Doctor-patient communication effect intelligent evaluation method based on multi-modal analysis

Abstract

The invention relates to the field of communication effect evaluation, and discloses a multi-modal analysis-based intelligent evaluation method for a doctor-patient communication effect, and the method comprises the following steps: constructing a standardized virtual patient scene library model, simulating a communication scene between a specialized training student and a patient, collecting conversation associated data, preprocessing the associated data, and obtaining a communication result; facial image feature information and voice information of the specialized training students are generated, so that multi-dimensional fusion analysis is carried out, the simulated interactive dialogue effect of the specialized training students is evaluated, and finally an evaluation report is generated. According to the invention, the artificial intelligence technology is utilized, the communication effect of specialized training students on patients is evaluated scientifically and standardly, and key technical support is provided for realizing high-quality and high-efficiency medical services.

Description

Technical Field

[0001] This invention relates to the field of communication effectiveness evaluation, and in particular to an intelligent evaluation method for doctor-patient communication effectiveness based on multimodal analysis. Background Technology

[0002] Doctor-patient communication is not simply a chat; it is a core component of modern healthcare, its importance manifesting in several ways. Up to 70% of clinical diagnoses rely on the patient's medical history. Effective communication is crucial for doctors to obtain accurate and comprehensive information about the patient's condition, directly impacting the accuracy and timeliness of diagnosis. The modern medical model has shifted from a "paternalistic" to a "collaborative" approach. Good communication ensures patients fully understand their condition, treatment plan, risks, and prognosis, enabling them to collaborate with doctors to make medical decisions that best align with their values ​​and wishes. Illness is often accompanied by fear and uncertainty. A doctor's empathy, patient explanations, and encouragement can greatly alleviate the psychological stress of patients and their families, building a strong doctor-patient trust relationship. Therefore, doctor-patient communication is a core clinical skill, as important as the medical technology itself.

[0003] Traditional doctor-patient communication training suffers from pain points such as high cost, difficulty in standardization, subjective feedback, and difficulty in scaling. This method addresses these pain points with revolutionary advantages and profound significance. Through multimodal analysis, the assessment dimensions are transformed from unimodal to multimodal, feedback information becomes more precise, and training scenarios can be scaled. Simultaneously, the system can record all historical data of trainees, personalizedly pinpoint their weaknesses, and recommend customized training programs, achieving "personalized instruction." This method utilizes artificial intelligence technology to transform the crucial "art" of doctor-patient communication into a standardized, trainable, and assessable "science," providing key technical support and solutions for achieving high-quality, efficient, and compassionate healthcare services. Summary of the Invention

[0004] This invention overcomes the shortcomings of existing technologies and provides an intelligent evaluation method for the effectiveness of doctor-patient communication based on multimodal analysis.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: The first aspect of this invention provides an intelligent evaluation method for the effectiveness of doctor-patient communication based on multimodal analysis, comprising the following steps: By combining large language models and patient history cases, a standardized virtual patient scenario library model was constructed and trained. Based on a standardized virtual patient scenario library model, all related data during the dialogue between the standardized virtual patient and the specialized training student were collected synchronously, and the related data were preprocessed to obtain the facial image feature information of the specialized training student and the preprocessed voice information of the specialized training student. Multi-dimensional fusion analysis was performed on the facial image feature information and voice information of preprocessed specialized training students. Based on the analysis results, the simulated interactive dialogue effect of specialized training students was evaluated in different dimensions. Based on the evaluation results of simulated interactive dialogues of specialized training students, a structured training evaluation report for specialized training students is generated. Speech prosody and emotion analysis was performed on the preprocessed speech information of students in specialized training programs, and the structured training evaluation reports of these students were updated based on the analysis results.

[0006] Furthermore, in a preferred embodiment of the present invention, the step of constructing and training a standardized virtual patient scenario library model by combining a large language model and patient historical cases specifically includes: Obtain information from the hospital information system and pre-determine the types of scenarios that require doctor-patient communication, and mark them as target doctor-patient communication scenario types. The target doctor-patient communication scenario types include typical medical history, standard personality type, gender, occupation, communication venue, and communication personnel. Within the hospital information system, based on the target doctor-patient communication scenario type, the corresponding electronic medical record is determined, key case information is extracted from the electronic medical record, and a case knowledge graph is constructed. The key case information includes the patient's medical history, examination results, and patient requests information corresponding to the electronic medical record. The case knowledge graph records the key case information of electronic medical records corresponding to all target doctor-patient communication scenario types. An LLM (Large Language Model) is introduced, and a standardized virtual patient role profile is created within the LLM based on the case knowledge graph. The standardized virtual patient role profile describes the key case information of different electronic medical records in the case knowledge graph. At the same time, the personality traits of the standardized virtual patient role profile are preset according to the patient's personality standard type. Based on the standardized virtual patient role profile, the LLM large language foundation model is subjected to model constraint adjustment training to obtain a standardized virtual patient scenario library model. The standardized virtual patient scenario library model can simulate the dialogue between the patient and the specialized training student and collect all related data in real time during the dialogue process. The specialized training student is the user using the standardized virtual patient scenario library model.

[0007] Furthermore, in a preferred embodiment of the present invention, the step of simultaneously collecting all relevant data during the dialogue between the standardized virtual patient and the specialized training student based on the standardized virtual patient scene library model, and preprocessing the relevant data to obtain the facial image feature information and preprocessed voice information of the specialized training student, specifically involves: Acquire a specialized training student interactive device, and integrate a standardized virtual patient scenario library model into the specialized training student interactive device, wherein the specialized training student interactive device enables specialized training students to conduct simulated interactive dialogue with standardized virtual patients; The interactive device for specialized training students was activated, and facial images and voice information of the specialized training students were collected in real time during the simulated interactive dialogue between the specialized training students and standardized virtual patients. During the data collection process, facial images and voice information of students undergoing specialized training were subjected to spatiotemporal synchronization matching. The data spatiotemporal synchronization matching processing method is as follows: a synchronization signal generator is set in the interactive device for specialized training students. When it is necessary to collect facial images and voice information of specialized training students, the synchronization signal generator sends a unified start pulse signal to different data acquisition devices in the interactive device for specialized training students. All data acquisition devices take the unified start pulse signal as the absolute time starting point and generate a recording timestamp to achieve data spatiotemporal synchronization matching. The facial images of students in specialized training programs are processed with standard pose alignment and illumination normalization. At the same time, the speech information of students in specialized training programs is separated by combining spectral subtraction. The speech information of students in specialized training programs is then processed by short-time frame segmentation and windowing based on the Hamming window function method, so as to obtain the preprocessed facial image feature information and the preprocessed speech information of students in specialized training programs.

