A remote acquisition system based on four examinations of traditional Chinese medicine

Abstract

The present application relates to the field of data acquisition, especially to a remote acquisition system based on four diagnostic methods of traditional Chinese medicine, the present application sets information acquisition module, multi-modal analyzer, pulse data arbitrator, data processing module and acquisition storage module, analyzes language data and facial data, determines the reaction consistent characteristic value and behavior consistent characteristic value of the person being diagnosed, calculates the speech representation consistency parameter, determines the data state, classifies and processes the data, determines the digital pulse characteristics, respectively compares the features corresponding to the inconsistent data received, locates the abnormal data points, determines the data abnormal tendency at the abnormal data points, adjusts the abnormal data with weak abnormal tendency, determines the collected data, stores the consistent data and the collected data. The present application is aimed at the paradigm of traditional isolated equipment which only collects but does not process, filters redundant and invalid data, processes the data specifically, and improves the efficiency and accuracy of data acquisition and analysis.

Description

Technical Field

[0001] This invention relates to the field of data acquisition, and in particular to a remote data acquisition system based on the four diagnostic methods of traditional Chinese medicine. Background Technology

[0002] With the continuous development of science and technology and artificial intelligence, telemedicine has become a new normal in healthcare, meeting the needs of people who are unable to seek medical treatment due to work, location, or other reasons. It provides scenario support for the remote application of traditional Chinese medicine. At the same time, advancements in sensor technology, such as pulse sensors and high-resolution diagnostic equipment, can accurately collect information from the four diagnostic methods. The high speed and low latency of 5G technology ensure real-time data transmission, such as cloud data synchronization in intelligent diagnosis and treatment platforms for chronic diseases in traditional Chinese medicine. Artificial intelligence and big data technologies can analyze and process the collected data, assist in diagnosis, and provide support for model training and disease pattern research. Multiple factors have jointly promoted the development of this system.

[0003] Chinese Patent Publication No. CN106691407A discloses a comprehensive remote diagnostic system for Traditional Chinese Medicine (TCM), comprising a data acquisition subsystem, a signal processing subsystem, a computer unit, a pulse simulation subsystem, and a calibration subsystem. The data acquisition subsystem collects data on the patient's body temperature, facial features, tongue coating, tongue base, voice, and radial artery pulse signals. The signal processing subsystem processes the data collected by the data acquisition subsystem and transmits it via a communication network to enable remote connection between the patient and physician. The physician can apply different pressures (superficial, medium, and deep) to the radial artery at the patient's wrist to palpate the pulse based on the received pulse data. This invention integrates the four diagnostic methods of TCM—inspection, auscultation, inquiry, and palpation—allowing physicians to accurately assess a patient's condition without face-to-face consultation, thus enabling remote medical care and facilitating patient access to medical care.

[0004] Chinese Patent Publication No. CN103876713A discloses a remote pulse diagnosis device, comprising five parts: a data acquisition system, a computer unit, a pulse wave simulation system, a calibration device, and a simulated hand. By detecting the patient's radial artery pulse signal and body temperature signal, and leveraging network technology to connect home and hospital information, the system utilizes modern electronic technology and hemodynamic principles to simulate remote pulse diagnosis signals, achieving comprehensive remote reproduction of pulse diagnosis information. This system can be applied to telemedicine in Traditional Chinese Medicine (TCM), allowing TCM practitioners to truly understand the patient's physiological state and facilitating patient access to medical care. When combined with visualization software, it can fully realize the remote integration of information from the four diagnostic methods of TCM. Furthermore, this invention can also be used in the teaching practice of TCM.

