CN120203527A - A method and system for predicting pulmonary edema risk based on electromagnetic field biodetection - Google Patents

Abstract

The present invention discloses a pulmonary edema risk prediction method and system based on electromagnetic field biological detection, wherein the method comprises: obtaining a disturbance coefficient in a preset rotation detection mode by means of six electrodes arranged on the chest of a user to be detected, and inputting the disturbance coefficient into a pre-trained lung water volume prediction model to predict a lung water volume prediction value of the user to be detected; matching the chest circumference data of the user to be detected to its corresponding preset chest circumference segment, comparing the lung water volume prediction value with a preset threshold value corresponding to the preset chest circumference segment, and obtaining a prediction result; or calculating a lung impedance spectrum of the user to be detected based on the disturbance coefficient obtained by detection; and matching the chest circumference data of the user to be detected to its corresponding preset chest circumference segment, calculating the Similarity between the lung impedance spectrum of the user to be detected and the standard lung impedance spectrum corresponding to the preset chest circumference segment based on the cosine similarity theory; and then making a prediction based on the Similarity.

Description

Pulmonary edema risk prediction method and system based on electromagnetic field biological detection

Technical Field

The invention relates to the technical field of artificial intelligence, in particular to a pulmonary edema risk prediction method and system based on electromagnetic field biological detection.

Background

Heart failure (heart failure) is a serious manifestation or advanced stage of various heart diseases. Heart failure is worldwide, especially in china, and the prevalence rate, the rate of readmission, the rate of death, and the cost of diagnosis and treatment are all very high, bringing a heavy burden to the medical system. According to the latest data, the global heart failure patient is about 6430 ten thousand. Heart-derived pulmonary edema is a common serious clinical manifestation of heart failure, and in recent years, the prevalence of heart-derived pulmonary edema rises year by year, and the rate of hospitalization fatality is as high as 10% -20%. Heart pulmonary edema can add to the condition of heart failure, and due to the accumulation of fluid in alveoli, the respiratory function of patients is seriously affected, resulting in acute dyspnea and hypoxia, and rapid progress to acute respiratory failure, even sudden cardiac death. Related studies indicate that heart failure patients with pulmonary edema have more than twice the mortality rate than heart failure patients without pulmonary edema (48% VS 21%).

Overload is an important pathophysiological process for heart failure development, and detection of pulmonary congestion symptoms is an important means for heart failure capacity management. Therefore, in the face of cardiogenic pulmonary edema, it is necessary to relieve and improve symptoms as soon as possible, stabilize hemodynamic state, maintain important organ functions, avoid recurrence, and improve prognosis. Early identification diagnosis, real-time dynamic monitoring and intelligent early warning assessment of pulmonary edema are key links of heart failure or cardiogenic pulmonary edema diagnosis and treatment. However, there is still a lack of accurate, dynamic and economical monitoring techniques. Currently, diagnosis and monitoring of cardiogenic pulmonary edema mainly depends on plasma brain natriuretic peptide level measurement, chest imaging examination such as X-ray, CT, MRI, bedside cardiopulmonary ultrasound, pulmonary Capillary Wedge Pressure (PCWP), pulse index continuous cardiac ejection measurement (PICCO) noninvasive hemodynamic monitoring equipment, etc., and these methods have limitations such as time and effort consumption, invasiveness, radiation exposure, irregular quality of doctor reading, lack of real-time and dynamic monitoring capability, poor specificity or sensitivity, complex operation, no parameter for pulmonary water, no big data analysis and artificial intelligence, and thus effective treatment measures cannot be taken in time, and the heart failure development is aggravated.

It is proposed in the prior art to monitor pulmonary fluid by making electrical resistance measurements across the lungs. The more fluid is present in the lungs, the lower the impedance and an implantable medical device, such as a cardiac pacemaker, is proposed. Typically, a resistance measurement is made between the right ventricular electrode connected to the implantable medical device via a lead and another electrode located at the implantable medical device itself. Such impedance measurements sample thoracic tissue, including the lungs.

For example, U.S. patent application publication No. US10303305 discloses a method for measuring pulmonary impedance using an implantable medical device, where the resistance measurement is made between an electrode placed epicardially on the left ventricle wall of the heart and connected to the implantable medical device, and another electrode located at the implantable medical device. Impedance is measured by machining a stimulation current larger built-in electrode and measuring the resulting voltage with its built-in electrode, and then calculating the ratio of voltage to current.