[0008] Furthermore, in a preferred embodiment of the present invention, the step of performing multi-dimensional fusion analysis on the preprocessed facial image feature information and preprocessed voice information of specialized training students, and evaluating the simulated interactive dialogue effect of specialized training students in different dimensions based on the analysis results, specifically includes: A big data network is introduced to retrieve the hesitant word knowledge graph of the preprocessed specialized training students' speech information. Based on the hesitant word knowledge graph, the speech information of the preprocessed specialized training students is analyzed, and the speech information sentences containing hesitant words are extracted and labeled as hesitant word speech information sentences. All speech information sentences with hesitant words are analyzed, the number of hesitant words and their duration are calculated, and speech information sentences with hesitant words whose number of hesitant words or duration is greater than the corresponding preset value are selected and marked as serious hesitant sentences. A key information knowledge graph for target doctor-patient communication scenarios is constructed to perform speech logic analysis on pre-processed speech information of specialized training students. Based on the results of speech logic analysis and severely hesitant sentences, a preliminary assessment of the simulated interactive dialogue effect of specialized training students is conducted.

[0009] Furthermore, in a preferred embodiment of the present invention, the construction of a key information knowledge graph of the target doctor-patient communication scenario type is used to perform speech logic analysis on the pre-processed speech information of specialized training students, and, in combination with the speech logic analysis results and severely hesitant statements, to conduct a preliminary evaluation of the simulated interactive dialogue effect of the specialized training students, specifically as follows: Based on big data networks, key information knowledge graphs of target doctor-patient communication scenarios are retrieved. Combined with key information knowledge graphs, the voice information of students in preprocessing specialized training is analyzed to locate the corresponding key information in the voice information of students in different preprocessing specialized training. A relation extraction model is introduced to perform deep semantic logical association analysis on the key information corresponding to all preprocessed specialized training students' speech information, and to determine whether there is a logical association between the key information corresponding to different preprocessed specialized training students' speech information. If there is no logical connection between the key information in the preprocessed specialized training student speech information, then the corresponding sentences in the preprocessed specialized training student speech information will be marked as sentences without logical connection. A communication scoring terminal for specialized training students is pre-set in the interactive device for specialized training students. Within the communication scoring terminal for specialized training students, statements with serious hesitation and statements without logical connection are analyzed. If the proportion of unrelated statements in all preprocessed specialized training students’ voice information is less than the corresponding predetermined value, or the number of seriously hesitant statements is greater than the corresponding predetermined value, then the preliminary score of the specialized training students in the specialized training students’ communication scoring terminal will be deemed as the preliminary unqualified simulated interactive dialogue effect. If the proportion of illogical statements in all preprocessed specialized training students’ voice information is less than the corresponding predetermined value, and the number of seriously hesitant statements is greater than the corresponding preset value, then the preliminary score of the specialized training students in the specialized training students’ communication scoring terminal is initially qualified for the simulated interactive dialogue effect. The facial image feature information of students in the preprocessed specialized training program was analyzed, and the simulated interactive dialogue effect of the students was evaluated based on the analysis results.

[0010] Furthermore, in a preferred embodiment of the present invention, the step of analyzing the facial image feature information of preprocessed specialized training students and conducting a secondary scoring of the simulated interactive dialogue effect of the specialized training students based on the analysis results specifically includes: Based on the preprocessed speech information of specialized training students, the preprocessed facial image feature information of specialized training students is processed by sentence segmentation to ensure that different preprocessed speech information sentences correspond to different preprocessed facial image feature information of specialized training students, thus obtaining different preprocessed facial image feature information sub-segments. In the preprocessing of facial image feature information sub-segments of specialized training students, specific pixels on the faces of specialized training students are encoded, and the intensity change status and corresponding duration of different characteristic pixels after encoding are recorded in real time. Among them, specific pixels include the pixels at the corners of the mouth and the pixels at the eyebrows and eyes of students in specialized training. Based on the intensity change state and corresponding duration of different characteristic pixels after encoding, the gaze direction vector and blink probability of specialized training students are calculated. At the same time, the attention area is obtained in the standardized virtual patient scene library model. The attention area is the standard gaze direction area of ​​specialized training students when they conduct simulated interactive dialogue with standardized virtual patients. The students in the specialized training program were given a second score through the communication and scoring terminal, and their gaze direction vector and blink probability were analyzed. The duration of the gaze direction vector of the specialized training student within the attention zone is judged. If the simulated interactive dialogue effect is initially unqualified, and the duration of the gaze direction vector of the specialized training student within the attention zone is less than the predetermined duration, or the blinking probability of the specialized training student is greater than the predetermined standard value, then the secondary score will be judged as unqualified for the simulated interactive dialogue effect. If the simulated interactive dialogue effect is initially satisfactory, and the duration of the specialized training student's gaze direction vector within the attention zone is less than the predetermined duration, or the specialized training student's blinking probability is greater than the predetermined standard value, then the second evaluation will be assessed as the simulated interactive dialogue effect needing improvement. If the simulated interactive dialogue effect is initially satisfactory, and the duration of the specialized training student's gaze direction vector within the attention zone is not less than the predetermined duration, and the specialized training student's blinking probability is not greater than the predetermined standard value, then the secondary evaluation will be deemed satisfactory for the simulated interactive dialogue effect.