[0005] It is evident that the existing technology still has the following problems:

[0006] In the current data acquisition paradigm, the digitization devices used for acquisition often work in isolation, lacking coordination and verification mechanisms. For the same diagnostic object, multiple heterogeneous devices will generate massive data streams in parallel. These data are prone to inconsistencies, incoordination, and even contradictions during the acquisition process. If the entire amount of raw data is acquired and transmitted without any processing, it will not only increase storage overhead and network load, but also affect the efficiency and accuracy of subsequent diagnostic analysis results. Summary of the Invention

[0007] To address this, the present invention provides a remote data acquisition system based on the four diagnostic methods of Traditional Chinese Medicine (TCM). This system overcomes the current data acquisition paradigm, where digital acquisition devices often operate in isolation, lacking coordination and verification mechanisms. For the same patient, multiple heterogeneous devices generate massive data streams in parallel, which are prone to inconsistencies, incoordination, and even contradictions during the acquisition process. If the entire raw data is acquired and transmitted without any processing, it will not only increase storage overhead and network load but also affect the efficiency and accuracy of subsequent diagnostic analysis.

[0008] To achieve the above objectives, the present invention provides a remote data acquisition system based on the four diagnostic methods of Traditional Chinese Medicine, comprising:

[0009] The information acquisition module is used to acquire the patient's language data, determine the sound features and semantic features of the language data, acquire the patient's facial data, and determine the facial wave texture features and local key pathological features.

[0010] A multimodal analyzer, connected to the information acquisition module, is used to analyze the language data and the facial data, determine the patient's response consistency feature value and behavior consistency feature value, calculate the speech representation consistency parameter, determine the data status, and classify the data.

[0011] A pulse data arbitrator, connected to the multimodal analyzer, is used to determine digital pulse characteristics based on diagnostic data acquired by the pulse diagnostic instrument, compare the characteristics corresponding to inconsistent data received, locate abnormal data points, and determine the abnormal data tendency at the abnormal data points.

[0012] A data processing module, which is connected to the pulse data arbitrator, is used to adjust for abnormal data with a weak abnormal tendency in order to determine the collected data.

[0013] The data acquisition and storage module is connected to the multimodal analyzer and the data processing module to store consistent data and the acquired data, and to perform data acquisition and transmission.

[0014] The diagnostic data refers to a textual description of the patient.

[0015] Furthermore, the information acquisition module determines the sound features and semantic features of the language data, including,

[0016] Used to construct the volume time-domain curve of the patient within a predetermined time period;

[0017] Used to determine the difference between the short-time average volume and the average volume;

[0018] The ratio of the difference to the average volume is used to determine the sound feature;

[0019] The semantic features used to determine the sentence keywords and sentiment words of the language data are used.

[0020] Furthermore, the information acquisition module determines facial wave texture features and local key pathological features, including,

[0021] Used to determine the new texture or reduced texture of the patient's face as a wavy texture;

[0022] This is used to determine the undulating texture length and the total texture length as facial undulating texture features;

[0023] Used to determine pathological keywords based on the facial color in a predetermined area;

[0024] Used to identify the pathological keywords as key local pathological features;

[0025] The predetermined area includes the forehead, cheeks, nose, and chin.

[0026] Furthermore, the multimodal analyzer determines the patient's response consistency characteristics and behavioral consistency characteristics, including,

[0027] This is used to calculate the similarity between the semantic features and the local key pathological features, and to determine the similarity as a response consistency feature value;

[0028] The average of the sum of the voice features and the facial wave texture features is used to determine the behavior consistency feature value.

[0029] Furthermore, the multimodal analyzer calculates speech representation consistency parameters, including:

[0030] The ratio of the reaction-consistent characteristic value to the baseline reaction-consistent characteristic value is used to determine the reaction influence factor;

[0031] The ratio of the behavioral consistency feature value to the benchmark behavioral consistency feature value is used to determine the behavioral influence factor;

[0032] The weighted sum of the response influence factor and the behavior influence factor is used to determine the verbal representation consistency parameter.

[0033] Furthermore, the multimodal analyzer determines the data state to classify the data, wherein,

[0034] If the speech representation consistency parameter is greater than the speech representation consistency parameter threshold, then it is determined that the language data and the facial data are consistent, and the consistent data is stored in the acquisition and storage module.

[0035] If the speech representation consistency parameter is less than or equal to the speech representation consistency parameter threshold, then the language data and the facial data are determined to be inconsistent, and the pulse data arbitrator is invoked to arbitrate the inconsistent data.

[0036] Furthermore, the pulse data arbitrator determines digital pulse characteristics, including:

[0037] This is used to traverse the pathology database and determine the corresponding pathology keywords;

[0038] This is used to identify the pathological keywords as digital pulse characteristics.

