Pelvic organ prolapse disease preoperative doctor-patient joint decision evaluation system based on artificial intelligence

Abstract

The invention relates to the technical field of decision support, in particular to a pelvic organ prolapse disease preoperative doctor-patient joint decision evaluation system based on artificial intelligence, which comprises an initial inquiry acquisition module, a communication behavior and expression evaluation module, a cognition test module, a patient feature analysis module and a decision communication module. According to the method, through structural induction of multi-dimensional health data, subjective feeling and psychological states of patients are meticulously quantified, interaction features and knowledge understanding short boards in the communication process are actively mined, label type feature affiliation is automatically generated, and self-adaptive information matching for individual characteristics and risk features is achieved; the method cooperatively promotes efficient fusion of willingness, cognition and schemes between doctors and patients, enhances participation and continuous tracking ability of the whole diagnosis and treatment process, optimizes pertinence of joint decision, strengthens dynamic information feedback, accurately recognizes weak links and specific appeals, supports high-level data closed-loop and behavior tracking, and further promotes scientific decision and improves patient experience.

Description

Technical Field

[0001] This invention relates to the field of decision support technology, and in particular to an artificial intelligence-based preoperative joint decision-making and assessment system for pelvic organ prolapse diseases. Background Technology

[0002] The decision support field involves providing relevant information and reference opinions to healthcare professionals and patients during the diagnosis and treatment process to assist them in making clinical decisions. This includes medical data collection, clinical information analysis, treatment suggestion generation, and doctor-patient communication. Its overall development aims to standardize medical processes, optimize treatment pathways, and improve the scientific rigor of clinical decision-making. Among these, the traditional preoperative consultation system for patients with pelvic organ prolapse involves doctors explaining the patient's condition, outlining surgical and non-surgical treatment options, and outlining potential risks and benefits through face-to-face communication before surgery. This assists the patient in choosing a treatment path based on their personal wishes and clinical recommendations. Information transmission and communication are typically accomplished through paper or oral inquiries, handwritten medical records, clinical forms, and the patient's past medical history.

[0003] Communication between doctors and patients during pelvic organ prolapse surgery is challenging. Currently, there is no efficient communication technology. Traditional conversations often exhibit problems such as heavy reliance on subjective judgment, fragmented data, poor communication efficiency, and difficulty in systematic quantification during patient interviews and information collection. The lack of a data integration mechanism for individualized patient behavior and cognition during doctor-patient communication makes it easy for symptoms and expectations to not be fully presented or important characteristics to be fully captured. In situations where patients have a weak willingness to express themselves, diverse cognitive blind spots, or limited interaction, subjective judgment and verbal communication are difficult to continuously monitor and trace, affecting the accurate matching and dynamic tracking of treatment recommendations. This hinders the improvement of the quality of joint doctor-patient decision-making and the level of personalized medical services, leading to doctor-patient decision bias. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention provides an artificial intelligence-based preoperative collaborative decision-making and assessment system for pelvic organ prolapse diseases. The technical solution is as follows:

[0005] On the one hand, an artificial intelligence-based preoperative doctor-patient collaborative decision-making assessment system for pelvic organ prolapse diseases is provided, including:

[0006] The initial inquiry and data collection module analyzes the patients' symptom distress content in the preoperative interview, interprets the patients' statements on the impact of symptoms on their lives, analyzes their psychological state, compares the proportion of symptoms and life impact, adjusts the parameter weights in combination with subjective narrative content, judges health status, and obtains subjective health score parameters.

[0007] The communication behavior and expression assessment module, based on the subjective health score parameters, determines the frequency of proactive expression, analyzes questioning and feedback performance, selects key issues in the communication process, compares the relationship between the number of proactive expressions and feedback, analyzes initiative, and obtains a communication initiative index.

[0008] The cognitive testing module analyzes the understanding of basic knowledge answers based on the aforementioned communication initiative index, screens the performance of answers to treatment methods, judges the mastery of postoperative risks and recovery priorities, compares the differences in answer performance, and statistically analyzes high-frequency knowledge blind spots to obtain characteristics of cognitive weaknesses.

[0009] Based on the cognitive weakness features, the patient feature analysis module compares the normalized performance of various symptoms and needs, analyzes the acceptance attitude during the postoperative recovery period, screens risk tolerance parameters, judges parameter combinations, adjusts the label classification order, and obtains the feature category label.

[0010] The decision communication module analyzes the doctor's surgical path parameters based on the feature category labels, determines the correspondence between the patient labels and decision suggestions, filters the matching statistical results, compares the consistency of information, integrates the doctor-patient joint decision-making adaptation, and obtains doctor-patient matching degree data.

[0011] On the other hand, the subjective health scoring parameters include physical condition score, daily activity ability score, and psychological state score; the communication initiative indicators include information expression activity, interaction frequency, and proactive communication attitude; the cognitive weakness characteristics include knowledge blind spots, risk perception misconceptions, and missing treatment information; the feature category labels include hierarchical classification number, recovery period attitude identifier, and risk acceptance category; and the doctor-patient matching data includes consistency in treatment plan selection, label association level, and effectiveness of joint doctor-patient decision-making.

[0012] On the other hand, the initial query and data collection module includes:

[0013] The symptom distress assessment submodule analyzes the symptom distress content of patients in the preoperative interview, compares the degree of influence of each symptom in the subjective narrative, judges the severity level of all symptom manifestations, and obtains the symptom influence balance factor.

[0014] The life impact analysis submodule compares the symptom impact balance factor with the patient's subjective expression of the degree of limitation in daily life, work, social life and exercise, analyzes the changes in the proportion of the two types of impact in the expression, screens the most prominent manifestation of life limitation, and obtains the life impact proportion parameter.

[0015] The psychological state assessment submodule optimizes the weighting relationship between psychological state and subjective life narratives based on the aforementioned life impact ratio parameter, combined with verbal and nonverbal expressions of psychological state during interviews, calculates and summarizes the performance of various influencing factors, and obtains subjective health score parameters.

[0016] On the other hand, the communication behavior and expression evaluation module includes:

[0017] The expression frequency identification submodule analyzes the language content spontaneously expressed by patients during the interview based on the subjective health score parameters, identifies statements that actively express needs and raise questions in each language segment, judges the continuity of the expression content and the fluency of the language, compares the occurrence ratio of different expression categories in the interview, and obtains the active communication activity level.

[0018] The interactive behavior extraction submodule analyzes the patient's questions and feedback during the consultation process based on the active communication activity, filters the number of questions and feedback details involved in the interactive language, compares the relationship between interactive behavior and active expression, and obtains the interactive response coupling coefficient.

[0019] Based on the interaction response coupling coefficient, the initiative assessment submodule analyzes the dominant characteristics of various expression methods in the interview content, determines the distribution among emotional expression, questioning language and declarative expression, adjusts the weight ratio of different expression types in the initiative assessment, and obtains the communication initiative index.

[0020] On the other hand, the cognitive testing module includes:

[0021] Based on the aforementioned communication initiative index, the disease understanding and identification submodule analyzes the patient's answers to basic knowledge questions about pelvic organ prolapse during the knowledge Q&A session, identifies the basic knowledge and statements for each question, compares the correspondence between the answer selection and the actual understanding, judges the consistency between the question answer and the patient's description, and obtains the basic knowledge mastery rate.

[0022] The treatment mastery screening submodule, based on the aforementioned basic knowledge mastery rate, screens patients' answers to questions related to various treatment methods and postoperative recovery, analyzes the mastery performance of answers under treatment plan questions, judges the correspondence between answer selection and key knowledge points, and obtains the degree of understanding of treatment points.

[0023] The knowledge blind spot statistics submodule, based on the understanding of the treatment points, filters out frequently erroneous knowledge content during the question-and-answer process, counts the locations of repeatedly erroneous knowledge points in the answers, compares the distribution of frequently erroneous questions, and obtains the characteristics of cognitive weaknesses.

[0024] On the other hand, the frequency of proactive expression refers to the frequency with which patients spontaneously express their demands, thoughts, questions, and needs during the communication process, while the initiative refers to the depth of participation, enthusiasm for expression, speed and quality of information feedback demonstrated by patients in doctor-patient interactions.