[0011] Furthermore, in a preferred embodiment of the present invention, the step of generating a structured training evaluation report for specialized training students by combining the evaluation results of simulated interactive dialogues with the students' performance in specialized training specifically includes: Based on the simulated interactive dialogue effects already assessed by specialized training students, a feedback report is generated within the specialized training student communication scoring terminal and labeled as a structured training evaluation report for specialized training students. The structured training evaluation report for specialized training students marks the simulated interactive dialogue effect of the specialized training students and scores different simulated interactive dialogue effects within a range of scores. Meanwhile, the structured training evaluation report for specialized training students generates a real-time change curve of the simulated interactive dialogue effect. The real-time change curve of the simulated interactive dialogue effect records the simulated interactive dialogue effect of specialized training students at different time points during the training process, and marks the time points when the simulated interactive dialogue effect is unsatisfactory and needs to be optimized.

[0012] Furthermore, in a preferred embodiment of the present invention, the step of performing prosodic and affective analysis on the preprocessed speech information of specialized training students, and updating the structured training evaluation report of specialized training students based on the analysis results, specifically includes: Acoustic features are extracted from each frame of the preprocessed specialized training students' speech information. The acoustic features include fundamental frequency coefficients, Mel frequency cepstral coefficients, speech rate, and short-time energy. The range of combinations of different acoustic features under standard conditions is preset and calibrated as the standard acoustic feature range; The acoustic features of preprocessed specialized training student speech information in different frames are combined and analyzed. If, after combining the acoustic features of preprocessed specialized training student speech information, the acoustic feature range of a frame does not remain within the standard acoustic feature range, then the preprocessed specialized training student speech information of the corresponding frame is marked as an abnormal speech information frame. When the proportion of abnormal voice information frames in all frames is greater than the preset proportion, the structured training evaluation report of specialized training students will be updated. The updating of the structured training evaluation report for specialized training students is as follows: if the simulated interactive dialogue effect in the structured training evaluation report for specialized training students is unsatisfactory, then the simulated interactive dialogue effect will continue to be marked as unsatisfactory. If the simulated interactive dialogue effect is deemed to need optimization, then the simulated interactive dialogue effect will continue to be labeled as needing optimization. If the simulated interactive dialogue effect is satisfactory, then the simulated interactive dialogue effect will be marked as needing optimization.

[0013] A second aspect of this invention also provides an intelligent evaluation system for doctor-patient communication effectiveness based on multimodal analysis. This intelligent evaluation system integrates a high-performance computing architecture and a data storage module, including a non-volatile memory consisting of a DDR4 RDIMM memory module with ECC verification and an NVMe solid-state storage array using 3D NAND flash memory, and a multi-core processor based on the Zen4 microarchitecture. The memory contains a program for an intelligent evaluation method of doctor-patient communication effectiveness with an intelligent evaluation system engine. When the program is executed in parallel through a superscalar pipeline execution unit within the processor, the following steps are implemented: By combining large language models and patient history cases, a standardized virtual patient scenario library model was constructed and trained. Based on a standardized virtual patient scenario library model, all related data during the dialogue between the standardized virtual patient and the specialized training student were collected synchronously, and the related data were preprocessed to obtain the facial image feature information of the specialized training student and the preprocessed voice information of the specialized training student. Multi-dimensional fusion analysis was performed on the facial image feature information and voice information of preprocessed specialized training students. Based on the analysis results, the simulated interactive dialogue effect of specialized training students was evaluated in different dimensions. Based on the evaluation results of simulated interactive dialogues of specialized training students, a structured training evaluation report for specialized training students is generated. Speech prosody and emotion analysis was performed on the preprocessed speech information of students in specialized training programs, and the structured training evaluation reports of these students were updated based on the analysis results.

[0014] This invention addresses the technical deficiencies in the existing technology and offers the following beneficial effects: By constructing a standardized virtual patient scenario library model to simulate communication scenarios between specialized training students and patients and collecting dialogue-related data, this invention preprocesses the related data to generate facial image feature information and voice information of the specialized training students. This data is then used for multi-dimensional fusion analysis to evaluate the simulated interactive dialogue effectiveness of the specialized training students, ultimately generating an evaluation report. This invention utilizes artificial intelligence technology to scientifically and systematically evaluate the communication effectiveness of specialized training students with patients, providing key technical support for achieving high-quality and efficient medical services. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other embodiments can be obtained from these drawings without creative effort.

[0016] Figure 1 A flowchart of an intelligent evaluation method for doctor-patient communication effectiveness based on multimodal analysis is shown. Figure 2 A flowchart illustrating the method for evaluating the effectiveness of simulated interactive dialogues on specialized training students from different dimensions is provided. Figure 3 The program view of an intelligent evaluation system for doctor-patient communication effectiveness based on multimodal analysis is shown. Detailed Implementation

[0017] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0018] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0019] Figure 1 A flowchart illustrating an intelligent evaluation method for doctor-patient communication effectiveness based on multimodal analysis is provided, including the following steps: S102: Combine large language models and patient history cases to build and train a standardized virtual patient scenario library model; S104: Based on the standardized virtual patient scenario library model, all related data in the dialogue process between the standardized virtual patient and the specialized training student are collected synchronously, and the related data are preprocessed to obtain the facial image feature information of the specialized training student and the preprocessed voice information of the specialized training student. S106: Perform multi-dimensional fusion analysis on the facial image feature information and voice information of preprocessed specialized training students, and evaluate the simulated interactive dialogue effect of specialized training students in different dimensions based on the analysis results. S108: Based on the evaluation results of the simulated interactive dialogue effect of specialized training students, generate a structured training evaluation report for specialized training students. S110: Perform prosodic and affective analysis on the preprocessed speech information of specialized training students, and update the structured training evaluation report of specialized training students based on the analysis results.

[0020] Furthermore, in a preferred embodiment of the present invention, the step of constructing and training a standardized virtual patient scenario library model by combining a large language model and patient historical cases specifically includes: Obtain information from the hospital information system and pre-determine the types of scenarios that require doctor-patient communication, and mark them as target doctor-patient communication scenario types. The target doctor-patient communication scenario types include typical medical history, standard personality type, gender, occupation, communication venue, and communication personnel. Within the hospital information system, based on the target doctor-patient communication scenario type, the corresponding electronic medical record is determined, key case information is extracted from the electronic medical record, and a case knowledge graph is constructed. The key case information includes the patient's medical history, examination results, and patient requests information corresponding to the electronic medical record. The case knowledge graph records the key case information of electronic medical records corresponding to all target doctor-patient communication scenario types. An LLM (Large Language Model) is introduced, and a standardized virtual patient role profile is created within the LLM based on the case knowledge graph. The standardized virtual patient role profile describes the key case information of different electronic medical records in the case knowledge graph. At the same time, the personality traits of the standardized virtual patient role profile are preset according to the patient's personality standard type. Based on the standardized virtual patient role profile, the LLM large language foundation model is subjected to model constraint adjustment training to obtain a standardized virtual patient scenario library model. The standardized virtual patient scenario library model can simulate the dialogue between the patient and the specialized training student and collect all related data in real time during the dialogue process. The specialized training student is the user using the standardized virtual patient scenario library model.