[0039] Furthermore, the pulse data arbitrator locates abnormal data points, including:

[0040] Used to identify the pathological keywords in the text description as digital pulse characteristics;

[0041] This is used to compare the digital pulse features with semantic features to obtain pulse language similarity;

[0042] This is used to compare the digital pulse characteristics with local key pathological features to obtain pulse behavior similarity.

[0043] If the pulse language similarity and / or pulse behavior similarity is less than the similarity threshold, then the data point corresponding to the digital pulse feature is determined to be an abnormal data point.

[0044] Furthermore, the pulse data arbitrator determines the data anomaly trend at the abnormal data points, wherein,

[0045] If the abnormal data point meets the preset conditions, the abnormal data point is determined to have a weak abnormal tendency, and the data processing module is invoked.

[0046] If the abnormal data point does not meet the preset conditions, then the abnormal data point is determined to have a strong abnormal tendency and will not be collected or stored.

[0047] The preset condition is that only one of the pulse language similarity and the pulse behavior similarity is less than the similarity threshold.

[0048] Furthermore, the data processing module makes adjustments for anomalous data with a weak tendency to anomaly, including:

[0049] This is used to replace the abnormal data corresponding to the abnormal data points with diagnostic data obtained based on the pulse diagnostic instrument.

[0050] Compared with existing technologies, this invention, by setting up an information acquisition module, a multimodal analyzer, a pulse data arbitrator, a data processing module, and a data acquisition and storage module, analyzes language data and facial data to determine the patient's response consistency feature value and behavioral consistency feature value, calculates speech representation consistency parameters, determines data status, classifies and processes the data, determines digital pulse characteristics based on diagnostic data acquired by the pulse diagnostic instrument, compares the characteristics corresponding to inconsistent data received, locates abnormal data points, determines the data abnormality tendency at the abnormal data points, adjusts abnormal data with weak abnormality tendency, determines the collected data, and stores consistent data and the collected data. This invention changes the traditional isolated device paradigm of "only collecting, not processing" data, filters redundant and invalid data, and improves the efficiency and accuracy of data acquisition and analysis.

[0051] In particular, by constructing a multimodal analyzer and a pulse data arbitrator, the acquired information is verified and processed. In practice, devices for collecting data from the four diagnostic methods of traditional Chinese medicine often operate in isolation, directly transmitting the entire raw data stream to analyze massive amounts of data. However, inconsistencies, incoordination, and even contradictions are easily observed during data collection. Transmitting such abnormal data not only wastes transmission links and storage space but also affects the efficiency and accuracy of data analysis. Based on this, this invention performs front-end processing to verify and process the data, rather than blindly collecting and transmitting the entire raw data stream from all devices. High-quality data deemed "consistent" by the arbitrator is directly stored; "inconsistent" data is adjusted or labeled before storage or deleted to filter redundant and invalid data, providing a data foundation for subsequent data analysis and improving the efficiency and accuracy of data collection and analysis.

[0052] In particular, by analyzing the language and facial data of patients and calculating the consistency parameter of speech representation, it is possible to determine whether there are any abnormalities in the patients and to process different categories of data in a targeted manner. In reality, patients may not be able to clearly describe their condition or may have misunderstood their condition, leading to discrepancies between language and facial expressions. If language or facial data is stored directly, these abnormal data will be hidden within the correct data, leading to errors in subsequent diagnostic analysis of the patients. Furthermore, facial data includes fluctuating texture and local facial color differences. It can be understood that fluctuating texture refers to the new or reduced texture caused by illness. For example, in the early stages of a fever, vasodilation in the patient may cause facial... Reduced facial texture, while increased facial texture due to dehydration from persistent fever, and different colors in different areas of the patient's face indicating abnormalities in different organs. By traversing the pathological database through local facial color differences, one or more pathological keywords can be obtained. Based on this, this invention extracts multimodal features from the acoustic features and semantic content of language signals, as well as the texture and color distribution in facial images, and further calculates their emotional consistency feature value and behavioral consistency feature value. Based on this, a speech representation consistency parameter is constructed. This parameter serves as a quantitative basis for judging whether the data is abnormal, providing a data foundation for subsequent data classification and targeted processing, and improving the efficiency and accuracy of data collection and analysis.