[0025] On the other hand, the patient feature analysis module includes:

[0026] The normalized performance comparison submodule, based on the cognitive weakness characteristics, converts each parameter into a standardized data set, compares the proportion of each parameter in the set, filters the dominant parameters, and adjusts the parameter sorting order to obtain a unified parameter trend factor.

[0027] The recovery period attitude analysis submodule analyzes the data related to the auxiliary treatment based on the unified parameter trend factor. By comparing the data distribution of each recovery period, it filters out the recovery period characteristics with correlation and obtains the recovery period acceptance.

[0028] The risk tolerance performance submodule analyzes the combination relationship of characteristic parameters based on the recovery period acceptance, determines the label distribution of each combination type, optimizes the label classification and sorting, and establishes a hierarchical structure of label classification to obtain the feature belonging category label.

[0029] On the other hand, the basic knowledge comprehension refers to the patient's correct answering performance in the knowledge test regarding the basic knowledge of pelvic organ prolapse. The answering performance refers to the patient's correctness of the answer, the logic of the response, and whether the patient has grasped the core information in the knowledge question and answer.

[0030] On the other hand, the decision communication module includes:

[0031] The surgical path analysis submodule, based on the feature category labels, determines the parameters involved when doctors recommend surgical paths for each label, compares the correspondence between each label and the surgical path parameters, identifies the association between labels and paths, and obtains the surgical path pairing quantity.

[0032] The parameter matching and filtering submodule filters the matching data of patient ID tags and doctor's suggested parameters based on the surgical path matching volume, judges the structural combination of tag parameters and suggested content, analyzes the distribution of consistent content in the matching process, and obtains the tag suggestion fit degree.

[0033] The information consistency determination submodule compares the information structure of the patient's expression and the doctor's recommended content based on the suggested fit of the tags, optimizes the process order of each step in the matching process, adjusts the operation standards of the determination step, integrates the information matching performance, and obtains the doctor-patient matching degree data.

[0034] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0035] By structuring and summarizing multidimensional health data, we can meticulously quantify patients' subjective feelings and psychological states, proactively uncover interactive features and knowledge gaps in the communication process, automatically generate tagged feature attributions, and achieve adaptive information matching based on individual traits and risk characteristics. This collaboratively promotes the efficient integration of doctors' and patients' wishes, cognition, and plans, enhances participation and continuous tracking capabilities throughout the entire treatment process, optimizes the relevance of joint doctor-patient decision-making outcomes, strengthens dynamic information feedback, accurately identifies weaknesses and specific demands, supports high-level data closure and behavior tracking, and further promotes scientific decision-making and improves patient experience. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 This is a schematic diagram of the system of the present invention;

[0038] Figure 2 This is a schematic diagram of the system framework of the present invention;

[0039] Figure 3 This is a flowchart of the initial query acquisition module of the present invention;

[0040] Figure 4 This is a flowchart of the communication behavior and expression evaluation module of the present invention;

[0041] Figure 5 This is a flowchart of the cognitive testing module of the present invention;

[0042] Figure 6 This is a flowchart of the patient feature analysis module of the present invention;

[0043] Figure 7 This is a flowchart of the decision communication module of the present invention. Detailed Implementation

[0044] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0045] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0046] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0047] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0048] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0049] This invention provides an artificial intelligence-based preoperative doctor-patient collaborative decision-making assessment system for pelvic organ prolapse diseases, such as... Figure 1 As shown, the system includes:

[0050] The initial inquiry and data collection module analyzes the patients' symptom distress content in the preoperative interview, interprets the expression of the impact on quality of life, analyzes the verbal and physical manifestations when the psychological state changes, compares the changes in the ratio between the patients' symptom distress and the impact on life, and judges the impact on health status by comparing emotional state and subjective narrative content to obtain subjective health score parameters.

[0051] The communication behavior and expression assessment module, based on subjective health score parameters, determines the frequency of patients' initiative to express their willingness, analyzes the performance of patients' questions and feedback during the communication process, screens key questions raised during the communication, compares the correlation between the frequency of initiative to express and the number of communication feedback, optimizes the judgment criteria for different expression types, analyzes patients' initiative, and obtains communication initiative indicators.

[0052] The cognitive testing module analyzes patients' understanding of basic disease knowledge based on communication initiative indicators, screens their performance on common treatment methods, judges the accuracy of their grasp of postoperative risks and recovery priorities, compares the differences in performance on knowledge points, optimizes key knowledge nodes in cognitive testing, and statistically analyzes frequently occurring knowledge blind spots to obtain characteristics of cognitive weaknesses.

[0053] The patient feature analysis module is based on cognitive weakness features. It compares the normalized performance of patients in various symptoms and needs expressions, analyzes the acceptance attitude during the postoperative recovery period, screens risk tolerance performance parameters, judges the combination relationship between parameters, optimizes the grouping and classification criteria, and adjusts the label sorting logic to obtain the feature category label.

[0054] The decision communication module analyzes the associated parameters of the surgical path recommended by the doctor based on the feature attribution category label, determines the correspondence between the label in the patient number and the doctor's decision suggestion, filters the statistical results of parameter matching, compares the consistency of information between the patient and the doctor, optimizes the standardized process of matching, integrates and evaluates the results of the doctor-patient joint decision-making adaptation, and obtains doctor-patient matching degree data.

[0055] Subjective health scoring parameters include physical condition score, daily activity ability score, and psychological state score; communication initiative indicators include information expression activity, interaction frequency, and proactive communication attitude; cognitive weakness characteristics include knowledge blind spots, risk perception misconceptions, and missing treatment information; feature classification labels include hierarchical classification number, recovery period attitude label, and risk acceptance category; and doctor-patient matching data include consistency in treatment plan selection, label association level, and effectiveness of joint doctor-patient decision-making.

[0056] In the initial inquiry and data collection module, symptom distress refers to the severity and impact of patients' subjective descriptions of their pelvic organ prolapse symptoms (such as vaginal prolapse, a feeling of heaviness, and discomfort during bowel movements / urination). This information is primarily collected using the PFDI-20 questionnaire (an internationally recognized standardized questionnaire for category A, and its reliability and validity have been verified in Chinese). The expression of impact on quality of life refers to patients' self-evaluations of limitations in daily life, work, social interactions, and exercise due to symptoms. This can be achieved using the PFIQ-7 questionnaire (an internationally recognized standardized questionnaire for category A, and its Chinese version has been verified). The scores of quality of life scales (such as those used for reliability and validity verification) are reflected in the data; changes in proportion refer to the respective weights of symptom distress and life impact in the overall subjective health description, and comparing the changes in the proportions of these two factors reveals the focus of the complaints; adjusted weights refer to the dynamic adjustment of the participation ratio of each type of complaint in the health status score based on the degree of emphasis on the impact of different complaints (such as emotional fluctuations / actual functional limitations) on the overall score during data processing; the impact of health status refers to the comprehensive judgment of the patient's current physical and mental state by summarizing various factors, which is used as the basis for stratification and personalized recommendations in subsequent modules.

[0057] In the communication behavior and expression assessment module, the frequency of proactive expression refers to the frequency with which patients spontaneously express their needs, ideas, questions, and demands during the communication process, measuring their willingness to actively participate; the performance of patient questioning and feedback refers to the specific behaviors of patients proactively asking questions, clarifying, and responding to feedback from doctors or the system during consultations, reflecting their enthusiasm for information reception and communication; key issues refer to the key questions raised by patients throughout the communication process that are closely related to and guide the decision-making of treatment plans; the criteria for judging different expression types refer to determining whether patients express themselves through direct statements, questions, emotional expressions, narrative explanations, etc., identifying and classifying different forms of expression; patient initiative refers to the comprehensive performance of patients in terms of the depth of participation, enthusiasm for expression, speed and quality of information feedback in doctor-patient interactions.

