System and method for providing pain analysis information using user free-text description and structured clinical data

Abstract

The present invention relates to a system and method for providing pain analysis information using user free descriptions and structured clinical data. The system receives free description data and structured clinical data from a user terminal, and a processor of an analysis server classifies the free description into multiple pattern categories, calculates KSI by multiplying VAS, pain diffusion score, and sensory hypersensitivity score, and calculates KLM by multiplying S, T, and A. Additionally, based on pattern and CDI data, one or more kernel structures and link structures are derived, and a structured analysis information object including a pattern identifier, a kernel identifier, and a link identifier is generated. In one embodiment, expressions such as "must do," "duty," and "responsibility" can be extracted through morphological analysis and matched with a responsibility burden pattern database. The present invention provides a means for providing reproducible pain analysis information that combines free descriptions and quantitative indicators.

Description

Technology Field

[0001] The present invention relates to information processing technology for processing user free descriptions and structured clinical data, and more specifically, to a system and method for providing structured pain analysis information by receiving a user's pain-related free description, pain data, and psychological profile data, and linking language patterns, kernel structures, link structures, and quantitative indicators. Background Technology

[0002] Traditional pain assessment has focused on certain physical indicators such as pain intensity, location, and duration. However, actual users often do not express pain merely as a simple physical sensation, but frequently describe it through linguistic structures such as burden of responsibility, threat to survival, emotional distress, suppression, or repetitive rumination. Although such free-form descriptions can reflect the structural aspects and fixation patterns of pain, existing systems [consist of] free-form descriptions and structured clinical data

[0003] There were limitations in generating reproducible analysis results by processing them together.

[0004] Furthermore, it is difficult to reflect the structural repetitiveness of pain-related narratives using only simple sentiment analysis or keyword extraction, and a technical pipeline for generating objective analytical information objects by directly linking free-form input with quantitative indicators has not been sufficiently presented.

[0005] Prior art literature related to the technology forming the background of the present invention includes U.S. Patent Application Publication US2010 / 0145720 and U.S. Patent Application Publication US2018 / 0260526. The problem to be solved

[0006] The objective of the present invention is to provide objective and reproducible pain analysis information by collecting users' pain-related free descriptions, structured clinical data, and psychological profile data together to classify repeatable language patterns, deriving major pain kernels and link structures, and linking them with quantitative indicators such as KSI and KLM.

[0007] ㆍ Classify free-form text into repeatable pattern categories.

[0008] ㆍ Derivative kernel and link structures by linking pattern classification results with structured clinical data.

[0009] Calculate KSI and KLM using mathematical operation rules.

[0010] Create a structured analysis information object that includes patterns, kernels, links, and quantitative indicators.

[0011] Ensure the possibility of output in the form of screens, reports, or data objects, and subsequent service development. means of solving the problem

[0012] A system according to one embodiment of the present invention may include the following configuration.

[0013] ㆍ (i) A communication unit that receives free-form data and structured clinical data from a user terminal

[0014] ㆍ (ii) an analysis server including memory and a processor that performs pattern classification, kernel derivation, link calculation, and quantitative linkage on the said data

[0015] ㆍ (iii) A storage unit or database that stores or transmits structured analysis information objects

[0016] ㆍ (iv) Interface section that receives visual input of the pain area and induces free-form sentence input (optional configuration)

[0017] The above structured clinical data may include, for example, the following items.