[0021] It's important to note that the first step is to construct a model storing standardized virtual patient communication scenarios—a standardized virtual patient scenario library model. Since intelligent evaluation of doctor-patient communication effectiveness is needed, but cannot be achieved through real-life patient communication (as this is insufficiently intelligent and difficult to assess), standardized virtual patients are constructed for use in specialized training students. First, the types of doctor-patient communication scenarios to be conducted are determined and labeled as target doctor-patient communication scenario types. Different communication scenarios are constructed based on different symptoms. Target doctor-patient communication scenario types include typical medical history, standard personality type, gender, occupation, communication location, and communication personnel. The patient's standard personality type is determined through questionnaires, consultations, and other methods using a personality assessment report. This report is used to set the patient's personality orientation in the subsequent construction of standardized virtual patients. Based on different types, historical cases of that type are identified for use in the construction of standardized virtual patients. The LLM (Large Language Model) is a large language model capable of storing and fusing data for analysis. By integrating a case knowledge graph into the model, standardized virtual patient profiles are improved. Different standardized virtual patient profiles require different communication methods, resulting in varying communication effectiveness among specialized training students. This demonstrates the diversity and comprehensiveness of the method. Intelligent evaluation of communication effectiveness is achieved by collecting relevant data during the dialogue process.

[0022] Furthermore, in a preferred embodiment of the present invention, the step of simultaneously collecting all relevant data during the dialogue between the standardized virtual patient and the specialized training student based on the standardized virtual patient scene library model, and preprocessing the relevant data to obtain the facial image feature information and preprocessed voice information of the specialized training student, specifically involves: Acquire a specialized training student interactive device, and integrate a standardized virtual patient scenario library model into the specialized training student interactive device, wherein the specialized training student interactive device enables specialized training students to conduct simulated interactive dialogue with standardized virtual patients; The interactive device for specialized training students was activated, and facial images and voice information of the specialized training students were collected in real time during the simulated interactive dialogue between the specialized training students and standardized virtual patients. During the data collection process, facial images and voice information of students undergoing specialized training were subjected to spatiotemporal synchronization matching. The data spatiotemporal synchronization matching processing method is as follows: a synchronization signal generator is set in the interactive device for specialized training students. When it is necessary to collect facial images and voice information of specialized training students, the synchronization signal generator sends a unified start pulse signal to different data acquisition devices in the interactive device for specialized training students. All data acquisition devices take the unified start pulse signal as the absolute time starting point and generate a recording timestamp to achieve data spatiotemporal synchronization matching. The facial images of students in specialized training programs are processed with standard pose alignment and illumination normalization. At the same time, the speech information of students in specialized training programs is separated by combining spectral subtraction. The speech information of students in specialized training programs is then processed by short-time frame segmentation and windowing based on the Hamming window function method, so as to obtain the preprocessed facial image feature information and the preprocessed speech information of students in specialized training programs.

[0023] It should be noted that the interactive device for specialized training students enables them to engage in simulated interactive dialogues with standardized virtual patients, and collects and analyzes their facial images and voice information to evaluate the effectiveness of doctor-patient communication. This is typically an immersive device, such as VR glasses. The data acquisition process requires synchronization of image and voice information because different voice messages are correlated with different facial images during the dialogue. Therefore, spatiotemporal synchronization processing is necessary, achieved by simultaneously acquiring data through the reception of pulse signals. In achieving this spatiotemporal synchronization, since different pulse signals are sent synchronously, all receiving devices use a unified start pulse signal as the absolute time starting point and record their own timestamps to achieve synchronized data processing. In other words, the timestamp records the absolute time when the device receives the pulse signal, and the device is synchronized to begin operation based on the timestamp. After data collection, the collected data needs to be preprocessed. For example, facial images need to be aligned to ensure that the facial images are always clear and in a neutral position during the process of students turning their heads in specialized training, so as to facilitate data analysis. Voice information needs to be separated into human voices and noise reduction is achieved through short-time frame segmentation and windowing.

[0024] Furthermore, in a preferred embodiment of the present invention, the step of generating a structured training evaluation report for specialized training students by combining the evaluation results of simulated interactive dialogues with the students' performance in specialized training specifically includes: Based on the simulated interactive dialogue effects already assessed by specialized training students, a feedback report is generated within the specialized training student communication scoring terminal and labeled as a structured training evaluation report for specialized training students. The structured training evaluation report for specialized training students marks the simulated interactive dialogue effect of the specialized training students and scores different simulated interactive dialogue effects within a range of scores. Meanwhile, the structured training evaluation report for specialized training students generates a real-time change curve of the simulated interactive dialogue effect. The real-time change curve of the simulated interactive dialogue effect records the simulated interactive dialogue effect of specialized training students at different time points during the training process, and marks the time points when the simulated interactive dialogue effect is unsatisfactory and needs to be optimized.

[0025] It should be noted that, given the known simulated interactive dialogue effects of specialized training students, constructing a structured training evaluation report for these students is to facilitate their understanding of their training results and the areas where points were deducted during the training process. This means that improvements may be needed in areas such as timing, dialogue style, and facial expressions during communication. Therefore, this report is necessary.