[0058] In the cognitive testing module, the understanding of basic disease knowledge refers to the patient's correct answers to questions about pelvic organ prolapse (definition, symptoms, etc.) during the knowledge test. Answer performance refers to the correctness of the patient's choices / answers, logical responses, and mastery of core information in the knowledge-based questions and answers. Accuracy of mastery refers to the patient's correct understanding and actual mastery of treatment options, surgical risks, and key points of postoperative recovery. Performance differences refer to the fluctuations and differences in the patient's mastery of different knowledge questions, used to identify cognitive gaps. Cognitive testing verifies the patient's actual mastery of disease knowledge, treatment options, risk awareness, and key points of recovery through questioning or questionnaires. Key knowledge points refer to important knowledge points necessary for decision-making, such as "surgery is not the only option," "postoperative recurrence," and "indications for conservative treatment." Frequently occurring knowledge blind spots refer to knowledge points that multiple patients answered incorrectly or did not fully understand; high frequency indicates areas of collective cognitive weakness that the system needs to address.

[0059] In the patient characteristic analysis module, each symptom refers to the subjective scale score or level covering all specific symptom parameters related to prolapse (such as pressure, urinary incontinence, constipation, etc.); needs expression refers to the patient's expression and scale record of specific needs such as treatment expectations, organ preservation, economic affordability, and life recovery; normalization performance refers to converting different parameters according to the same standard / score range for easy subsequent aggregation analysis; acceptance attitude refers to the patient's acceptance and cooperation with inconveniences during the postoperative recovery period (such as bed rest, reduced activity, vaginal bleeding, use of adjuvant drugs, etc.); risk tolerance performance parameters refer to the patient's subjective acceptance level of uncertain or negative outcomes such as surgical complications, recurrence, pain, and reoperation; the combination relationship between parameters refers to the comprehensive characteristics of the patient (such as high symptoms, low risk tolerance, high needs) after aggregation of multiple parameters, providing a grouping basis for the attribution label; grouping and classification criteria refer to the feature grouping standards constructed based on parameters such as symptoms, needs, acceptance, and risk, used for patient stratification and recommendation strategies within the system; label sorting logic refers to the systematic sorting of attribution categories (such as high need type, high anxiety type, low knowledge type, etc.) to facilitate priority attention by doctors and the system.

[0060] In the decision-making and communication module, the associated parameters for recommended surgical pathways refer to the key information (such as contraindications, expectations, and high-risk factors for complications) referenced by doctors when recommending surgical treatment methods based on the patient's comprehensive characteristics; the tags in the patient ID refer to the patient characteristic tags automatically generated by the system (e.g., "high knowledge, high needs," "risk-sensitive"), with each tag belonging to one or more grouping characteristics; the correspondence refers to comparing the patient characteristic tags with the key information in the doctor's recommendations one by one to determine whether they are consistent or conflicting; the statistical results refer to the specific statistical analysis results of the quantity and proportion of consistent and conflicting items in the matching process; the consistency of information at both ends refers to the degree of matching between the patient's self-expression and the doctor's professional judgment in the choice of treatment plan, serving as a measure of the degree of achievement of joint doctor-patient decision-making; the standardized process refers to the decision-making and communication process being carried out according to unified system standards (e.g., information verification first, then treatment plan recommendation, and finally two-way confirmation); the evaluation of the fit of joint doctor-patient decision-making refers to comprehensively considering key data such as consistency and deviation to determine whether the current joint doctor-patient decision-making is smooth and whether key intervention is needed.

[0061] like Figure 2 and Figure 3 As shown, the initial query and data collection module includes:

[0062] The symptom distress assessment submodule analyzes the symptom distress content of patients in the preoperative interview, compares the degree of influence of each symptom in the subjective narrative, judges the severity level of all symptom manifestations, and obtains the symptom influence balance factor.

[0063] Read the interview transcripts and locate the core symptom words appearing in key phrases, such as "feeling of heaviness," "difficulty defecating," "urinary incontinence," "discomfort during sexual intercourse," and "vulvar prolapse." Record the frequency of each symptom; for example, "feeling of heaviness" appeared 3 times in one interview, "urinary incontinence" appeared 2 times, and "discomfort during sexual intercourse" appeared once. Then, extract the impact words from the subjective descriptions accompanying each symptom, such as "I am in great pain, and it bothers me so much that I can't do anything," "It has a significant impact, greatly affecting my life," "It has some impact, but I can still carry out daily activities normally," and "It has no impact." Based on the impact words, a rating scale of 1 to 4 points is set. For example, "no impact" is assigned 1 point, "some impact, but I can still carry out daily activities" is assigned 2 points, "significant impact, greatly affecting my life" is assigned 3 points, and "I am in great pain, and it is troubling me so much that I can't do anything" is assigned 4 points. The symptom is then scored according to its corresponding subjective description. For example, if "a feeling of heaviness" appears 3 times, and each instance of heaviness corresponds to 4 points, the total score is 12, and the average score is 4. If "urinary incontinence" appears 2 times, and each instance corresponds to 2 points, the total score is 4, and the average score is 2. All symptoms are then sorted according to their average scores, and the following are selected. The three symptoms with the highest average scores were selected as the primary distress items. If symptom scores were tied, the higher frequency symptom was prioritized. The ratio of the total score for these three symptoms to the total score for all symptoms was then calculated. Assuming the total score for these three symptoms was 30 and the total score for all symptoms was 45, the ratio was 66.7%. Weighting coefficients were set based on this ratio; for example, items exceeding 60% were weighted at 1.2. After weighting, the weighted score for "feeling of pressure" was 4.8, and for "urinary incontinence" it was 4.2. These weighted values ​​were then re-aggregated and averaged to obtain the mean subjective distress score. This mean score was then averaged with the symptom scores of all preoperative interview patients in the built-in sample database. By comparing the mean scores, if the current patient's score is 4.5 and the sample mean is 3.5, then the current score is about 28% higher than the mean, and the patient is classified into the "high severity" group. If the score is 2.8, the patient is classified into the "low severity" group. Finally, based on the consistency of all symptoms in terms of intensity, subjective perception, and score distribution, it is determined whether the patient has a single symptom-dominant problem or multiple balanced problems. If the scores of the three main symptoms in the same patient are less than 1 point and their weights are similar, the patient is defined as "balanced"; otherwise, the patient is defined as "dominant". Based on this, the symptom influence balance factor value and label are formed.

[0064] The life impact analysis submodule compares the symptom impact balance factor with patients' subjective statements on the degree of limitation in daily life, work, social life, and exercise. It analyzes the changes in the proportion of the two types of impact in the expression, screens the most prominent manifestations of life limitation, and obtains the life impact proportion parameter.

[0065] Based on a comparison of the symptom impact balance factor and patients' descriptions of limitations in daily life, social life, and physical activity during interviews, the PFIQ questionnaire's item rules were applied to map patient statements to three dimensions (daily life, social life, and physical activity). Rankings were assigned according to content intensity, with scores ranging from 1 to 4. For example, in the daily life dimension, a patient stating "Recently I can't cook or mop the floor; I always feel like something is about to fall" was categorized as "no impact" and assigned 1 point; a statement like "I feel my daily life is somewhat affected, such as shopping or doing housework" was categorized as "somewhat affected" and assigned 2 points; "I feel my daily life is affected" was categorized as "quite affected" and assigned 3 points; and "I feel I can't take care of myself, and it's too much of a problem" was categorized as "very" affected. The "Impact" score is assigned 4 points, with a total score of 10 points and an average score of 2.5 points. This average score is then compared to the symptom impact balance factor. Assuming the symptom impact balance factor is 0.58 and the average life impact is 2.5, the calculated ratio is 2.5 / 0.58 ≈ 4.31. A score greater than 5 indicates a life-restriction-dominant type, 2 to 5 indicates a balanced type, and less than 2 indicates a symptom-disturbing-dominant type. The current value is 4.31, classifying it as a life-restriction-balanced type. The highest score is then marked as the life-dominant restriction item. In this example, "Very Impact" scores 4 points, accounting for 40% of the total score, exceeding the set 25% dominance threshold, therefore it is marked as "Restriction-dominant type." The percentage of the dominant item is recorded as the life impact percentage parameter and combined with the dominant label in subsequent health score calculations.

[0066] The psychological state assessment submodule optimizes the weighting relationship between psychological state and subjective life narratives based on the life impact ratio parameter and the verbal and nonverbal expressions of psychological state during interviews. It calculates and summarizes the performance of various influencing factors to obtain subjective health score parameters.