For another example, chinese patent application CN101163443a discloses a pathology assessment by impedance measurement using convergent bioelectric lead fields by implanting a first electrode and a second electrode in a living body, then injecting a current to generate a second electric lead field for the first electric lead, then calculating an impedance value based on the potential difference and the injected current, and then assessing the pathology based on the impedance value. However, the above prior art is to implant electrodes in the lung or the lesion to be detected, i.e. invasive detection is used.

In view of this, the present invention provides a non-invasive pulmonary edema risk prediction method.

Disclosure of Invention

The invention aims to provide a pulmonary edema risk prediction method and a pulmonary edema risk prediction system based on electromagnetic field biological detection, which partially solve or alleviate the defects in the prior art and can realize noninvasive pulmonary edema risk prediction.

In order to solve the technical problems, the invention adopts the following technical scheme:

In a first aspect, the present invention provides a pulmonary edema risk prediction method based on electromagnetic field biological detection, comprising:

The method comprises the steps of acquiring disturbance coefficients in a preset rotation detection mode through six electrode plates arranged on the chest of a user to be detected at six specified positions, wherein the preset six positions are respectively that a first electrode is positioned between 3 rd ribs and 4 th ribs of a right collarbone midline in front of the chest, a second electrode and a third electrode are arranged between 6 th ribs and 7 th ribs of right armpit midline in the chest side by side, a fourth electrode and a fifth electrode are arranged between 6 th ribs and 7 th ribs of left armpit midline in the chest side by side, and a sixth electrode is positioned between 3 rd ribs and 4 th ribs of the left collarbone midline in front of the chest, and the disturbance coefficients are the average value of the disturbance coefficients on the propagation path from all electrodes serving as transmitting ends to each electrode serving as a receiving end;

Inputting the disturbance coefficient obtained by detection in the rotation detection mode into a pre-trained lung water volume prediction model to predict a lung water volume predicted value of the user to be detected, comparing the lung water volume predicted value with a preset threshold corresponding to the preset chest circumference section based on the chest circumference data of the user to be detected which is obtained in advance, if the lung water volume predicted value is smaller than a first preset threshold, the predicted result is risk-free, if the lung water volume predicted value is larger than a second preset threshold, the predicted result is high, if the lung water volume predicted value is larger than the first preset threshold and smaller than the third preset threshold, the predicted result is low, and if the lung water volume predicted value is larger than or equal to the third preset threshold and smaller than the second preset threshold, the predicted result is risk of stroke;

Or calculating the lung impedance spectrum of the user to be detected based on the disturbance coefficient detected in the rotation detection mode And based on the pre-acquired chest circumference data of the user to be detected being matched with the corresponding preset chest circumference segment, calculating the lung impedance spectrum Z _b of the user to be detected based on the cosine similarity theory and the pre-calculated standard lung impedance spectrum corresponding to the preset chest circumference segmentThe method comprises the steps of obtaining a prediction result of pulmonary edema of a user to be detected based on cosine Similarity, obtaining the prediction result of pulmonary edema of the user to be detected based on the cosine Similarity, wherein the prediction result is risk-free if the cosine Similarity is 1, high risk if the cosine Similarity is-1, low risk if the cosine Similarity is more than or equal to 0 and less than 1, and medium risk if the cosine Similarity is less than 0 and less than 1;

Wherein, the ,Is thatIs used for the frequency component of the (i) th frequency component,Is thatN is the total number of frequency components.

In some embodiments, the preset rotation detection mode refers to detection in a plurality of detection modes within a preset period time, and excitation signals in a frequency band of 10KHz-100KHz are sequentially applied in each detection mode, where the plurality of detection modes specifically include:

a first detection mode, in which any two electrodes of six electrodes are used as transmitting ends, any two electrodes of the remaining four electrodes are used as receiving ends, and the other two electrodes are used as grounding ends, and/or,

A second detection mode in which any two electrodes of the six electrodes are used as a transmitting end and a receiving end at the same time, and the remaining four electrodes are used as grounding ends, and/or,

In the third detection mode, any two electrodes in the six electrodes are used as a transmitting end and a receiving end at the same time, any two electrodes in the remaining four electrodes are grounded, and the other two electrodes are suspended.