[0053] In particular, this invention achieves precise location of abnormal data points by determining digital pulse characteristics and comparing them with received inconsistent data. This provides a data foundation for subsequent classification of data anomaly trends and precise anomaly handling. In practice, anomaly data processing is mostly concentrated in the subsequent analysis stage. This approach not only reduces analysis efficiency but may also affect the accuracy of analysis results due to interference from hidden anomaly data. Based on this, this invention introduces diagnostic data from a pulse diagnostic instrument as an arbitration basis for situations where language data and facial data are inconsistent. By determining digital pulse characteristics, it achieves the identification of abnormal locations and the determination of abnormal trends in massive amounts of data, thereby classifying and processing the anomaly data, ensuring the usability of the collected data, and improving the efficiency and accuracy of data collection and analysis.

[0054] In particular, by acquiring data from three different dimensions—language, facial features, and pulse—the collected data can be identified. It is understandable that language and facial data belong to the subjectively controllable "behavioral layer" information, while pulse is an objective "physiological layer" information that is difficult to fake. This multimodal design achieves information complementarity and cross-validation, enabling the construction of a complete, multi-dimensional health status analysis method that ranges from overt behavior to internal physiology. This overcomes the limitations of single-modal information, which is prone to distortion and bias, and improves the efficiency and accuracy of data collection and analysis. Attached Figure Description

[0055] Figure 1 This is a schematic diagram of the structure of a remote data acquisition system based on the four diagnostic methods of traditional Chinese medicine, as described in an embodiment of the invention.

[0056] Figure 2 This is a logic block diagram illustrating the determination of data status for data classification processing in an embodiment of the invention.

[0057] Figure 3 This is a logical block diagram illustrating the location of abnormal data points in an embodiment of the invention.

[0058] Figure 4 This is a logic block diagram illustrating the determination of data anomaly trends at the anomalous data points, as described in an embodiment of the invention. Detailed Implementation

[0059] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0060] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0061] It should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the term "connection" should be interpreted broadly. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium; it can be a connection within two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0062] Please see Figure 1 , Figure 1 This is a schematic diagram of the remote data acquisition system based on the four diagnostic methods of Traditional Chinese Medicine (TCM) according to an embodiment of the invention. The remote data acquisition system based on the four diagnostic methods of TCM of the present invention includes:

[0063] The information acquisition module is used to acquire the patient's language data, determine the sound features and semantic features of the language data, acquire the patient's facial data, and determine the facial wave texture features and local key pathological features.

[0064] A multimodal analyzer, connected to the information acquisition module, is used to analyze the language data and the facial data, determine the patient's response consistency feature value and behavior consistency feature value, calculate the speech representation consistency parameter, determine the data status, and classify the data.

[0065] A pulse data arbitrator, connected to the multimodal analyzer, is used to determine digital pulse characteristics based on diagnostic data acquired by the pulse diagnostic instrument, compare the characteristics corresponding to inconsistent data received, locate abnormal data points, and determine the abnormal data tendency at the abnormal data points.

[0066] A data processing module, which is connected to the pulse data arbitrator, is used to adjust for abnormal data with a weak abnormal tendency in order to determine the collected data.

[0067] The data acquisition and storage module is connected to the multimodal analyzer and the data processing module to store consistent data and the acquired data, and to perform data acquisition and transmission.

[0068] The diagnostic data refers to a textual description of the patient.

[0069] Specifically, there are no restrictions on the method of acquiring language data. For example, a device with recording function can be used to acquire it, as long as it can capture the speech content and volume parameters of the patient. Those skilled in the art can determine the method based on the actual application scenario or needs, which will not be elaborated here.

[0070] Specifically, there are no restrictions on the method of acquiring facial data. For example, a 2D camera, a 3D vision sensor, an infrared thermal imager, or other devices with image acquisition capabilities can be used to capture multimodal data, including facial geometry, texture features, temperature distribution, and dynamic changes. It is only necessary to ensure that the facial data of the patient can be acquired. Those skilled in the art can determine the method based on the actual application scenario or accuracy requirements, which will not be elaborated here.