[0067] The system automatically identifies negative emotional statements in interview texts, such as "feeling depressed," "very annoyed," and "don't know what to do," and records corresponding body language responses, such as frowning, sighing, rubbing hands, or remaining silent with a downcast head during the description. An emotion rating scale is set from 1 to 5, where level 1 indicates no emotional speech or body language, and level 5 indicates strong emotional statements accompanied by multiple negative body language signals. For example, if a patient's statement includes "I'm so anxious I can't sleep at night," and they exhibit obvious frowning, sighing, or rapid speech during the interview, their score would be 4. The average emotion rating score for all patients in the sample is set at 3.2. If the current value is more than 20% higher than this average (i.e., exceeding 3.84), it is considered an emotional level. The weight of factors in the health score should be increased. At this time, the weight of emotion is adjusted from the default 0.4 to 0.5, and the weight of life impact is reduced from 0.6 to 0.5. After the weight adjustment, the emotion score and the life impact weight are multiplied by their respective weights. For example, if the emotion score is 4 and the life impact weight is 0.75, the total score calculated according to the new weight is 2 plus 0.375, resulting in 2.375. This value is set as the subjective health score parameter for the current patient and compared with the mean health score in the same patient sample. Those with a score greater than 3 are defined as those with higher health risk, 2 to 3 as the intermediate group, and those with a score less than 2 as the mild impact group, thus providing a health score level and specific numerical basis for subsequent classification.

[0068] like Figure 2 and Figure 4 As shown, the communication behavior and expression assessment module includes:

[0069] The expression frequency identification submodule analyzes the language content spontaneously expressed by patients during interviews based on subjective health score parameters, identifies statements that actively express needs and raise questions in each language segment, judges the continuity of the expression content and the fluency of the language, compares the occurrence ratio of different expression categories in the interview, and obtains the active communication activity level.

[0070] The process involves reading patient interview texts and segmenting each segment into sentences. Individual sentences are marked to indicate whether they contain expressions of active desire, treatment needs, or questions. Each segment is then iterated to determine if it begins with subjective, active words such as "hope," "want to know," "can I," or "what to do." If such expressions are found, they are counted as one active expression event. The total number of such statements across all interviews is tallied, and the segment numbers and order of their occurrence are recorded. For example, in a 15-minute interview, 18 segments were identified, of which 6 contained expressions of active needs and 4 contained questions. Ten paragraphs showed instances of active expression, resulting in an active expression rate of 10 divided by 18, which equals 0.556. Subsequently, language coherence analysis was performed on each paragraph to determine the presence of sentence interruptions, frequent word repetitions, and language jumps. A fluency threshold was set as follows: each paragraph contained more than 20 effective information words, had no repeated keywords more than three times, and exhibited logical coherence. A paragraph was considered fluent if it met any two or more of these three conditions. Of the patient's 18 paragraphs, 11 were identified as fluent, representing 61.1%. The active expression rate was then compared with the language fluency rate. The comparison is performed. If the difference between the two values ​​is less than 10%, it is defined as "consistent expression type"; otherwise, it is defined as "skewed expression type". Then, the proactive expression events are categorized into three types: "subjective intention", "questioning and discussion", and "proactive clarification". The percentage of each type in all proactive expression statements is calculated. For example, if subjective intention accounts for 6 statements, questions for 3, and clarification for 1, the percentages are 60%, 30%, and 10% respectively. A single type accounting for more than 60% is defined as "skewed expression type", and a balanced distribution of the three types is defined as "balanced expression type". The "skewed / balanced expression type" labels are used to evaluate the patient's proactive expression. The distribution of emphasis in active expression reflects whether the user focuses more on a single need or multiple needs during communication. Ultimately, based on the frequency of expression, paragraph distribution, language completeness, and type distribution, combined with the score level in the subjective health rating parameters, those with a score greater than 3.5 are defined as having a proactive expression rate of over 50%, a language fluency rate of over 50%, and a subjective intention expression rate of over 40%. Those who meet all three conditions are defined as having "high expressive activity," those who meet only two conditions are defined as having "medium expressive activity," and those who meet only one condition or none are defined as having "low expressive activity." This label and the numerical value constitute the active communication activity level.

[0071] The interactive behavior extraction submodule analyzes patients' questions and feedback during the consultation process based on the level of active communication, filters the number of questions and feedback details involved in the interactive language, compares the relationship between interactive behavior and active expression, and obtains the interactive response coupling coefficient.

[0072] First, extract all complete dialogues between questions and doctor / system responses from the interview transcripts. Calculate the total number of questions initiated by the patient and the total number of responses. Set a question-and-answer session with a structure of "question + answer + patient feedback." Number the questions. For example, questions such as "Is surgery not possible?", "Is the recurrence rate high?", and "How long after surgery can I start exercising?" appearing in paragraphs 1, 4, and 7 are counted as 3 initiated questions. Then, determine whether there is verbal feedback following the corresponding paragraph in response to the doctor's answer, such as "I understand," "I see," or "I'm still a little worried." The occurrence of such a statement is defined as a feedback event. A total of 3 feedback events were identified: 3 questions and 3 feedbacks. The interaction matching ratio was 100%. A matching ratio greater than 80% was defined as "high response type," between 50% and 80% as "medium response type," and less than 50% as "low response type." The current patient's matching value was 100%, classifying them as high response type. Subsequently, sentences containing key information words from all interactive dialogues were extracted, and the depth of the patient's response in the dialogue was analyzed. For example, "I understood that the mesh panel was about reinforcement, but what about exposure?" was defined as "deep feedback." Distinguishing between "superficial feedback" such as "okay" and those with a deep feedback ratio higher than 50%, the patient is marked as having "high feedback detail." Since two out of three feedback responses included supplementary or extended questions, the patient is categorized as high detail (deep feedback ratio > 50%), medium detail (deep feedback ratio between 20% and 50%), or low detail (deep feedback ratio < 20%). The total number of interaction rounds, number of questions, number of responses, and average feedback depth are calculated (each feedback is assigned a value: deep feedback = 1, superficial feedback = 0; average feedback depth = number of deep feedback responses / total number of responses; e.g., in three feedback responses...). The input parameters are: 2 (deep level → mean = 2 / 3 = 0.667) and the pairing ratio. These parameters are standardized (range 0-1) and input as vectors. They are then correlated with the active expression activity score. If the difference between the two activity scores is no more than 20%, it is considered "interaction consistent"; otherwise, it is considered "interaction differing". The overall result of the interaction response is combined with the active expression score proportionally. For example, if the active communication activity score is 0.6 and the interaction matching score is 0.9, the interaction response coupling coefficient is the average of the two, which is 0.75. This coefficient is used for subsequent initiative assessment.

[0073] The initiative assessment submodule analyzes the dominant characteristics of various expression modes in the interview content based on the interaction response coupling coefficient, judges the distribution among emotional expression, questioning language and declarative expression, adjusts the weight ratio of different expression types in initiative assessment, and obtains the communication initiative index.

[0074] All expressions in the interview text were extracted, including emotional expressions, interrogative expressions, and factual statements. Each sentence was labeled and categorized; for example, "I'm very worried about problems with the surgery" was labeled as an emotional expression, "Will this affect my sex life?" as an interrogative expression, and "Currently, the prolapse is mainly due to difficulty in defecation" as a declarative expression. The proportion of each type of expression in the total sentences was statistically analyzed. For example, out of 35 sentences identified, there were 9 emotional expressions, 11 interrogative expressions, and 15 declarative expressions, accounting for 25.7%, 31.4%, and 42.9% respectively. A dominant expression type was defined as one where any one expression type accounted for more than 45% of the total; otherwise, it was defined as a multi-expression type. Since the patient's proportion did not exceed the 45% threshold, it was classified as a multi-expression type. The distribution of expression types was jointly analyzed with the interaction response coupling coefficient. If it was a dominant expression type, it was assigned a dominant category. The first category has a weight of 0.5, while the other two categories each have a weight of 0.25. If it is a multi-expression type, then all three categories have a weight of 0.33. In subsequent scoring, the frequency of the three types of expression is multiplied by the corresponding weight, and then averaged with the coupling coefficient. A coupling coefficient higher than 0.8 is considered a high level of initiative, lower than 0.5 is considered a low level, and the middle level is considered a medium level. For example, if the frequency of emotional expression is 25.7% with a weight of 0.33, questioning expression is 31.4%, and declarative expression is 42.9% with a coupling coefficient of 0.75, then the comprehensive initiative score is the weighted average of the three multiplied by their respective weights and the average of the coupling coefficient, with a final value of 0.74. The scoring range is set as follows: 0.71 is high communication initiative, 0.5-0.7 is medium communication initiative, and lower than 0.5 is low initiative. This patient's score is 0.74, which is classified as high communication initiative, and the corresponding communication initiative index is output.