[0026] Furthermore, in a preferred embodiment of the present invention, the step of performing prosodic and affective analysis on the preprocessed speech information of specialized training students, and updating the structured training evaluation report of specialized training students based on the analysis results, specifically includes: Acoustic features are extracted from each frame of the preprocessed specialized training students' speech information. The acoustic features include fundamental frequency coefficients, Mel frequency cepstral coefficients, speech rate, and short-time energy. The range of combinations of different acoustic features under standard conditions is preset and calibrated as the standard acoustic feature range; The acoustic features of preprocessed specialized training student speech information in different frames are combined and analyzed. If, after combining the acoustic features of preprocessed specialized training student speech information, the acoustic feature range of a frame does not remain within the standard acoustic feature range, then the preprocessed specialized training student speech information of the corresponding frame is marked as an abnormal speech information frame. When the proportion of abnormal voice information frames in all frames is greater than the preset proportion, the structured training evaluation report of specialized training students will be updated. The updating of the structured training evaluation report for specialized training students is as follows: if the simulated interactive dialogue effect in the structured training evaluation report for specialized training students is unsatisfactory, then the simulated interactive dialogue effect will continue to be marked as unsatisfactory. If the simulated interactive dialogue effect is deemed to need optimization, then the simulated interactive dialogue effect will continue to be labeled as needing optimization. If the simulated interactive dialogue effect is satisfactory, then the simulated interactive dialogue effect will be marked as needing optimization.

[0027] It should be noted that the prosodic and affective analysis of specialized training students are also important characteristics for expressing the effectiveness of communication between specialized training students and patients. Specifically, the fundamental frequency represents the pitch variation of the specialized training student's speech, the Mel frequency cepstral coefficient represents timbre, and the short-time energy represents volume. Different sets of acoustic features can constitute the prosodic and affective expression of specialized training students. If the combination is not within the standard range, it may result in abnormal timbre due to fear, or increased volume and slowed speech due to emotional instability, thus affecting the effectiveness of doctor-patient communication. Different frames correspond to different acoustic features; if the acoustic features of multiple frames are problematic, it indicates poor doctor-patient communication, and the structured training assessment report of the specialized training student will be updated.

[0028] Figure 2 A flowchart illustrating the method for evaluating the effectiveness of simulated interactive dialogues on specialized training students across different dimensions is provided, including the following steps: S202: Perform multi-dimensional fusion analysis on the facial image feature information and voice information of preprocessed specialized training students, and evaluate the simulated interactive dialogue effect of specialized training students in different dimensions based on the analysis results. S204: Perform multi-dimensional fusion analysis on the facial image feature information and voice information of preprocessed specialized training students, and evaluate the simulated interactive dialogue effect of specialized training students in different dimensions based on the analysis results. S206: Analyze the facial image feature information of students in preprocessed specialized training, and conduct a second scoring of the simulated interactive dialogue effect of students in specialized training based on the analysis results.

[0029] Furthermore, in a preferred embodiment of the present invention, the step of performing multi-dimensional fusion analysis on the preprocessed facial image feature information and preprocessed voice information of specialized training students, and evaluating the simulated interactive dialogue effect of specialized training students in different dimensions based on the analysis results, specifically includes: A big data network is introduced to retrieve the hesitant word knowledge graph of the preprocessed specialized training students' speech information. Based on the hesitant word knowledge graph, the speech information of the preprocessed specialized training students is analyzed, and the speech information sentences containing hesitant words are extracted and labeled as hesitant word speech information sentences. All speech information sentences with hesitant words are analyzed, the number of hesitant words and their duration are calculated, and speech information sentences with hesitant words whose number of hesitant words or duration is greater than the corresponding preset value are selected and marked as serious hesitant sentences. A key information knowledge graph for target doctor-patient communication scenarios is constructed to perform speech logic analysis on pre-processed speech information of specialized training students. Based on the results of speech logic analysis and severely hesitant sentences, a preliminary assessment of the simulated interactive dialogue effect of specialized training students is conducted.

[0030] It should be noted that the knowledge graph of hesitation words in the preprocessed speech information of specialized training students is identified in the big data network. These hesitation words include, but are not limited to, "um..." and "oh..." that appear during dialogue. When specialized training students use these hesitation words, it indicates that they lack a deep understanding of the patient's current condition, or that there are communication barriers, preventing them from clearly and directly explaining the condition to the patient. Statements with hesitation words appearing more than a preset number are marked as severely hesitant statements. A high frequency of severely hesitant statements indicates inadequate professional competence among specialized training students.

[0031] Furthermore, in a preferred embodiment of the present invention, the construction of a key information knowledge graph of the target doctor-patient communication scenario type is used to perform speech logic analysis on the pre-processed speech information of specialized training students, and, in combination with the speech logic analysis results and severely hesitant statements, to conduct a preliminary evaluation of the simulated interactive dialogue effect of the specialized training students, specifically as follows: Based on big data networks, key information knowledge graphs of target doctor-patient communication scenarios are retrieved. Combined with key information knowledge graphs, the voice information of students in preprocessing specialized training is analyzed to locate the corresponding key information in the voice information of students in different preprocessing specialized training. A relation extraction model is introduced to perform deep semantic logical association analysis on the key information corresponding to all preprocessed specialized training students' speech information, and to determine whether there is a logical association between the key information corresponding to different preprocessed specialized training students' speech information. If there is no logical connection between the key information in the preprocessed specialized training student speech information, then the corresponding sentences in the preprocessed specialized training student speech information will be marked as sentences without logical connection. A communication scoring terminal for specialized training students is pre-set in the interactive device for specialized training students. Within the communication scoring terminal for specialized training students, statements with serious hesitation and statements without logical connection are analyzed. If the proportion of unrelated statements in all preprocessed specialized training students’ voice information is less than the corresponding predetermined value, or the number of seriously hesitant statements is greater than the corresponding predetermined value, then the preliminary score of the specialized training students in the specialized training students’ communication scoring terminal will be deemed as the preliminary unqualified simulated interactive dialogue effect. If the proportion of illogical statements in all preprocessed specialized training students’ voice information is less than the corresponding predetermined value, and the number of seriously hesitant statements is greater than the corresponding preset value, then the preliminary score of the specialized training students in the specialized training students’ communication scoring terminal is initially qualified for the simulated interactive dialogue effect. The facial image feature information of students in the preprocessed specialized training program was analyzed, and the simulated interactive dialogue effect of the students was evaluated based on the analysis results.