[0071] Specifically, there are no restrictions on the specific structure of the pulse diagnosis instrument, as long as it can realize the acquisition and digital conversion of pulse information. For example, a single-point pressure sensor, a multi-element array sensor, a photoplethysmography pulse wave sensor, or a multimodal composite sensing system can be used to collect multi-dimensional features including pulse position, pulse strength, pulse frequency, and pulse shape. Those skilled in the art can determine the appropriate method based on the diagnostic scenario or accuracy requirements, which will not be elaborated here.

[0072] Specifically, by constructing a multimodal analyzer and a pulse data arbitrator, the acquired information is verified and processed. In practice, devices for collecting data from the four diagnostic methods of traditional Chinese medicine often operate in isolation, directly transmitting the entire raw data stream for analysis. However, inconsistencies, incoordination, and even contradictions are easily observed during data collection. Transmitting such abnormal data not only wastes transmission links and storage space but also affects the efficiency and accuracy of data analysis. Therefore, this invention performs front-end processing to verify and process the data, rather than blindly collecting and transmitting the entire raw data stream from all devices. High-quality data deemed "consistent" by the arbitrator is directly stored; "inconsistent" data is adjusted or labeled before storage or deleted to filter redundant and invalid data, providing a data foundation for subsequent data analysis and improving the efficiency and accuracy of data collection and analysis.

[0073] Specifically, the information acquisition module determines the sound features and semantic features of the language data, including,

[0074] Used to construct the volume time-domain curve of the patient within a predetermined time period;

[0075] Used to determine the difference between the short-time average volume and the average volume;

[0076] The ratio of the difference to the average volume is used to determine the sound feature;

[0077] The semantic features used to determine the sentence keywords and sentiment words of the language data are used.

[0078] Specifically, there is no limit to the exact length of the predetermined time period. In practice, the predetermined time is set to 1 / 3 of the total collection time. Of course, those skilled in the art can also determine it according to the actual situation, as long as it is reasonable. This will not be elaborated further.

[0079] It is understandable that the horizontal axis of the volume time domain curve represents time, and the vertical axis represents volume.

[0080] Specifically, the short-term average volume characterizes the emotional fluctuations of the patient. Fluctuations in volume occur when the patient wants to disclose information or emphasize information. There are no restrictions on how the short-term average volume is determined. In practice, the initial fluctuation point and the end fluctuation point are determined, and the short-term average volume is calculated based on the volume between the two fluctuation points.

[0081] Specifically, there are no restrictions on how keywords and sentiment words are determined. For example, they can be identified by the parts of speech in the language data. In the sentence "I've been very sad lately, my head has been hurting," the noun is "head," the adjective is "ache," and the sentiment word is "sad."

[0082] Specifically, the information acquisition module determines facial texture features and key local pathological features, including...

[0083] Used to determine the new texture or reduced texture of the patient's face as a wavy texture;

[0084] This is used to determine the undulating texture length and the total texture length as facial undulating texture features;

[0085] Used to determine pathological keywords based on the facial color in a predetermined area;

[0086] Used to identify the pathological keywords as key local pathological features;

[0087] The predetermined area includes the forehead, cheeks, nose, and chin.

[0088] Specifically, fluctuating texture refers to the new or reduced texture caused by illness. It is understandable that in the early stages of a fever, vasodilation can lead to a decrease in facial texture, while prolonged fever can lead to an increase in facial texture due to dehydration.

[0089] Specifically, the total texture length is the sum of all textures within the facial image area of ​​the patient in a normal state. There are no restrictions on how the facial image in a normal state is acquired. For example, it can be an image uploaded by the patient himself or an image from other sources authorized by the patient. This will not be elaborated further.

[0090] It is understandable that different colors in different areas of the patient's face represent different organs. For example, the forehead corresponds to the heart and throat, the cheeks to the lungs and liver, the tip of the nose to the spleen and stomach, and the chin to the kidneys and reproductive system. Based on the color distribution in different areas of the face, one or more pathological keywords can be obtained by traversing the pathology database.