[0075] like Figure 2 and Figure 5 As shown, the cognitive testing module includes:

[0076] The disease understanding and identification submodule analyzes patients' answers to basic knowledge questions about pelvic organ prolapse based on the communication initiative index. It identifies the basic knowledge and statements for each question, compares the correspondence between the answer choices and the actual understanding, judges the consistency between the question answers and the patient's description, and obtains the basic knowledge mastery rate.

[0077] The system retrieves all questions related to basic knowledge of pelvic organ prolapse from the knowledge-based Q&A section, extracting the options and basic knowledge tags for each question. For example, if the question is "Which of the following is pelvic organ prolapse?", and the options include "vaginal prolapse," "urinary frequency and urgency," and "anal mass," this question is identified as a basic knowledge question, with the corresponding tag being "definition of prolapse." The correct option is "vaginal prolapse." The system compares the patient's choices to determine if they match the tag. If the option content matches the medical terminology corresponding to the basic knowledge name in the question, it is considered a correct answer; if the patient chooses "anal mass," it is marked as incorrect. Furthermore, the system considers whether the patient can describe the characteristics of pelvic organ prolapse in their open-ended questions. For example, statements like "vaginal prolapse causing friction and discomfort" or "vaginal prolapse may be bladder prolapse" can be mapped to an understanding of basic knowledge of pelvic organ prolapse. If this description corroborates the chosen option, the system is considered correct. Further assessment is made to determine if the patient possesses a basic understanding. Consistency statistics are then performed on the matching of each answer to a given question with the corresponding description. With a total of 5 questions, if the patient answers 3 questions correctly and their description in 2 of those questions matches the correct basic knowledge, the comprehension matching rate is marked as 60%, and the expression comprehension rate as 40%. The weighted average of these two scores yields a comprehensive comprehension score of 50%. The mastery rate is categorized as follows: 0-40% is low, 41-70% is medium, and above 71% is high. The current score of 50% falls into the medium mastery range. Finally, the following three calculations are weighted and averaged: answer accuracy (answer accuracy = number of correct answers ÷ total number of questions), description coverage (description coverage = number of questions containing the correct basic knowledge name ÷ total number of questions), and consistency ratio (consistency ratio = number of questions with correct answers and matching descriptions ÷ total number of questions). The combined evaluation output is the basic knowledge mastery rate.

[0078] The treatment mastery screening submodule is based on the basic knowledge mastery rate. It screens patients' answers to questions related to various treatment methods and postoperative recovery, analyzes the mastery performance of answers under treatment plan questions, judges the correspondence between answer choices and key knowledge points, and obtains the degree of understanding of treatment points.

[0079] The system retrieves patient responses related to surgical treatment methods, postoperative recovery, and risks and complications. It then sequentially filters the key knowledge points corresponding to the correct answer for each question. For example, in the question "Which treatment method is suitable for patients who wish to preserve their sexual function?", the key points are identified as "preservation of sexual function" and "treatment method selection." The correct answer is "pelvic floor reconstruction surgery (autologous or mesh implantation)." The system semantically compares the patient's choice with this key knowledge point. If the patient chooses "vaginal closure surgery," it is considered incorrect as it does not meet the key knowledge point. If the patient chooses the correct answer but stated in the interview record or questionnaire that they "did not understand the specific details of the closure surgery," the answer is correct but the understanding is insufficient. This type of situation is marked as "partially mastered" according to the set rules. The response results for each question are categorized into three types based on whether the key points are mastered: "correctly mastered," "incorrectly mastered," and "partially mastered." "Correctly mastered" is assigned a value of 1, "partially mastered" is 0.5, and "incorrectly mastered" is 0. The system accumulates the scores of all treatment-related questions. For example, if a patient answers 6 questions, with 3 correctly understood, 2 partially understood, and 1 incorrectly, the calculated score is 3 multiplied by 1 plus 2 multiplied by 0.5, which equals 4, for a total score of 6. The comprehension level is 4 divided by 6, which equals 66.7%. The thresholds are set as low comprehension (0-50%), medium comprehension (51%-75%), and high comprehension (76%-100%). The current patient's score is in the medium range, and they are judged as a medium-level master. In addition, the system further analyzes the coverage of key knowledge points. If a patient makes more than 2 mistakes on a knowledge point related to a question, that knowledge point is marked as a cognitive blind spot. The current patient is numbered according to the postoperative knowledge points, such as "judgment of surgical type", "risk acceptance", "postoperative recovery process", "impact on sexual life", etc. If the patient answers 2 questions incorrectly in a certain type of recovery question, the "postoperative recovery" category is marked as a cognitive blind spot. Based on the comprehension score and error distribution, the system outputs the patient's comprehension performance of the key treatment points and generates the comprehension parameters of the key treatment points.

[0080] The knowledge blind spot statistics submodule filters frequently missed knowledge content during the question-and-answer process based on the understanding of key treatment points, counts the location of repeatedly missed knowledge points in the answers, compares the distribution of frequently missed questions, and obtains the characteristics of cognitive weaknesses.

[0081] The formula for identifying the locations of repeatedly incorrect knowledge points in answers is as follows:

[0082] ;

[0083] Calculate the error distribution parameters of knowledge points By comparing the distribution of frequently missed questions, we can identify characteristics of cognitive weaknesses. Representing the Number of errors per knowledge point This represents the average number of errors across all knowledge points. Representing the The total number of times each knowledge point was mentioned in the questions. This represents the total number of knowledge points included in this round of statistics.

[0084] The knowledge point error distribution parameter is a comprehensive parameter reflecting the correlation between the distribution differences of answering errors for each knowledge point within the statistical scope and the total number of answers. If the number of errors for certain knowledge points is significantly higher than the average level, and these knowledge points are also involved in more questions, then... The value will shift in the positive direction, indicating the existence of concentrated areas with high error rates. If the number of errors for all knowledge points is evenly distributed and close to the average, then... A value close to zero indicates a relatively uniform distribution of errors overall. This value is used to quantify the differences in the distribution of errors across different knowledge points, and comprehensively reflects the position and impact of frequently erroneous knowledge points within the overall statistical analysis.

[0085] Records of answer performance for each knowledge point are collected, and each knowledge point is associated with a number. Extract the number of individual errors. Total number of responses At the same time, the error counts for all knowledge points are summed and averaged to obtain the average number of errors. In a practical application, patients were tasked with completing a quiz covering four knowledge points: basic knowledge of anterior pelvic prolapse (No. 1), understanding of surgical risks (No. 2), adaptation to conservative treatment (No. 3), and key points of postoperative recovery (No. 4). The original number of errors for each knowledge point was as follows: , , , The corresponding original number of responses are as follows: , , , The system normalizes the original error counts, and the normalized error counts are as follows: , , , The normalized number of responses are as follows: , , , Next, calculate the average number of errors. Its normalized value is:

[0086] ;

[0087] Substitute the above parameters into the formula:

[0088] ;

[0089] The numerator, calculated item by item, is:

[0090] , ;

[0091] , ;

[0092] Summing the four terms, we get:

[0093] ;

[0094] The denominator is: The calculation yields:

[0095] ;

[0096] The result indicates that the obtained knowledge point error distribution parameters The result is below the lower limit of the baseline interval [0.1, 0.3], indicating that the overall error distribution deviation in this round of patient knowledge quizzes is small, and there are no obviously concentrated knowledge points with errors. This means that the degree of deviation in the occurrence of errors by patients on different knowledge points is low, and the errors in the questions are relatively evenly distributed. After subsequent logical judgment, this parameter result does not meet the condition of "high-frequency concentrated bias". Therefore, the system does not trigger the aggregation and attribution of "cognitive weakness feature", but only marks it as "evenly distributed type" for recording and saving. The formula can highlight the relative performance of specific knowledge blind spots in the entire knowledge structure by introducing the product operation of the normalized difference of error frequency and coverage, and after the total weight is adjusted, thereby enhancing the accuracy of subsequent identification and the coordination of parameter application.