[0032] It should be noted that the key information knowledge graph retrieved for the target doctor-patient communication scenario type includes, but is not limited to, etiology, treatment plan, and recovery time. A relation extraction model is used to perform deep semantic logical association analysis on the key information corresponding to all pre-processed specialized training students' speech information. This determines whether there is a logical relationship between the corresponding key information in the speech information of different pre-processed specialized training students, i.e., analyzing whether the students' statements have logical coherence. If the logic is flawed, it is judged that the specialized training students have limited expressive ability in communicating with patients, i.e., the communication effect is not high. Statements without logical connection are identified and analyzed in conjunction with statements of significant hesitation to determine the number of statements without logical connection. Based on the number of both, a preliminary judgment is made on the communication effect, determining whether it is initially satisfactory or initially unsatisfactory.

[0033] Furthermore, in a preferred embodiment of the present invention, the step of analyzing the facial image feature information of preprocessed specialized training students and conducting a secondary scoring of the simulated interactive dialogue effect of the specialized training students based on the analysis results specifically includes: Based on the preprocessed speech information of specialized training students, the preprocessed facial image feature information of specialized training students is processed by sentence segmentation to ensure that different preprocessed speech information sentences correspond to different preprocessed facial image feature information of specialized training students, thus obtaining different preprocessed facial image feature information sub-segments. In the preprocessing of facial image feature information sub-segments of specialized training students, specific pixels on the faces of specialized training students are encoded, and the intensity change status and corresponding duration of different characteristic pixels after encoding are recorded in real time. Among them, specific pixels include the pixels at the corners of the mouth and the pixels at the eyebrows and eyes of students in specialized training. Based on the intensity change state and corresponding duration of different characteristic pixels after encoding, the gaze direction vector and blink probability of specialized training students are calculated. At the same time, the attention area is obtained in the standardized virtual patient scene library model. The attention area is the standard gaze direction area of ​​specialized training students when they conduct simulated interactive dialogue with standardized virtual patients. The students in the specialized training program were given a second score through the communication and scoring terminal, and their gaze direction vector and blink probability were analyzed. The duration of the gaze direction vector of the specialized training student within the attention zone is judged. If the simulated interactive dialogue effect is initially unqualified, and the duration of the gaze direction vector of the specialized training student within the attention zone is less than the predetermined duration, or the blinking probability of the specialized training student is greater than the predetermined standard value, then the secondary score will be judged as unqualified for the simulated interactive dialogue effect. If the simulated interactive dialogue effect is initially satisfactory, and the duration of the specialized training student's gaze direction vector within the attention zone is less than the predetermined duration, or the specialized training student's blinking probability is greater than the predetermined standard value, then the second evaluation will be assessed as the simulated interactive dialogue effect needing improvement. If the simulated interactive dialogue effect is initially satisfactory, and the duration of the specialized training student's gaze direction vector within the attention zone is not less than the predetermined duration, and the specialized training student's blinking probability is not greater than the predetermined standard value, then the secondary evaluation will be deemed satisfactory for the simulated interactive dialogue effect.

[0034] It should be noted that facial information analysis is necessary after analyzing the voice information. Facial changes in specialized training students are typically visible between the eyebrows and eyes, and at the corners of the mouth. Therefore, after obtaining different preprocessed facial image feature fragments of specialized training students, pixels at the eyebrows, eyes, and corners of the mouth are marked, and the changes in pixel status are analyzed in real time to determine whether there are changes such as the corners of the mouth being raised or the eyebrows and eyes changing, which is used to analyze the effectiveness of doctor-patient communication. First, the relationship between the specialized training student's gaze direction vector and attention area is analyzed. If the gaze direction is not within the attention area, it proves that the specialized training student is not looking at the patient during communication, which will greatly reduce the effectiveness of communication. Moreover, the higher the blink rate, the less confident the specialized training student is, and the worse the communication response. Based on the duration of the specialized training student's gaze direction vector within the attention area and the blink rate, the simulated interactive dialogue effect is scored a second time to determine whether the specialized training student is qualified or needs to continue to improve their communication skills.

[0035] like Figure 3As shown, the second aspect of the present invention also provides an intelligent evaluation system for the effectiveness of doctor-patient communication based on multimodal analysis. The intelligent evaluation system integrates a high-performance computing architecture and a data storage module, including a non-volatile memory consisting of a DDR4 RDIMM memory module with ECC verification and an NVMe solid-state storage array using 3D NAND flash memory, and a multi-core processor based on the Zen4 microarchitecture. The memory contains a program for intelligent evaluation of the effectiveness of doctor-patient communication, which has an intelligent evaluation system engine. When the program is executed in parallel through a superscalar pipeline execution unit within the processor, the following steps are implemented: By combining large language models and patient history cases, a standardized virtual patient scenario library model was constructed and trained. Based on a standardized virtual patient scenario library model, all related data during the dialogue between the standardized virtual patient and the specialized training student were collected synchronously, and the related data were preprocessed to obtain the facial image feature information of the specialized training student and the preprocessed voice information of the specialized training student. Multi-dimensional fusion analysis was performed on the facial image feature information and voice information of preprocessed specialized training students. Based on the analysis results, the simulated interactive dialogue effect of specialized training students was evaluated in different dimensions. Based on the evaluation results of simulated interactive dialogues of specialized training students, a structured training evaluation report for specialized training students is generated. Speech prosody and emotion analysis was performed on the preprocessed speech information of students in specialized training programs, and the structured training evaluation reports of these students were updated based on the analysis results.

[0036] A third aspect of the present invention provides a computer-readable storage medium, characterized in that the computer-readable storage medium includes a program for intelligent evaluation of the effectiveness of doctor-patient communication, wherein when the program for intelligent evaluation of the effectiveness of doctor-patient communication is executed by a processor, it implements the steps of the intelligent evaluation of the effectiveness of doctor-patient communication as described in any of the preceding claims.