[0018] ㆍ VAS (Pain Intensity)

[0019] ㆍ Pain spread score

[0020] ㆍ Sensory hypersensitivity score

[0021] ㆍ CDI_E, CDI_I, CDI_S, CDI_R, CDI_M CDI (Core Domain Index) is the patient's free-form and structured As a core domain-based multiaxial clinical indicator derived from clinical data, simple pain intensity Not the score, but the emotions, identity, survival pressure, burden of responsibility, and loss of meaning included in the patient's free narrative It is an indicator for evaluating structures by decomposing them into independent domain dimensions. Here, the domain is sentiment, It refers to areas of analysis such as identity, survival, responsibility, and meaning, and the Kernel corresponds to each Domain It refers to the unit of analysis or the structure of analysis derived from free-form and structured clinical data. CDI_E is As an Emotional Domain-based indicator, it is a metric representing emotional burden. CDI_I is an Identity Domain As a foundational indicator, it is an indicator representing the burden associated with identity impairment or changes in self-perception. CDI_S is As a Survival Domain-based indicator, it is a metric related to survival pressure, safety threats, and economic survival burden. CDI_R is a Responsibility Domain-based indicator covering responsibility overload, caregiving burden, occupational duties, and family. It is an indicator related to role burden. CDI_M is a Meaning Domain-based indicator regarding loss of meaning and life's It is an indicator related to a sense of emptiness and pessimism about the future. CDI_total is the above CDI_E, CDI_I, CDI_S, CDI_R and It is a comprehensive clinical indicator calculated by integrating CDI_M.

[0022] ㆍ Inhibition intensity (S), duration (T), emotional resonance (A)

[0023] ㆍ Pain location, age, gender, and other basic clinical information Effects of the invention

[0024] According to the present invention, by processing a user's free description, pain data, and psychological profile data in a single information processing pipeline, objective analysis information that simultaneously reflects the physical severity and maintenance burden of pain can be provided. Furthermore, by generating language patterns, kernels, and link structures as structured analysis information objects, reproducible data processing results for pain-related tasks can be provided. Moreover, the possibility of generating additional empathy aid phrases, follow-up questions, or risk-related flags provides scalability during the service and productization stages. Brief explanation of the drawing

[0025] Figure 1 shows the overall system configuration diagram. Figure 2 shows a processing flowchart from the input of free description and structured data to the creation of a structured analysis information object. Figure 3 shows an example diagram of a major pattern link structure. Figure 4 shows an example of a structured output object. Figure 5 shows a scatter plot of the correlation between CDI_total and KSI. Figure 6 shows a bar chart comparing the average KSI by major language patterns. Specific details for implementing the invention

[0026] The calculation unit can calculate the pain index and control index using the following mathematical formula.

[0027] ㆍ KSI = VAS × Pain Diffusion Score × Hypersensitivity Score

[0028] ㆍ KLM = S (Inhibitory Strength) × T (Duration) × A (Emotional Resonance)

[0029] The above formula is intended to quantify physical pain severity and maintenance burden according to repeatable operation rules, and enables the generation of structured analysis information by combining simple free text with mathematical output values. However, the present invention does not exclude modified embodiments including variations in coefficients or weights of specific formulas.

[0030] The analysis unit of the analysis server can classify free narratives into at least one pattern among Burden of Responsibility (RB), Threat to Survival (ST), Emotional Distress (ED), Inhibition / Repression (SI), Loss of Meaning (ML), Identity Instability (II), Repetitive Rumination (RR), and Relational Sensitivity (RS). In this specification, a narrative pattern refers to a pattern identified in a patient's free narrative, and RB stands for Responsibility Burden, SI for Suppression / Inhibition, and ST is Survival Threat, ED is Emotional Distress, and II is Identity Instability and ML stand for Meaning Loss.

[0031] In addition, the analysis unit can derive one or more kernel structures among E, I, S, R, and M based on the above classification results and structured clinical data, and can calculate one or more link structures among RB→SI, RB→ST, ST→ED, and RB→II. In this specification, the E, I, S, R, and M kernel structures are respectively the Emotional Domain, Identity Domain, kernel structures corresponding to the Survival Domain, Responsibility Domain, and Meaning Domain It signifies. Additionally, the link structure between patterns represents the transition relationship between narrative patterns, and RB→SI represents responsibility. In the relationship where burden leads to suppressed or withdrawn behavior, RB→ST is the burden of responsibility as survival pressure. Expanding relationships, ST→ED is a relationship where survival threats transfer into emotional distress, and RB→II is responsibility It refers to relationships where the burden leads to damage to identity.

