresponding to the address symptoms decrease in all consecutive treatment cycles, the accompanying symptoms are also alleviated, and the immediate response to the treatment stimulus and time-series persistence are all good, the feedback generating unit (162) can determine the feedback step as 'recommended'.

[0212] Accordingly, the feedback generation unit (162) can generate feedback data that induces maintaining the current treatment direction or applying the same treatment stimulation conditions first.

[0214] In addition, if the above feedback step is determined to be 'caution', the feedback generating unit (162) may determine that the patient's condition is generally stable or partially improving, but that deterioration or variability is observed in a specific symptom axis.

[0215] For example, if the intensity of the main symptom decreases but the change in frequency is minimal, or if the main pain is relieved but accompanying symptoms such as dizziness, fatigue, or sleep disorder newly increase, the feedback generating unit (162) may determine the feedback stage as 'caution'.

[0216] In addition, even if the initial response to the same therapeutic stimulus was good, if the persistence of the response deteriorates with repeated treatment or sensitivity to changes in stimulus amount appears unstable, it can be judged as a 'caution' stage.

[0217] Accordingly, the feedback generation unit (162) can generate feedback data including the need for adjustment of stimulation amount, partial change of acupoint combination, resetting of treatment interval, or enhancement of progress tracking.

[0219] In addition, if the above feedback step is determined to be 'concern about side effects,' the feedback generating unit (162) may determine that the current treatment stimulus or treatment direction is causing a negative reaction to the patient's progress or at least failing to induce stable improvement.

[0220] For example, if the intensity or frequency of the symptoms of the address increase in consecutive treatment cycles, and if adverse reactions such as dizziness, nausea, fatigue, insomnia, or pain spreading newly appear or intensify after the treatment stimulus, the feedback generating unit (162) may determine the feedback stage as a 'concern about side effects'.

[0221] In addition, if negative responses accumulate repeatedly under the same stimulus condition, or if the integrated progress judgment value is continuously maintained below the preset second threshold value θ2, it can be classified as a 'concern for side effects' state.

[0222] Accordingly, the feedback generating unit (162) can generate feedback data indicating the need to stop the current treatment stimulation, reduce the amount of stimulation, review alternative acupoints, or switch to another treatment strategy.

[0224] As another example, the feedback generation unit (162) may generate auxiliary feedback information indicating the basis for judgment together with the feedback step.

[0225] For example, the above auxiliary feedback information may include which symptom axis showed significant improvement, which symptom axis exacerbation occurred, how responsiveness to a specific therapeutic stimulus changed, or whether continuity of response was maintained across consecutive treatment cycles.

[0226] Accordingly, the feedback generation unit (162) can more effectively support the clinical judgment of medical staff by providing not only the step but also the cause and basis for the step.

[0228] <Example 2>

[0229] Meanwhile, the method for managing the progress of unstructured Korean medical clinical data based on semantic tagging and multi-axis vector mapping of the unstructured Korean medical clinical data progress management system (100) according to one embodiment of the present invention may include an input step, an analysis step, a tagging step, a vectorization step, a progress recording step, and a progress analysis step.

[0230] In the input step, the input unit (110) may receive unstructured clinical data including at least one of the patient's symptoms, medical history, and treatment history; in the analysis step, the analysis unit (120) may process the unstructured clinical data into at least one clinical semantic item; and in the tagging step, the tagging unit (130) may generate tagging data by semantically tagging the clinical semantic item to correspond to a preset symptom axis.

[0231] Additionally, in the vectorization step, the vectorization unit (140) may convert the tagging data into multi-axis vector data including extracted values ​​and weights for each symptom axis, and in the progress recording step, the progress recording unit (150) may store the multi-axis vector data by treatment stage and generate progress data by treatment stage including changes in the patient's condition after treatment, and in the progress analysis step, the progress analysis unit (160) may compare and analyze the progress data by treatment stage to generate feedback data for at least one of changes in symptoms, treatment effects, and responses to treatment stimuli.

[0233] More specifically, in the above input step, the input unit (110) can receive unstructured Korean medical clinical data including a patient's natural language statement, a Korean medical doctor's medical text input, a medical history response, a record of reactions before and after treatment, and a past medical history.

[0234] At this time, the above-mentioned atypical Korean medicine clinical data can be entered by classifying it into chief symptom, important symptom, major symptom, and other symptom, and additional information such as the treatment time, treatment cycle, combination of applied acupoints, stimulation amount, and whether tonification or sedation methods were applied can also be entered.