[0091] The method of obtaining the pathology database is not limited. For example, it can be an existing open-source database containing pathology keywords, corresponding symptom manifestations, corresponding area colors, and corresponding diagnostic data.

[0092] Specifically, the multimodal analyzer determines the respondent consistency characteristics and behavioral consistency characteristics of the patient, including,

[0093] This is used to calculate the similarity between the semantic features and the local key pathological features, and to determine the similarity as a response consistency feature value;

[0094] The average of the sum of the voice features and the facial wave texture features is used to determine the behavior consistency feature value.

[0095] Specifically, there are no restrictions on the method of calculating similarity. For example, the similarity between semantic features and local key pathological features can be calculated using the cosine similarity method. Of course, those skilled in the art can also choose a method according to the actual situation, as long as it is reasonable, which will not be elaborated here.

[0096] Specifically, the multimodal analyzer calculates speech representation consistency parameters, including,

[0097] The ratio of the reaction-consistent characteristic value to the baseline reaction-consistent characteristic value is used to determine the reaction influence factor;

[0098] The ratio of the behavioral consistency feature value to the benchmark behavioral consistency feature value is used to determine the behavioral influence factor;

[0099] The weighted sum of the response influence factor and the behavior influence factor is used to determine the verbal representation consistency parameter.

[0100] Specifically, the baseline response consistency characteristic value is calculated in advance. Several historical response consistency characteristic values ​​are obtained in advance, and the average of each historical response consistency characteristic value is determined as the baseline response consistency characteristic value.

[0101] Specifically, the baseline behavior consistency feature value is calculated in advance. Several historical behavior consistency feature values ​​are obtained in advance, and the average of each historical behavior consistency feature value is determined as the baseline behavior consistency feature value.

[0102] Specifically, the weight coefficients of the response consistency impact factor and the behavior consistency impact factor follow the normalization constraint, and their sum is 1. When weighting, considering that response consistency and behavior consistency are equally important during data collection, the weight coefficients of both the response impact factor and the behavior impact factor are set to 0.5 to reflect their equal contribution in the comprehensive evaluation.

[0103] Specifically, by analyzing the language and facial data of patients, a speech representation consistency parameter is calculated to determine whether any abnormalities exist. This allows for targeted processing of different categories of data. It is understandable that in reality, patients may not clearly describe their condition or misunderstand it, leading to discrepancies between their language and facial expressions. If language or facial data is stored directly, these abnormal data will be hidden within the correct data, causing errors in subsequent diagnostic analysis. Therefore, this invention extracts multimodal features from the acoustic features and semantic content of language signals, as well as the texture and color distribution in facial images. It further calculates emotional consistency and behavioral consistency feature values, thereby constructing a speech representation consistency parameter. This parameter serves as a quantitative basis for determining whether data is abnormal, providing a data foundation for subsequent data classification and targeted processing, and improving the efficiency and accuracy of data collection and analysis.

[0104] Please see Figure 2 , Figure 2 This is a logic block diagram illustrating the determination of data state for data classification processing according to an embodiment of the invention. Specifically, a multimodal analyzer determines the data state for data classification processing, wherein...

[0105] If the speech representation consistency parameter is greater than the speech representation consistency parameter threshold, then it is determined that the language data and the facial data are consistent, and the consistent data is stored in the acquisition and storage module.

[0106] If the speech representation consistency parameter is less than or equal to the speech representation consistency parameter threshold, then the language data and the facial data are determined to be inconsistent, and the pulse data arbitrator is invoked to arbitrate the inconsistent data.

[0107] Specifically, the speech representation consistency parameter threshold represents a boundary that can be directly collected and stored during the collection process. It is calculated in advance by obtaining the speech representation consistency of several historical collection processes in advance, and determining the product of the mean of the speech representation consistency of each historical speech representation and the accuracy coefficient as the speech representation consistency parameter threshold. The accuracy coefficient is selected in the interval [1.0, 1.2]. In practice, in order to improve the analysis accuracy, the accuracy coefficient is determined to be 1.1.