[0097] like Figure 2 and Figure 6 As shown, the patient feature analysis module includes

[0098] The normalized performance comparison submodule, based on the characteristics of cognitive weaknesses, converts each parameter into a standardized dataset, compares the proportion of each parameter in the dataset, filters the dominant parameters, and adjusts the parameter sorting order to obtain a unified parameter trend factor.

[0099] All symptom and treatment need parameters expressed by the patient during the subjective expression phase were retrieved. Each parameter was standardized, with a uniform range of 0 to 1. Parameters from different scales were converted to standard units using a linear ratio. For example, a feeling of heaviness in the PFDI score is 4 points out of 12, standardized to 0.3; no influence in the PFIQ score is 1 point out of 10, standardized to 0.1. After standardizing all parameters, they were categorized into "symptom expression parameter group" and "needs expression parameter group." The needs expression parameter group refers to the quantitative results of patient needs-oriented scale items and interview needs statements, for example... In the PFIQ questionnaire, "degree of activity limitation" is a symptom, while "expectation of postoperative quality of life improvement" is a need. Scoring is based on the symptom severity rating of the scale. The PFDI primarily reflects the degree of symptom distress, such as a feeling of heaviness or difficulty urinating; the PFIQ primarily reflects the degree of limitation in quality of life, but it is essentially still an impact caused by symptoms, therefore it is categorized as a "symptom expression parameter." Different scale items need to be classified according to their attributes (symptoms and needs). Each item within both parameter groups should be ranked, and the maximum, minimum, and mean values ​​of each group should be calculated. Then, the numerical differences of each parameter at the same position within the two groups should be compared. For example, the first... The first symptom item, "feeling of heaviness," scored 0.8; the first demand item, "expectation of complete improvement," scored 0.65, with a difference of 0.15; the second items scored 0.7 and 0.45 respectively, with a difference of 0.25. All differences were statistically analyzed and averaged to obtain the average performance deviation value of the two parameter groups under the same ranking. If the deviation value is greater than 0.3, it is classified as "significant difference in expression tendency"; if it is between 0.1 and 0.3, it is "moderate difference in expression tendency"; and if it is less than or equal to 0.1, it is "approaching difference in expression tendency." The current patient's average parameter deviation value is 0.22, classifying it as moderate difference. Further analysis of the changing trends of various parameters within the two groups is needed. Whether the trend is consistent or not, for example, if the symptom parameters decrease from high to low, and the demand parameters also decrease, it is marked as "consistent trend". If there is an intersection of high before low and low before high, it is marked as "reverse trend". The type of trend difference is determined by calculating the positive and negative values ​​between the slopes of the two groups. If two or more nodes change in opposite directions, it is a reverse trend. The trend type (such as consistent, reverse, interrupted) is combined with the expression offset level and assigned a value. The trend factor corresponding to consistent trend + low offset is 0.9, and the factor corresponding to reverse trend + medium offset is 0.6. This patient meets the latter, so the trend factor is set to 0.6. The output is the unified parameter trend factor.

[0100] The recovery period attitude analysis submodule analyzes data related to auxiliary treatment cooperation based on a unified parameter trend factor. By comparing the data distribution of performance in each recovery period, it filters the performance data of cooperation behavior and restrictions to obtain the recovery period acceptance.

[0101] The formula is used to filter performance data for cooperative behaviors and constraints:

[0102] ;

[0103] Calculate the characteristic parameters of the recovery period To determine the acceptance status under each requirement, the recovery period acceptance rate is obtained. This represents the total number of cooperative behaviors and constraints being analyzed. Representing the The weighting of each requirement in the recovery recommendations. Representing the patient on the first The actual performance score for each behavior or restriction. Representing the Recommended standard scores for each restriction or cooperative behavior. The parameters representing the patient's subjective health score are related to the first The psychological state score associated with each item.

[0104] The recovery period cooperation characteristic parameter refers to the comprehensive quantitative parameter reflecting the patient's overall cooperation and acceptance of various behaviors and restrictions during the recovery period. This parameter is obtained by weighting, normalizing, and absolutizing the actual performance scores of patients with various cooperation behaviors and restrictions and the recommended standard scores during the postoperative recovery period, and combining them with the corresponding psychological state rating parameters.

[0105] Based on the standard management checklist for the postoperative recovery period, a total of 8 behaviors and restrictions requiring monitoring were extracted. These include, for example, controlling the duration of postoperative bed rest, avoiding strenuous exercise, abstaining from sexual activity, avoiding lifting heavy objects, adhering to a diet to prevent constipation, limiting daily walking distance, compliance with psychological relaxation training, and frequency of topical medication use. These are numbered sequentially as follows: to Correspondingly set recommended standard scores The criteria were developed and quantitatively assigned according to a standard guidance protocol, with all items set on a 10-point scale. Patient's actual postoperative behavior was collected through interview records and questionnaires, and converted into performance scores. For example, if a patient scores 9 points on "avoiding lifting heavy objects" and 6 points on "daily walking restriction," the system will simultaneously retrieve the psychological state scores related to each behavior item obtained from the initial inquiry and data collection module. For example, in the "Daily Walking Limitation" item, patients reported that anxiety affected their adherence, scoring 3 points. The system assigns weights based on the importance of the behavior. This setting is based on expert evaluation or experience-weighted processing of the degree of behavioral intervention in the recovery suggestion path. For example, the weight is set to 0.2 for the "abstaining from sexual activity" item and 0.05 for the "psychological relaxation training" item. All parameters need to be processed on a unified scale. The actual score, standard score, psychological score and weight are each normalized to the [0, 1] interval before participating in the calculation. "Abstaining from sexual activity" (set as...) For example, if the behavior has a weight Patient's actual performance score Recommended standard score Psychological state score The corresponding normalized value is: , , , Then, the term to be substituted into the calculation is:

[0106] ;

[0107] Continue with the "daily walking limit" (set) For example, the parameters are set as follows: , , , Then the value of this item is:

[0108] ;

[0109] Similarly, after processing the remaining 6 items, let the calculation results for the other items be 0.023, 0, 0.019, 0.03, 0.004, and 0.015, respectively. Then, these 8 data items will be used in the overall formula calculation:

[0110] ;

[0111] This result indicates that the recovery period is related to characteristic parameters. Located at the lowest segment of the recovery period behavioral cooperation assessment interval, this parameter, according to clinical assessment standards, is numerically lower than the upper limit of the system's baseline interval of 0.05, falling within the stable cooperation interval of [0, 0.05]. This indicates that the patient's adherence to various behavioral norms and restrictions after surgery has minimal deviation, with a highly consistent level of cooperation. Furthermore, the influence of psychological state on cooperation performance is within an acceptable range. The system's assessment of the patient's cooperation acceptance level tends towards the "good" category. Subsequently, the system will connect this parameter with the acceptance category determination logic, according to the interval boundaries [0, 0.05], [0.05, 0.1], [0.1, 0.2], [0.2... [1] Corresponding to four categories of acceptance labels: “good”, “moderate”, “fluctuating”, and “unstable”, respectively. Since the patient’s cooperation characteristic parameters are in the “good” range, the system automatically labels their acceptance level as “high cooperation”. The formula uses a triple structure of behavioral weight, behavioral deviation, and psychological score to participate in the calculation. It adopts mathematical operations such as addition and subtraction of differences, multiplication of weights, square root normalization, absolute value stabilization, and summation and averaging to construct a multi-dimensional quantitative reflection system of the recovery of behavioral cooperation. This makes the original behavioral data comparable after structured processing, forming a unified indicator that meets the system’s requirements for use in subsequent decision-making suggestions and attribution label generation modules.