[0235] Accordingly, input data can be structured to enable time-series analysis of the same patient.

[0237] In the above analysis step, the analysis unit (120) can process the unstructured clinical data of Korean medicine received in the above input step into natural language and analyze it into one or more clinical semantic items.

[0238] At this time, the above natural language processing may include at least one of morphological analysis, syntactic analysis, semantic decomposition, named entity recognition, and matching with a clinical term dictionary, and may separate and derive symptom expressions, site expressions, intensity expressions, frequency expressions, expressions related to trigger or alleviation, expressions related to concomitant symptoms, and expressions related to patient history included in the patient's free-form sentences.

[0239] In addition, in the above analysis step, the atypical Korean medicine clinical data can be analyzed into clinical semantic items having a hierarchical structure including major, medium, and minor categories, so multiple semantic elements included in the same sentence can be arranged in a structure suitable for tagging.

[0241] In the above tagging step, the tagging unit (130) can generate tagging data by semantically tagging the clinical meaning items derived in the above analysis step so that they correspond to a preset symptom axis. At this time, the symptom axis may include location, character, intensity, frequency, triggering factors, alleviating factors, associated symptoms, and patient history.

[0242] For example, posterior neck pain can be tagged by location, stinging or tingling can be tagged by pattern, VAS 7 can be tagged by intensity, and expressions such as continuous or intermittent can be tagged by frequency.

[0243] Additionally, exacerbation upon movement can be tagged as a trigger, relief upon rest can be tagged as a relief factor, dizziness, nausea, or sleep disturbances can be tagged as concomitant symptoms, and past diagnosis, treatment, medication, or occupational history can be tagged as patient history.

[0244] Accordingly, in the above tagging step, free-form clinical data can be converted into structured tagging data corresponding to multiple semantic axes.

[0246] In the above vectorization step, the vectorization unit (140) can convert the tagging data generated in the above tagging step into multi-axis vector data including symptom axis extraction values ​​and weights.

[0247] More specifically, clinical semantic items corresponding to each symptom axis can be converted into pre-set numeric values, categorical values, code values, or combination values, and each item can be assigned a weight reflecting clinical importance, pathological association, presence of the main symptom, or contribution to treatment judgment.

[0248] For example, high weights may be assigned to axis-specific items corresponding to the chief complaint symptoms, while low weights may be assigned to items with low clinical significance or unclear significance.

[0249] Accordingly, in the above vectorization step, unstructured clinical data can be converted into structured multi-axial vector data comparable by treatment cycle.

[0251] In addition, in the above progress recording step, the above progress recording unit (150) can store multi-axis vector data generated in the above vectorization step according to the treatment stage and generate progress data according to the treatment stage including changes in the patient's condition after treatment.

[0252] At this time, in the aforementioned progress recording stage, multi-axis vector data corresponding to each treatment point from the initial consultation to the nth treatment can be linked and stored in chronological order, and the pre-treatment state, post-treatment state, applied treatment stimulus, stimulus amount, acupoint combination, and amount of change between treatment sessions for the same patient can be recorded together.

[0253] In addition, in the above progress recording stage, data for comparing the effects of each acupoint within the same patient, comparing the effects of each treatment day, recording response curves by stimulation amount, and cumulatively reflecting a 3D surface plot can be stored together.

[0254] Accordingly, in the aforementioned progress recording step, structured time-series progress data according to the treatment flow can be generated beyond simple record preservation.

[0256] In addition, in the above progress analysis step, the progress analysis unit (160) can compare and analyze the progress data for each treatment stage generated in the above progress recording step to generate feedback data for at least one of symptom change, treatment effect, and response to treatment stimulus.

[0257] For example, in the above progress analysis step, the degree of symptom improvement, degree of deterioration, persistence, and treatment response trends can be determined by comparing the amount of change between multi-axial vector data of each treatment course in a time series.

[0258] In addition, in the above progress analysis step, an integrated progress judgment value can be calculated by reflecting the amount of change and weight for each axis, and based on the integrated progress judgment value, a feedback step corresponding to recommendation, caution, or concern for side effects can be determined.

[0259] Accordingly, medical staff can intuitively and quantitatively check the patient's treatment progress and receive judgment information necessary for setting the direction of subsequent treatment.