[0108] Specifically, the pulse data arbitrator determines digital pulse characteristics, including...

[0109] This is used to traverse the pathology database and determine the corresponding pathology keywords;

[0110] This is used to identify the pathological keywords as digital pulse characteristics.

[0111] Specifically, this invention determines digital pulse characteristics and compares them with received inconsistent data to accurately locate abnormal data points. This provides a data foundation for subsequent classification of data anomalies and precise anomaly handling. In practice, anomaly data processing is often concentrated in the subsequent analysis stage. This approach not only reduces analysis efficiency but may also affect the accuracy of analysis results due to interference from hidden anomalies. Therefore, this invention introduces diagnostic data from a pulse diagnostic instrument as an arbitration basis for situations where language data and facial data are inconsistent. By determining digital pulse characteristics, it identifies abnormal locations and determines abnormal tendencies in massive amounts of data, thereby classifying and processing the anomaly data, ensuring the usability of the collected data, and improving the efficiency and accuracy of data collection and analysis.

[0112] Please see Figure 3 , Figure 3 This is a logical block diagram illustrating the location of abnormal data points according to an embodiment of the invention. Specifically, the pulse data arbitrator locates abnormal data points, including:

[0113] Used to identify the pathological keywords in the text description as digital pulse characteristics;

[0114] This is used to compare the digital pulse features with semantic features to obtain pulse language similarity;

[0115] This is used to compare the digital pulse characteristics with local key pathological features to obtain pulse behavior similarity.

[0116] If the pulse language similarity and / or pulse behavior similarity is less than the similarity threshold, then the data point corresponding to the digital pulse feature is determined to be an abnormal data point.

[0117] Alternatively, data points corresponding to the digital pulse feature values ​​may not be collected.

[0118] Specifically, the similarity threshold represents a boundary where the pulse result differs from the language result and / or facial result. It is pre-calculated by acquiring the historical pulse language similarity and historical pulse behavior similarity corresponding to several normal data points in advance, calculating the average of the sum of the historical pulse language similarity and the historical pulse behavior similarity, and determining the product of the average and the similarity coefficient as the similarity threshold. The similarity coefficient is selected within the range [0.8, 1.0]. In order to improve the accuracy of data acquisition, the similarity coefficient is set to 0.9.

[0119] Please see Figure 4 , Figure 4 This is a logic block diagram illustrating the determination of data anomaly trends at the anomalous data points according to an embodiment of the invention. Specifically, the pulse data arbitrator determines the data anomaly trends at the anomalous data points, wherein...

[0120] If the abnormal data point meets the preset conditions, the abnormal data point is determined to have a weak abnormal tendency, and the data processing module is invoked.

[0121] If the abnormal data point does not meet the preset conditions, then the abnormal data point is determined to have a strong abnormal tendency and will not be collected or stored.

[0122] The preset condition is that only one of the pulse language similarity and the pulse behavior similarity is less than the similarity threshold.

[0123] Specifically, the data processing module makes adjustments for anomalous data with a weak tendency to anomaly, including:

[0124] This is used to replace the abnormal data corresponding to the abnormal data points with diagnostic data obtained based on the pulse diagnostic instrument.

[0125] Specifically, by acquiring data from three different dimensions—language, facial features, and pulse—the collected data is determined. It is understood that language and facial data belong to the subjectively controllable "behavioral layer" information, while pulse is the objective "physiological layer" information that is difficult to fake. This multimodal design realizes the complementarity and cross-validation of information, and can construct a complete health status analysis method from external behavior to internal physiology. It overcomes the limitations of single-modal information being prone to distortion and one-sidedness, and improves the efficiency and accuracy of data collection and analysis.