[0032] The above pattern-kernel-link structure can be stored as a core field of a structured analysis information object and, depending on the embodiment, can be utilized as basic information for generating subsequent interpretation memos or risk-related flags. A mixed kernel can be determined when multiple kernel values ​​are above a threshold; for example, if two or more kernel scores among five dimensions are 7 points or higher, it can be determined as a mixed kernel. However, the threshold value can be adjusted according to system settings, and the present invention is not limited to a specific numerical value of 7 points. In the claims, it can be expressed as "above a preset threshold value" to allow for adaptation to future changes in the criteria.

[0033] The analysis unit can classify narrative patterns using a natural language processing engine. In one embodiment, the natural language processing engine can classify free narratives into pattern categories using one or more of (i) matching based on a predefined keyword dictionary, (ii) analysis of co-occurrence frequency within a context window, or (iii) similarity analysis based on embedding vectors. Additionally, expressions such as "should," "duty," and "responsibility" can be extracted from a sentence through morphological analysis and classified by matching them with a Responsibility Burden (RB) pattern database. The above implementation examples do not limit the invention and may be implemented using a rule-based method, a learning-based method, or a hybrid method thereof.

[0034] The system may further include an interface unit that receives visual input of the pain area through a user terminal and induces the input of free-form sentences. The interface unit transmits the input data to an analysis server or a local processing module and can output the calculated analysis information in the form of a screen, report, or data object. According to an embodiment, the analysis information object may include interpretation memos, empathy support phrases, follow-up questions, or risk-related flags, which can be utilized as differentiation features during the service commercialization phase.

[0035] As a result of analyzing a dataset of 110 subjects according to one embodiment of the present invention, the Pearson correlation coefficient between CDI_total and KSI was found to be r = 0.53, and the Pearson correlation coefficient between CDI_total and VAS was found to be r = 0.61. In addition, high correspondence was observed between primary patterns and link outputs; for example, the ED-primary pattern and the E link showed a 100.0% correspondence, the ST-primary pattern and the S link showed a 97.7% correspondence, and the RB-primary pattern and the R link showed a 90.4% correspondence. This can be utilized as an algorithm reliability indicator demonstrating that the present invention can convert free-form input into structured analysis information according to repeatable rules. The scatter plot in Fig. 5 visually represents the positive correlation trend between CDI_total and KSI, and the bar chart in Fig. 6 visually represents the average KSI difference by major language patterns.

[0036] When a user inputs a free-form sentence and provides pain data and CDI data, the analysis unit of the analysis server determines primary and secondary patterns using a natural language processing engine. The computation unit calculates KSI and KLM based on the structured clinical data, and the analysis unit derives key kernels and link structures based on the pattern and CDI values. A structured analysis information object containing pattern identifiers, kernel identifiers, link identifiers, and quantitative indicators is recorded in the storage unit or database, and if necessary, interpretation memos, empathy auxiliary phrases, follow-up questions, or risk-related flags can be additionally generated as dependent functions. In the embodiment, the free-form sentence input by the user is mapped to a specific pattern category with a high correspondence rate, and the quantitative computation result is included in the structured object and output, thereby providing repeatable information processing results according to the same rules. Explanation of the symbols

[0037] 110 : Input section 111: Input data (free description, VAS, KSI, CDI, S / T / A, etc.) 120 Narrative Pattern Analysis Department 130 : Kernel derivation section 140 : Link output section 150 : Quantitative linkage section 160 : Output section 161 Structured output information (main kernel, pattern code, pattern link) 162 : Dependency analysis objects (interpretation memo, empathy phrase, follow-up question / flag) 210~260 : Each processing step (process) from input to output 310~330 : Each pattern link (RB, SI, ST, ED, etc.) and connection relationships 410 : clinician-support output screen example