[0261] Meanwhile, regarding the control method of the unstructured Korean medicine clinical data progress management system based on semantic tagging and multi-axis vector mapping according to the present invention, the living organism can be understood as a non-linear dynamic system in which extremely complex organic interactions occur continuously. However, conventional methods of interpreting the living organism often model the structure based on the state observed at a specific point in time, or interpret examination results at a single point in time or pre- and post-treatment results in a segmental manner. While such static modeling methods may be useful for understanding the state at a specific point in time, they have limitations in that they fail to sufficiently reflect the dynamic characteristics of biological responses that change continuously over time.

[0262] In particular, a patient's condition does not simply change discretely between the time points before and after treatment, but can exhibit continuous and cumulative changes over multiple treatment cycles. Nevertheless, conventionally, only the onset and end states of the disease were often recorded in a fragmentary manner, or the results of individual treatment cycles were treated as independent information, making it difficult to precisely track the trajectory of changes formed throughout the entire treatment process. Consequently, the complex system characteristics and time-series continuity of actual biological responses were not sufficiently reflected, leading to limitations in that clinical decision-making also relied on static judgment structures.

[0263] A control method for an unstructured Korean medicine clinical data progress management system based on semantic tagging and multi-axis vector mapping according to an embodiment of the present invention can be configured to improve these limitations by constructing the n-th treatment progress as time-series data and accumulating treatment result feedback data collected at each treatment stage as digital progress record data, thereby enabling real-time tracking of the continuous trajectory of changes in the patient's condition. That is, the present invention does not simply record the condition at a specific point in time individually, but rather continuously reconstructs the bio-response that changes throughout the entire treatment process on a time axis and reflects this in a digital twin model and a reinforcement learning structure, thereby enabling more precise clinical decision support.

[0264] Furthermore, the control method of an unstructured Korean medicine clinical data progression management system based on semantic tagging and multi-axis vector mapping according to an embodiment of the present invention goes beyond simple diagnostic assistance; by reconstructing clinical progression data and state change data from a dynamic perspective, it enables structural determination of the direction in which a patient's state change is progressing, which of multiple prescription pathways is more effective, and the likelihood of future prognosis unfolding in a particular manner. Accordingly, the present invention provides a clinical decision support structure that reflects the time-series changes and dynamic characteristics of biological responses, thereby enabling more precise Korean medicine clinical navigation and personalized treatment support.

[0266] As described above, according to the present invention, a system and method for managing the progress of unstructured Korean medical clinical data based on semantic tagging and multi-axis vector mapping can be provided, which can standardize Korean medical clinical information recorded in a free-form narrative into a structured form by processing unstructured Korean medical clinical data including a patient's symptoms, medical history, and treatment history into natural language, analyzing it into clinical semantic items, and performing semantic tagging.

[0267] In addition, a system and method for managing the progress of unstructured Korean medical clinical data based on semantic tagging and multi-axis vector mapping can be provided, which can multidimensionally represent a patient's symptom state and quantitatively compare the state before and after treatment by converting clinical semantic items included in unstructured Korean medical clinical data into multi-axis vector data by corresponding them to multiple pre-established symptom axes.

[0268] In addition, a system and method for managing the progress of unstructured Korean medical clinical data based on semantic tagging and multi-axis vector mapping can be provided, which can manage the progress of treatment over multiple cycles in a time-series manner and continuously track changes in the patient's condition by generating and accumulating progress data including multi-axis vector data and changes in the patient's condition after treatment for each treatment cycle.

[0269] In addition, a semantic tagging and multi-axis vector mapping-based atypical Korean medicine clinical data progress management system and method can be provided, which can more objectively identify the patient's treatment response trends and generate feedback data for subsequent treatment by comparing and analyzing progress data by treatment stage to determine symptom changes, treatment effects, and responses to treatment stimuli.

[0270] In addition, by calculating weights reflecting the clinical importance and pathological association of clinical significance items for each symptom axis and applying them to multi-axis vector data, a system and method for managing the progress of unstructured Korean medicine clinical data based on semantic tagging and multi-axis vector mapping can be provided, which enables integrated analysis of the patient's progress status centered on information meaningful for clinical judgment, going beyond simple record storage.

[0272] In addition, a control method for an unstructured Korean medicine clinical data progress management system based on semantic tagging and multi-axis vector mapping according to an embodiment of the present invention may be recorded on a computer-readable medium containing program instructions for performing operations implemented by various computers. The computer-readable medium may include program instructions, data files, data structures, etc., either alone or in combination. The program instructions on the medium may be those specifically designed and configured for the present invention, or they may be those known and available to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes; optical recording media such as CD-ROMs and DVDs; magneto-optical media such as floptical disks; and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, and flash memory. Examples of program instructions include machine code, such as that generated by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc.