[0037] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0038] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0039] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0040] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0041] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

[0042] The above are merely specific embodiments 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. An intelligent evaluation method for doctor-patient communication effectiveness based on multimodal analysis, characterized in that, Includes the following steps: By combining large language models and patient history cases, a standardized virtual patient scenario library model was constructed and trained. Based on a standardized virtual patient scenario library model, all related data during the dialogue between the standardized virtual patient and the specialized training student were collected synchronously, and the related data were preprocessed to obtain the facial image feature information of the specialized training student and the preprocessed voice information of the specialized training student. Multi-dimensional fusion analysis was performed on the facial image feature information and voice information of preprocessed specialized training students. Based on the analysis results, the simulated interactive dialogue effect of specialized training students was evaluated in different dimensions. Based on the evaluation results of simulated interactive dialogues of specialized training students, a structured training evaluation report for specialized training students is generated. Speech prosody and emotion analysis was performed on the preprocessed speech information of students in specialized training programs, and the structured training evaluation reports of these students were updated based on the analysis results.

2. The intelligent evaluation method for doctor-patient communication effectiveness based on multimodal analysis as described in claim 1, characterized in that, The process of constructing and training a standardized virtual patient scenario library model by combining a large language model with patient history cases is as follows: Obtain information from the hospital information system and pre-determine the types of scenarios that require doctor-patient communication, and mark them as target doctor-patient communication scenario types. The target doctor-patient communication scenario types include typical medical history, standard personality type, gender, occupation, communication venue, and communication personnel. Within the hospital information system, based on the target doctor-patient communication scenario type, the corresponding electronic medical record is determined, key case information is extracted from the electronic medical record, and a case knowledge graph is constructed. The key case information includes the patient's medical history, examination results, and patient requests information corresponding to the electronic medical record. The case knowledge graph records the key case information of electronic medical records corresponding to all target doctor-patient communication scenario types. An LLM (Large Language Model) is introduced, and a standardized virtual patient role profile is created within the LLM based on the case knowledge graph. The standardized virtual patient role profile describes the key case information of different electronic medical records in the case knowledge graph. At the same time, the personality traits of the standardized virtual patient role profile are preset according to the patient's personality standard type. Based on the standardized virtual patient role profile, the LLM large language foundation model is subjected to model constraint adjustment training to obtain a standardized virtual patient scenario library model. The standardized virtual patient scenario library model can simulate the dialogue between the patient and the specialized training student and collect all related data in real time during the dialogue process. The specialized training student is the user using the standardized virtual patient scenario library model.

3. The intelligent evaluation method for doctor-patient communication effectiveness based on multimodal analysis as described in claim 1, characterized in that, The standardized virtual patient scenario library model synchronously collects all relevant data during the dialogue between the standardized virtual patient and the specialized training student, and preprocesses the relevant data to obtain the facial image feature information and preprocessed voice information of the specialized training student. Specifically: Acquire a specialized training student interactive device, and integrate a standardized virtual patient scenario library model into the specialized training student interactive device, wherein the specialized training student interactive device enables specialized training students to conduct simulated interactive dialogue with standardized virtual patients; The interactive device for specialized training students was activated, and facial images and voice information of the specialized training students were collected in real time during the simulated interactive dialogue between the specialized training students and standardized virtual patients. During the data collection process, facial images and voice information of students undergoing specialized training were subjected to spatiotemporal synchronization matching. The data spatiotemporal synchronization matching processing method is as follows: a synchronization signal generator is set in the interactive device for specialized training students. When it is necessary to collect facial images and voice information of specialized training students, the synchronization signal generator sends a unified start pulse signal to different data acquisition devices in the interactive device for specialized training students. All data acquisition devices take the unified start pulse signal as the absolute time starting point and generate a recording timestamp to achieve data spatiotemporal synchronization matching. The facial images of students in specialized training programs are processed with standard pose alignment and illumination normalization. At the same time, the speech information of students in specialized training programs is separated by combining spectral subtraction. The speech information of students in specialized training programs is then processed by short-time frame segmentation and windowing based on the Hamming window function method, so as to obtain the preprocessed facial image feature information and the preprocessed speech information of students in specialized training programs.

4. The intelligent evaluation method for doctor-patient communication effectiveness based on multimodal analysis as described in claim 1, characterized in that, The process involves multi-dimensional fusion analysis of preprocessed facial image feature information and preprocessed voice information of specialized training students. Based on the analysis results, the simulated interactive dialogue effects of the specialized training students are evaluated from different dimensions. Specifically: A big data network is introduced to retrieve the hesitant word knowledge graph of the preprocessed specialized training students' speech information. Based on the hesitant word knowledge graph, the speech information of the preprocessed specialized training students is analyzed, and the speech information sentences containing hesitant words are extracted and labeled as hesitant word speech information sentences. All speech information sentences with hesitant words are analyzed, the number of hesitant words and their duration are calculated, and speech information sentences with hesitant words whose number of hesitant words or duration is greater than the corresponding preset value are selected and marked as serious hesitant sentences. A key information knowledge graph for target doctor-patient communication scenarios is constructed to perform speech logic analysis on pre-processed speech information of specialized training students. Based on the results of speech logic analysis and severely hesitant sentences, a preliminary assessment of the simulated interactive dialogue effect of specialized training students is conducted.

5. The intelligent evaluation method for doctor-patient communication effectiveness based on multimodal analysis as described in claim 4, characterized in that, The constructed key information knowledge graph for the target doctor-patient communication scenario type is used to perform speech logic analysis on the pre-processed speech information of specialized training students. Based on the speech logic analysis results and severely hesitant statements, a preliminary assessment of the simulated interactive dialogue effect of the specialized training students is conducted. Specifically: Based on big data networks, key information knowledge graphs of target doctor-patient communication scenarios are retrieved. Combined with key information knowledge graphs, the voice information of students in preprocessing specialized training is analyzed to locate the corresponding key information in the voice information of students in different preprocessing specialized training. A relation extraction model is introduced to perform deep semantic logical association analysis on the key information corresponding to all preprocessed specialized training students' speech information, and to determine whether there is a logical association between the key information corresponding to different preprocessed specialized training students' speech information. If there is no logical connection between the key information in the preprocessed specialized training student speech information, then the corresponding sentences in the preprocessed specialized training student speech information will be marked as sentences without logical connection. A communication scoring terminal for specialized training students is pre-set in the interactive device for specialized training students. Within the communication scoring terminal for specialized training students, statements with serious hesitation and statements without logical connection are analyzed. If the proportion of unrelated statements in all preprocessed specialized training students’ voice information is less than the corresponding predetermined value, or the number of seriously hesitant statements is greater than the corresponding predetermined value, then the preliminary score of the specialized training students in the specialized training students’ communication scoring terminal will be deemed as the preliminary unqualified simulated interactive dialogue effect. If the proportion of illogical statements in all preprocessed specialized training students’ voice information is less than the corresponding predetermined value, and the number of seriously hesitant statements is greater than the corresponding preset value, then the preliminary score of the specialized training students in the specialized training students’ communication scoring terminal is initially qualified for the simulated interactive dialogue effect. The facial image feature information of students in the preprocessed specialized training program was analyzed, and the simulated interactive dialogue effect of the students was evaluated based on the analysis results.