Claims

Claim 1 A communication unit that receives free-form descriptive data and structured clinical data related to a user's pain from a user terminal; including a memory and a processor, wherein the processor receives the free-form descriptive data Responsibility Burden(RB), Suppression / Inhibition(SI), Survival Threat(ST), At least one narrative among Emotional Distress (ED), Identity Instability (II), and Meaning Loss (ML) as a pattern Classify and based on the above patterns and structured clinical data Emotional Domain At least one of the Identity Domain, Survival Domain, Responsibility Domain, and Meaning Domain Deriving one or more kernel structures, One or more links indicating the transition relationship between the above narrative patterns Calculate the structure, and the pain intensity (VAS), pain diffusion score, and sensation included in the above structured clinical data. Calculate the Pain Index (KSI) using the hypersensitivity score, and the inhibition intensity (S), duration (T), and emotional The control index (KLM) is calculated using resonance (A), and A pattern identifier representing the above narrative pattern, a kernel identifier representing the above kernel structure, a link identifier representing the above link structure, and the above pain Index (KSI) and the above-mentioned control index (KLM) A pain analysis information provision system using user free description and structured clinical data, characterized by including an analysis server that generates a structured analysis information object including, and a storage unit or database that stores or transmits said structured analysis information object. Claim 2 In paragraph 1, the pain index (KSI) comprises the pain intensity (VAS), the pain diffusion score, and the sensation A system characterized by being calculated as the product of hypersensitivity scores Claim 3 In paragraph 1, the control index (KLM) comprises the inhibition intensity (S), the duration (T), and the emotional A system characterized by being calculated as a product of resonance (A). Claim 4 A system according to claim 1, characterized in that the processor determines a mixed kernel when two or more of the five-dimensional kernel values ​​are greater than or equal to a preset threshold. Claim 5 A system according to claim 4, characterized in that the threshold value is 7 points in one embodiment. Claim 6 A system according to claim 1, wherein the processor utilizes one or more of keyword dictionary-based matching, contextual co-occurrence frequency analysis, or embedding vector-based similarity analysis within the free-form data through a natural language processing engine, and classifies patterns by extracting expressions such as "should," "duty," and "responsibility" from a sentence through morphological analysis and matching them with a responsibility burden (RB) pattern database. Claim 7 In claim 1, the structured clinical data includes pain intensity (VAS), pain diffusion score, sensory hypersensitivity score, Includes CDI_E, CDI_I, CDI_S, CDI_R, CDI_M, inhibition intensity (S), duration (T), and emotional resonance (A), The above CDI_E is an Emotional Domain-based metric, and the above CDI_I is an Identity Domain-based metric, and The above CDI_S is a Survival Domain-based metric, and the above CDI_R is a Responsibility Domain-based metric. It is an indicator, and the above CDI_M is a Meaning Domain-based indicator, and the above analysis server is the above CDI_E, CDI_I, A system characterized by integrating CDI_S, CDI_R, and CDI_M to calculate CDI_total. Claim 8 In claim 1, the link structure comprises one or more of RB→SI, RB→ST, ST→ED, or RB→II. Including, the above RB→SI is a transference relationship in which the burden of responsibility leads to suppression or endurance behavior, the above RB→ST represents a transition relationship where the burden of responsibility extends to survival pressure, and the aforementioned ST→ED represents a survival threat Regarding the transference relationship that leads to emotional distress, the aforementioned RB→II connects the burden of responsibility to identity impairment. A system characterized by each representing a transition relationship. Claim 9 A system according to claim 1, further comprising an interface unit that receives visual input of a pain area through a user terminal and induces input of a free-form sentence. Claim 10 A system according to claim 1, characterized in that the structured analysis information object further includes one or more of an interpretation memo, an empathy auxiliary phrase, a follow-up question, or a risk-related flag as a dependent function.

Images