[0112] The risk tolerance performance submodule analyzes the combination relationship of characteristic parameters based on the acceptance level during the recovery period, determines the label distribution of each combination type, optimizes the label classification and sorting, and establishes a hierarchical structure for label classification to obtain the feature category label.

[0113] The patient's acceptance scores for various postoperative inconveniences (such as vaginal bleeding, bed rest, and topical medications) in the recovery period questions were read. Each item was assigned a score range of 0 to 1, with a full score indicating complete acceptance and 0 indicating complete rejection. For example, a patient scored 0.8 on "reduced activity 3 months post-surgery," 0.4 on "long-term estrogen use," 0.3 on "risk of mesh exposure," and 0.6 on "postoperative pain." All scores were categorized into the "recovery period acceptance parameter group." This parameter group was ranked, and its mean and variability were calculated. Then, a risk tolerance performance parameter was introduced as the second input, including the patient's acceptance of the risks of postoperative complications such as infection, intraoperative bleeding, and nerve damage. Each item was also standardized to a 0-1 range. For example, a patient's scores in this group were 0.5 for infection, 0.7 for bleeding, 0.6 for complications, and 0.3 for sexual dysfunction, with an average of 0.525. The means of the two parameter groups were then compared side-by-side. A two-dimensional combination vector is constructed. For example, the mean acceptance during the recovery period is 0.525, and the mean risk tolerance is 0.525, which is considered a balanced level. A score above 0.7 is defined as high acceptance, 0.4-0.7 as medium acceptance, and below 0.4 as low acceptance. In this example, both groups have a mean of medium acceptance. The extreme value difference in each parameter group is compared. For example, the maximum value of the parameter during the recovery period is 0.8, the minimum value is 0.3, and the range is 0.5, which is judged as "large expression difference". If the range is less than 0.2, it is defined as "concentrated expression". Then, the expression difference, mean level, and extreme value range of the two groups are combined into a three-dimensional feature. The grouping logic table is called for matching. If there is a combination of high acceptance + low risk + concentrated expression, it is marked as "low anxiety type". If it is medium acceptance + high risk + large expression difference, it is "anxious and sensitive type". If it is medium acceptance + medium risk + medium difference, it is "neutral response type". This example meets the last combination and is classified as "neutral response type".

[0114] like Figure 2 and Figure 7 As shown, the decision communication module includes:

[0115] The surgical path analysis submodule, based on feature-assigned category labels, determines the parameters involved when doctors recommend surgical paths for each label, compares the correspondence between each label and surgical path parameters, identifies the association between labels and paths, and obtains the surgical path pairing quantity.

[0116] The system retrieves the patient feature vector sets corresponding to each label from the label numbering library. For example, label T1, "High Knowledge, High Need," includes feature dimensions of knowledge mastery (0.9), risk acceptance (0.8), and recovery acceptance (0.7), while label T3, "Neutral Response," shows 0.6, 0.5, and 0.6. Then, it retrieves the recommended surgical path parameter set by the doctor for each label. For example, the T1 label is associated with the path "pelvic floor mesh reconstruction surgery," and its associated parameters include three items: "acceptable long-term foreign body implantation," "high expected results," and "willingness to tolerate long postoperative checkups and recovery periods." The parameter vector under each label is compared item by item with the doctor's recommended path parameters. The difference between each dimension of the parameter in the label and each surgical feature requirement in the path is calculated and then... The matching degree is defined by setting rules. A difference of less than 0.15 is considered a strong match, 0.15 to 0.3 is considered a medium match, and greater than 0.3 is considered a weak match. For example, the "risk tolerance" of T1 is 0.8. If the required value for this value in the path parameters is 0.7, the difference is 0.1, which is considered a strong match. If the matching results of the three parameters under this label with the path are strong, medium, and strong, then the overall matching level is "strong match". The matching level between each label and path pair is marked and the number of matches is recorded. Two out of the three parameters in the current label and path pairing are strong matches, so the pairing quantity is 2. The pairing quantity is set as the sum of the matching items divided by the total number of parameters. The surgical path pairing quantity between this label and this path is 0.67. The pairing quantities of all labels with multiple paths are summarized to form a data sequence.

[0117] The parameter matching and filtering submodule filters the matching data of patient ID tags and doctor suggested parameters based on the surgical path matching volume, judges the structural combination of tag parameters and suggested content, analyzes the distribution of consistent content in the matching process, and obtains the tag suggestion fit degree.

[0118] Extract the patient's assigned label information from their patient ID and retrieve the surgical recommendation parameters recorded by the doctor in the consultation log. For example, patient A is assigned to T3 "Neutral Responsive Type," and the doctor's recommended parameters are "Physical therapy is needed during recovery," "Uterus can be considered for preservation," and "Wire mesh implantation is not recommended." Compare these parameters with those already included in the T3 label, such as "Acceptance during recovery is 0.6," "No strong inclination to preserve the uterus," and "Acceptance of foreign body implantation is 0.4." Structure the patient's label parameters with the doctor's recommendations, comparing each recommendation parameter with the direction and intensity of the intention expressed in the label. For example, if the doctor's recommendation to "preserve the uterus" matches the patient's label "No clear intention," the matching level is defined as "acceptable." Match types are set as completely consistent, acceptable, and inconsistent. Each comparison records the occurrence of a matching type. Frequency: For example, if a patient currently compares 6 suggested parameters, with 3 completely consistent, 2 acceptable, and 1 inconsistent, the consistency rate is 50%. A consistency rate greater than 70% is considered high fit, 50%–70% is medium fit, and less than 50% is low fit. This patient's data is classified as medium fit. All matching positions between suggested parameters and the labels are numbered and their distribution is statistically analyzed. If the matching items are mainly concentrated in the two dimensions of risk preference and recovery period requirements, but less so in the treatment goal dimension, the fit is labeled as "structurally biased"; otherwise, it is labeled as "structurally balanced." The current matching items are "recovery acceptance," "foreign body rejection attitude," and "attitude towards uterine preservation," evenly distributed across the three dimensions, thus classified as structurally balanced. The output label suggestion fit is medium fit and structurally balanced.

[0119] The information consistency determination submodule compares the information structure of the patient's expression and the doctor's recommended content based on the tag suggestion fit, optimizes the process sequence of each step in the matching process, adjusts the operation standards of the determination step, integrates information matching performance, and obtains doctor-patient matching degree data.

[0120] The original subjective statements expressed by patients in all interview questionnaires were extracted, including five categories: treatment expectations, acceptable risks, recovery methods, and organ preservation preferences. Then, the suggestions recorded by doctors were extracted by field analysis. Key content items mentioned in the two text segments were structurally matched. For example, if a patient stated, "I hope to retain the uterus, but if the doctor feels it's okay not to," and the doctor suggested "Uterus preservation is preferred," this group was identified as a structurally consistent item. If a patient stated, "I cannot accept prolonged postoperative bleeding," and the doctor suggested, "Local hormone therapy is needed for 1-2 months postoperatively," this was identified as a partially conflicting item. All structural items were matched and scored. Each item with complete consistency was assigned 1 point, partial conflict 0.5 points, and complete conflict 0 points. If a total of 12 items were identified, with 7 consistent, 3 partially conflicting, and 2 completely conflicting, the score would be 7 + 1.5, totaling 8.5. The consistency ratio is 8.5 divided by 12, resulting in 0.71. High consistency is defined as above 0.75, and medium consistency as 0. The consistency score is 0.5 to 0.75, with low consistency being less than 0.5. Currently, the consistency level is medium. Next, the process of information extraction and comparison steps during matching is reviewed to see if there is a reversal in the execution order, resulting in some fields not being compared. If it is detected that the doctor's suggestion includes filling in the treatment path first and then recording the preference explanation paragraph, while the patient's expression order is to express intention first and then raise specific concerns, this is marked as a inconsistent process order. The process is adjusted to: first classify structural items, then sort and normalize, then uniformly number them before matching. Subsequently, all field mapping errors in the matching steps are corrected, and the judgment threshold setting is calibrated to ensure it covers ambiguous expressions. For example, "I think it's okay" is mapped to "acceptable" instead of "completely agree." Finally, all comparison scores are re-summarized, and after confirming there are no omissions or misjudged items, the overall consistency score is output. Simultaneously, the information matching performance is integrated with the structural classification results of the label suggestion fit. Finally, the doctor-patient matching score is calculated to be 0.71, corresponding to a "medium matching level." This result serves as one of the core parameters for the subsequent output of doctor-patient joint decision support results.