6. The intelligent evaluation method for doctor-patient communication effectiveness based on multimodal analysis as described in claim 5, characterized in that, The process involves analyzing the facial image feature information of preprocessed specialized training students, and then conducting a secondary scoring of the simulated interactive dialogue performance based on the analysis results. Specifically: Based on the preprocessed speech information of specialized training students, the preprocessed facial image feature information of specialized training students is processed by sentence segmentation to ensure that different preprocessed speech information sentences correspond to different preprocessed facial image feature information of specialized training students, thus obtaining different preprocessed facial image feature information sub-segments. In the preprocessing of facial image feature information sub-segments of specialized training students, specific pixels on the faces of specialized training students are encoded, and the intensity change status and corresponding duration of different characteristic pixels after encoding are recorded in real time. Among them, specific pixels include the pixels at the corners of the mouth and the pixels at the eyebrows and eyes of students in specialized training. Based on the intensity change state and corresponding duration of different characteristic pixels after encoding, the gaze direction vector and blink probability of specialized training students are calculated. At the same time, the attention area is obtained in the standardized virtual patient scene library model. The attention area is the standard gaze direction area of ​​specialized training students when they conduct simulated interactive dialogue with standardized virtual patients. The students in the specialized training program were given a second score through the communication and scoring terminal, and their gaze direction vector and blink probability were analyzed. The duration of the gaze direction vector of the specialized training student within the attention zone is judged. If the simulated interactive dialogue effect is initially unqualified, and the duration of the gaze direction vector of the specialized training student within the attention zone is less than the predetermined duration, or the blinking probability of the specialized training student is greater than the predetermined standard value, then the secondary score will be judged as unqualified for the simulated interactive dialogue effect. If the simulated interactive dialogue effect is initially satisfactory, and the duration of the specialized training student's gaze direction vector within the attention zone is less than the predetermined duration, or the specialized training student's blinking probability is greater than the predetermined standard value, then the second evaluation will be assessed as the simulated interactive dialogue effect needing improvement. If the simulated interactive dialogue effect is initially satisfactory, and the duration of the specialized training student's gaze direction vector within the attention zone is not less than the predetermined duration, and the specialized training student's blinking probability is not greater than the predetermined standard value, then the secondary evaluation will be deemed satisfactory for the simulated interactive dialogue effect.

7. The intelligent evaluation method for doctor-patient communication effectiveness based on multimodal analysis as described in claim 1, characterized in that, The evaluation results of simulated interactive dialogues for specialized training students are combined to generate a structured training assessment report for these students, specifically: Based on the simulated interactive dialogue effects already assessed by specialized training students, a feedback report is generated within the specialized training student communication scoring terminal and labeled as a structured training evaluation report for specialized training students. The structured training evaluation report for specialized training students marks the simulated interactive dialogue effect of the specialized training students and scores different simulated interactive dialogue effects within a range of scores. Meanwhile, the structured training evaluation report for specialized training students generates a real-time change curve of the simulated interactive dialogue effect. The real-time change curve of the simulated interactive dialogue effect records the simulated interactive dialogue effect of specialized training students at different time points during the training process, and marks the time points when the simulated interactive dialogue effect is unsatisfactory and needs to be optimized.

8. The intelligent evaluation method for doctor-patient communication effectiveness based on multimodal analysis as described in claim 1, characterized in that, The process of performing prosodic and affective analysis on the preprocessed speech information of specialized training students, and updating the structured training evaluation report of the specialized training students based on the analysis results, specifically involves: Acoustic features are extracted from each frame of the preprocessed specialized training students' speech information. The acoustic features include fundamental frequency coefficients, Mel frequency cepstral coefficients, speech rate, and short-time energy. The range of combinations of different acoustic features under standard conditions is preset and calibrated as the standard acoustic feature range; The acoustic features of preprocessed specialized training student speech information in different frames are combined and analyzed. If, after combining the acoustic features of preprocessed specialized training student speech information, the acoustic feature range of a frame does not remain within the standard acoustic feature range, then the preprocessed specialized training student speech information of the corresponding frame is marked as an abnormal speech information frame. When the proportion of abnormal voice information frames in all frames is greater than the preset proportion, the structured training evaluation report of specialized training students will be updated. The updating of the structured training evaluation report for specialized training students is as follows: if the simulated interactive dialogue effect in the structured training evaluation report for specialized training students is unsatisfactory, then the simulated interactive dialogue effect will continue to be marked as unsatisfactory. If the simulated interactive dialogue effect is deemed to need optimization, then the simulated interactive dialogue effect will continue to be labeled as needing optimization. If the simulated interactive dialogue effect is satisfactory, then the simulated interactive dialogue effect will be marked as needing optimization.

9. An intelligent evaluation system for doctor-patient communication effectiveness based on multimodal analysis, characterized in that, The intelligent assessment system integrates a high-performance computing architecture and a data storage module, including a non-volatile memory consisting of a DDR4 RDIMM memory module with ECC verification and an NVMe solid-state storage array using 3D NAND flash memory, and a multi-core processor based on the Zen4 microarchitecture; the memory is embedded with an intelligent assessment method program for the effectiveness of doctor-patient communication, which has an intelligent assessment system engine. When the program is executed in parallel through the superscalar pipeline execution unit in the processor, it realizes the intelligent assessment steps for the effectiveness of doctor-patient communication as described in any one of claims 1-8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a program for intelligent evaluation of the effectiveness of doctor-patient communication. When the program is executed by a processor, it implements the steps for intelligent evaluation of the effectiveness of doctor-patient communication as described in any one of claims 1-8.

Images