[0274] As described above, although an embodiment of the present invention has been explained by limited embodiments and drawings, the embodiment of the present invention is not limited to the embodiments described above, and various modifications and variations are possible from this description by those skilled in the art to which the present invention pertains. Accordingly, an embodiment of the present invention should be understood only by the claims described below, and all equivalent or analogous variations thereof shall be considered to be within the scope of the inventive concept. Explanation of the symbols

[0276] 110 : Input section 120 : Analysis Department 130 : Tagging section 140 : Vectorization part 141 : Weight calculation unit 142 : Vector generation section 150 : Progress Record Book 160 : Progress Analysis Section 161 : Treatment Response Judgment Section 162 : Feedback generation section

Claims

Claim 1 A semantic tagging and multi-axis vector mapping-based system for managing the progress of non-standard Korean medical data, comprising: an input unit for receiving non-standard Korean medical clinical data including at least one of a patient's symptoms, medical history, and treatment history; an analysis unit for analyzing the non-standard Korean medical clinical data into at least one clinical semantic item by natural language processing; a tagging unit for generating tagging data by semantically tagging the clinical semantic item to correspond to a preset symptom axis; a vectorization unit for converting the tagging data generated by the tagging unit into multi-axis vector data including extracted values ​​and weights for each symptom axis; a progress record unit for storing the multi-axis vector data converted by the vectorization unit by treatment stage and generating progress data by treatment stage including changes in the patient's condition after treatment; and a progress analysis unit for generating feedback data regarding at least one of symptom changes, treatment effects, and responses to treatment stimuli by comparing and analyzing the progress data by treatment stage generated by the progress record unit. Claim 2 A semantic tagging and multi-axis vector mapping-based atypical Korean medicine clinical data progression management system according to claim 1, wherein the aforementioned preset symptom axis includes location, pattern, intensity, frequency, triggering factor, alleviating factor, associated symptoms, and patient history. Claim 3 A semantic tagging and multi-axis vector mapping-based system for managing the progress of unstructured Korean medical clinical data according to claim 1, wherein the analysis unit analyzes the unstructured Korean medical clinical data into clinical semantic items having a hierarchical structure including major, medium, and minor classifications; the tagging unit generates tagging data by automatically matching the clinical semantic items to correspond to the symptom axis; the vectorization unit comprises: a weight calculation unit that calculates weights for each symptom axis by reflecting at least one of the clinical importance and pathological association of the clinical semantic items included in the tagging data; and a vector generation unit that generates multi-axis vector data based on the weights for each symptom axis calculated by the weight calculation unit. Claim 4 A semantic tagging and multi-axis vector mapping-based unstructured Korean medicine clinical data progress management system, wherein, in claim 1, the progress analysis unit comprises: a treatment response judgment unit that generates judgment result data by comparing and analyzing progress data for each treatment stage corresponding to the treatment stage in a time-series manner to determine at least one of symptom improvement, deterioration, persistence, and treatment response trend; and a feedback generation unit that calculates integrated judgment data indicating the patient's progress status based on the judgment result data generated by the treatment response judgment unit, and generates feedback data based on the integrated judgment data. Claim 5 In paragraph 4, the above-mentioned progress analysis unit comprises an integrated progress judgment value G representing the patient's progress status based on the following [Mathematical Formula 1]. t Calculate and [Mathematical Formula 1] (Here, G t is the integrated course judgment value, N is the number of symptom axes, and w i represents a weight indicating the clinical importance of the i-th symptom axis, and ΔS i,t represents the amount of change in symptoms at the treatment cycle, and ΔR i,t represents the change in response before and after treatment within the same treatment cycle, and ΔT i,t represents the amount of change in time-series trend over multiple treatment cycles, and α, β, and γ represent the weights of each term) The feedback generation unit above, the integrated progress judgment value G t Based on, determine the feedback stage, wherein the above integrated progress judgment value G t If is greater than or equal to the preset first reference value θ1, the feedback step is determined as 'recommended', and the integrated progress judgment value G t If α is smaller than the first reference value θ1 and greater than or equal to the second reference value θ2, the feedback step is judged as 'caution', and the integrated progress judgment value G t A semantic tagging and multi-axis vector mapping-based unstructured Korean medicine clinical data progress management system characterized by determining the feedback step as 'concern about side effects' when the above second reference value θ2 is smaller.

Images