[0121] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A preoperative doctor-patient collaborative decision-making assessment system for pelvic organ prolapse diseases based on artificial intelligence, characterized in that, The system includes: The initial inquiry and data collection module analyzes the patients' symptom distress content in the preoperative interview, interprets the expression of the impact on quality of life, analyzes the psychological state, compares the proportion of symptoms and life impact, adjusts the parameter weights in combination with subjective narrative content, judges the health status, and obtains subjective health score parameters. The communication behavior and expression assessment module, based on the subjective health score parameters, determines the frequency of proactive expression, analyzes questioning and feedback performance, selects key issues in the communication process, compares the relationship between the number of proactive expressions and feedback, analyzes initiative, and obtains a communication initiative index. The cognitive testing module analyzes the understanding of basic knowledge answers based on the aforementioned communication initiative index, screens the performance of answers to treatment methods, judges the mastery of postoperative risks and recovery priorities, compares the differences in answer performance, and statistically analyzes high-frequency knowledge blind spots to obtain characteristics of cognitive weaknesses. Based on the cognitive weakness features, the patient feature analysis module compares the normalized performance of various symptoms and needs, analyzes the acceptance attitude during the postoperative recovery period, screens risk tolerance parameters, judges parameter combinations, adjusts the label classification order, and obtains the feature category label. Based on the feature-assigned category labels, the decision-making and communication module analyzes the doctor's surgical path parameters, determines the correspondence between the patient's labels and decision suggestions, filters the matching statistical results, compares the consistency of information, integrates the doctor-patient joint decision-making adaptation, and obtains doctor-patient matching degree data.

2. The preoperative doctor-patient collaborative decision-making assessment system for pelvic organ prolapse disease based on artificial intelligence as described in claim 1, characterized in that, The subjective health scoring parameters include physical condition score, daily activity ability score, and psychological state score. The communication initiative indicators include information expression activity, interaction frequency, and proactive communication attitude. The cognitive weakness characteristics include knowledge blind spots, risk perception misconceptions, and missing treatment information. The feature category labels include hierarchical classification number, recovery period attitude identifier, and risk acceptance category. The doctor-patient matching data includes consistency in treatment plan selection, label association level, and effectiveness of joint doctor-patient decision-making.

3. The preoperative doctor-patient collaborative decision-making assessment system for pelvic organ prolapse diseases based on artificial intelligence as described in claim 1, characterized in that, The initial query data collection module includes: The symptom distress assessment submodule analyzes the symptom distress content described by patients in preoperative interviews, compares the degree of influence of each symptom in subjective narratives, and determines... The severity level of all symptoms is used to obtain a symptom impact balance factor. The life impact analysis submodule compares the symptom impact balance factor with the patient's subjective expression of the degree of limitation in daily life, work, social life and exercise, analyzes the changes in the proportion of the two types of impact in the expression, screens the most prominent manifestation of life limitation, and obtains the life impact proportion parameter. The psychological state assessment submodule optimizes the weighting relationship between psychological state and subjective life narratives based on the aforementioned life impact ratio parameter, combined with verbal and nonverbal expressions of psychological state during interviews, calculates and summarizes the performance of various influencing factors, and obtains subjective health score parameters.

4. The preoperative doctor-patient collaborative decision-making assessment system for pelvic organ prolapse diseases based on artificial intelligence as described in claim 1, characterized in that, The communication behavior and expression assessment module includes: The expression frequency identification submodule analyzes the language content spontaneously expressed by patients during the interview based on the subjective health score parameters, identifies statements that actively express needs and raise questions in each language segment, judges the continuity of the expression content and the fluency of the language, compares the occurrence ratio of different expression categories in the interview, and obtains the active communication activity level. The interactive behavior extraction submodule analyzes the patient's questions and feedback during the consultation process based on the active communication activity, filters the number of questions and feedback details involved in the interactive language, compares the relationship between interactive behavior and active expression, and obtains the interactive response coupling coefficient. Based on the interaction response coupling coefficient, the initiative assessment submodule analyzes the dominant characteristics of various expression methods in the interview content, determines the distribution among emotional expression, questioning language and declarative expression, adjusts the weight ratio of different expression types in the initiative assessment, and obtains the communication initiative index.

5. The preoperative doctor-patient collaborative decision-making assessment system for pelvic organ prolapse diseases based on artificial intelligence as described in claim 1, characterized in that, The cognitive testing module includes: Based on the aforementioned communication initiative index, the disease understanding and identification submodule analyzes the patient's answers to basic knowledge questions about pelvic organ prolapse during the knowledge Q&A session, identifies the basic knowledge and statements for each question, compares the correspondence between the answer selection and the actual understanding, judges the consistency between the question answer and the patient's description, and obtains the basic knowledge mastery rate. The treatment mastery screening submodule, based on the aforementioned basic knowledge mastery rate, screens patients' answers to questions related to various treatment methods and postoperative recovery, analyzes the mastery performance of answers under treatment plan questions, judges the correspondence between answer selection and key knowledge points, and obtains the degree of understanding of treatment points. The knowledge blind spot statistics submodule, based on the understanding of the treatment points, filters out frequently erroneous knowledge content during the question-and-answer process, counts the locations of repeatedly erroneous knowledge points in the answers, compares the distribution of frequently erroneous questions, and obtains the characteristics of cognitive weaknesses.

6. The preoperative doctor-patient collaborative decision-making assessment system for pelvic organ prolapse disease based on artificial intelligence as described in claim 1, characterized in that, The frequency of spontaneous expression refers to the frequency with which patients spontaneously express their needs, ideas, questions, and demands during the communication process. The initiative refers to the depth of participation, enthusiasm for expression, speed and quality of information feedback demonstrated by patients in doctor-patient interactions.

7. The preoperative doctor-patient collaborative decision-making assessment system for pelvic organ prolapse diseases based on artificial intelligence as described in claim 1, characterized in that, The patient feature analysis module includes: The normalized performance comparison submodule, based on the cognitive weakness characteristics, converts each parameter into a standardized data set, compares the proportion of each parameter in the set, filters the dominant parameters, and adjusts the parameter sorting order to obtain a unified parameter trend factor. The recovery period attitude analysis submodule analyzes the data related to the auxiliary treatment based on the unified parameter trend factor. By comparing the data distribution of each recovery period, it filters out the recovery period characteristics with correlation and obtains the recovery period acceptance. The risk tolerance performance submodule analyzes the combination relationship of characteristic parameters based on the recovery period acceptance, determines the label distribution of each combination type, optimizes the label classification and sorting, and establishes a hierarchical structure of label classification to obtain the feature belonging category label.

8. The preoperative doctor-patient collaborative decision-making assessment system for pelvic organ prolapse diseases based on artificial intelligence as described in claim 1, characterized in that, The basic knowledge comprehension refers to the patient's correct answering performance on various types of questions about pelvic organ prolapse in a knowledge test. The answering performance refers to the patient's correctness of the answer, the logic of the response, and whether the patient has grasped the core information in the knowledge question and answer.

9. The preoperative doctor-patient collaborative decision-making assessment system for pelvic organ prolapse diseases based on artificial intelligence according to claim 1, characterized in that, The decision communication module includes: The surgical path analysis submodule, based on the feature category labels, determines the parameters involved when doctors recommend surgical paths for each label, compares the correspondence between each label and the surgical path parameters, identifies the association between labels and paths, and obtains the surgical path pairing quantity. The parameter matching and filtering submodule filters the matching data of patient ID tags and doctor suggested parameters based on the surgical path matching volume, judges the structural combination of tag parameters and suggested content, analyzes the distribution of consistent content during the matching process, and obtains the tag suggestion fit degree. The information consistency determination submodule compares the information structure of the patient's expression and the doctor's recommended content based on the suggested fit of the tags, optimizes the process sequence of each step in the matching process, adjusts the operation standards of the determination step, integrates the information matching performance, and obtains the doctor-patient matching degree data.

Images