[0126] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0127] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A remote acquisition system based on four examinations of traditional Chinese medicine, characterized in that, The information acquisition module is used to acquire language data of a subject, determine sound features and semantic features of the language data, acquire facial data of the subject, determine facial fluctuation texture features and local key pathological features; The multi-modal analyzer connected with the information acquisition module is used to analyze the language data and the facial data, determine reaction consistency feature values and behavior consistency feature values of the subject, calculate speech representation consistency parameters, determine data states, and perform classification processing on the data; The pulse data arbitrator connected with the multi-modal analyzer is used to determine digital pulse feature values based on diagnostic data acquired by a pulse diagnosis instrument, compare the feature values corresponding to inconsistent data received from the multi-modal analyzer, locate abnormal data points, and determine data abnormality tendencies at the abnormal data points; The data processing module connected with the pulse data arbitrator is used to adjust abnormal data with weak abnormality tendencies to determine collected data; The collection storage module connected with the multi-modal analyzer and the data processing module is used to store consistent data and collected data, and perform data collection and transmission; The diagnostic data is a textual description of the subject; The information acquisition module determines sound features and semantic features of the language data, including, a time domain curve of the volume of the subject in a predetermined time period is constructed; a difference between a short-time average volume and an average volume is determined; a ratio of the difference to the average volume is determined as a sound feature; sentence keywords and emotional words of the language data are determined as semantic features; The information acquisition module determines facial fluctuation texture features and local key pathological features, including, newly generated textures or reduced textures on the face of the subject are determined as fluctuation textures; a length of the fluctuation textures and a total length of textures are determined as facial fluctuation texture features; pathological keywords are determined based on facial colors in predetermined areas; the pathological keywords are determined as local key pathological features; The predetermined areas include a forehead, a cheek, a nose, and a lower jaw; The multi-modal analyzer determines reaction consistency feature values and behavior consistency feature values of the subject, including, a similarity between the semantic features and the local key pathological features is calculated, and the similarity is determined as a reaction consistency feature value; an average value of a sum of the sound features and the facial fluctuation texture features is determined as a behavior consistency feature value; The multi-modal analyzer calculates speech representation consistency parameters, including, a ratio of the reaction consistency feature value to a reference reaction consistency feature value is determined as a reaction influence factor; a ratio of the behavior consistency feature value to a reference behavior consistency feature value is determined as a behavior influence factor; a weighted sum of the reaction influence factor and the behavior influence factor is determined as a speech representation consistency parameter; The multi-modal analyzer determines data states to perform classification processing on the data, wherein, ​ If the speech representation consistency parameter is greater than the speech representation consistency parameter threshold value, it is determined that the language data and the face data remain consistent, and the consistent data is stored to the collection storage module; If the speech representation consistency parameter is less than or equal to the speech representation consistency parameter threshold value, it is determined that the language data and the face data are inconsistent, and a pulse data arbitrator is called to arbitrate the inconsistent data.

2. The remote acquisition system based on the four examinations of traditional Chinese medicine according to claim 1, characterized in that, The pulse data arbitrator determines digital pulse features, including, to traverse a pathology database to determine corresponding several pathology keywords; to determine the pathology keywords as digital pulse features.

3. The remote acquisition system based on the four examinations of traditional Chinese medicine according to claim 1, characterized in that, The pulse data arbitrator locates abnormal data points, including, to determine that the pathology keywords of the textual description are digital pulse features; to compare the digital pulse features with semantic features for similarity to obtain pulse language similarity; to compare the digital pulse features with local key pathology features for similarity to obtain pulse behavior similarity; If the pulse language similarity and / or the pulse behavior similarity is less than a similarity threshold value, it is determined that the data point corresponding to the digital pulse feature is an abnormal data point.

4. The remote acquisition system based on the four diagnostic methods of traditional Chinese medicine according to claim 3, characterized in that, The pulse data arbitrator determines data abnormality tendency at the abnormal data point, wherein, If the abnormal data point meets a preset condition, it is determined that the abnormal data point is weakly abnormal, and a data processing module is called; If the abnormal data point does not meet the preset condition, it is determined that the abnormal data point is strongly abnormal, and no collection and storage are performed; Wherein, the preset condition is that the pulse language similarity and the pulse behavior similarity have only one similarity less than the similarity threshold value.

5. The remote acquisition system based on the four examinations of traditional Chinese medicine of claim 1, characterized in that, The data processing module adjusts the abnormal data of the weakly abnormal data, including, to replace the abnormal data corresponding to the abnormal data point with diagnosis data obtained based on a pulse diagnosis instrument.

Images