In some embodiments, a standard lung impedance spectrum corresponding to the preset chest circumference segment is calculatedSpecifically comprising the steps of:

setting at least three bust segments based on a preset maximum bust threshold and minimum bust threshold;

Screening based on each chest circumference segment to obtain corresponding healthy subject groups, and obtaining at least three healthy subject groups;

Acquiring a health disturbance coefficient set of all healthy subjects in a preset rotation detection mode, and calculating the lung tissue impedance spectrum of each healthy subject based on the respective health disturbance coefficient of each healthy subject;

Fitting the lung tissue impedance spectrums of all healthy subjects in each group by using a least square method to obtain respective standard lung impedance spectrums of each group 。

In some embodiments, the step of training the lung water volume prediction model specifically includes:

The method comprises the steps of constructing a training data set, setting at least three chest circumference sections based on a preset maximum chest circumference threshold value and a preset minimum chest circumference threshold value, acquiring a healthy disturbance coefficient training set of healthy subjects with different chest circumferences in each chest circumference section in a preset rotation detection mode, and acquiring an abnormal disturbance coefficient training set of pulmonary edema subjects with different chest circumferences and different pulmonary water volumes in each chest circumference section in the preset rotation detection mode;

And model training, namely learning the healthy disturbance coefficient training set and the abnormal disturbance coefficient data set by using a machine learning algorithm to obtain the lung water volume prediction model.

In some embodiments, when obtaining an abnormal disturbance coefficient training set of pulmonary edema subjects with different chest circumference and different pulmonary water volumes in each chest circumference section in a preset rotation detection mode, respectively obtaining a first abnormal disturbance coefficient and a second abnormal disturbance coefficient of the pulmonary edema subjects in sitting postures and prone postures aiming at the same pulmonary edema subject, and taking the average value of the first abnormal disturbance coefficient of the sitting postures and the second abnormal disturbance coefficient corresponding to the prone postures under the same excitation signals as the abnormal disturbance coefficient of the pulmonary edema subjects to obtain the abnormal disturbance coefficient training set corresponding to each chest circumference section.

In a second aspect, the present invention provides a pulmonary edema risk prediction system based on electromagnetic field biological detection, comprising:

the six preset positions are respectively that a first electrode is positioned between 3 rd and 4 th ribs of a right collarbone midline in front of the chest of the user, a second electrode and a third electrode are arranged between 6 th and 7 th ribs of right armpit midline in side by side of the chest of the user, a fourth electrode and a fifth electrode are arranged between 6 th and 7 th ribs of left armpit midline in side by side of the chest of the user, and a sixth electrode is positioned between 3 rd and 4 th ribs of left collarbone midline in front of the chest of the user;

The data processing module is used for calculating the mean value based on disturbance coefficients on all propagation paths in a preset rotation detection mode and taking the mean value as a disturbance coefficient of a user to be detected;

The excitation generator is used for respectively sending electric stimulation signals to any two electrodes of the six electrode plates based on a preset rotation detection mode;

A first prediction module, configured to input the disturbance coefficient detected in the rotation detection mode into a pre-trained lung water volume prediction model to predict a lung water volume predicted value of a user to be detected, calculate a comparison between the lung water volume predicted value and a preset threshold corresponding to the preset chest circumference segment based on the chest circumference data of the user to be detected being matched to the corresponding preset chest circumference segment, if the lung water volume predicted value is smaller than the first preset threshold, the predicted result is risk-free, if the lung water volume predicted value is larger than the second preset threshold, the predicted result is high risk, if the lung water volume predicted value is larger than the second preset threshold, the predicted result is low risk, if the lung water volume predicted value is larger than or equal to the first preset threshold and smaller than the third preset threshold, the predicted result is stroke risk, or,

A second prediction module for calculating the lung impedance spectrum of the user to be detected based on the disturbance coefficient detected in the rotation detection modeAnd matching chest circumference data of the user to be detected to corresponding preset chest circumference segments, and calculating the lung impedance spectrum of the user to be detected based on cosine similarity theoryAnd the pre-calculated standard lung impedance spectrum of the preset chest circumference sectionAnd finally, predicting the occurrence of pulmonary edema of the user to be detected based on the cosine Similarity, wherein the prediction result is risk-free if the cosine Similarity is 1, the prediction result is high if the cosine Similarity is-1, the prediction result is low if the cosine Similarity is more than or equal to 0 and less than 1, and the prediction result is risk-free if the cosine Similarity is less than 0 and less than-1;

Wherein, the ,Is thatIs used for the frequency component of the (i) th frequency component,Is thatN is the total number of frequency components.

In some embodiments, the preset rotation detection mode refers to detection in a plurality of detection modes within a preset period time, and excitation signals in a frequency range of 10KHz-100KHz are sequentially applied to each detection mode, wherein the plurality of detection modes specifically comprise a first detection mode, wherein any two electrodes of six electrodes are used as transmitting ends, any two electrodes of the remaining four electrodes are used as receiving ends, and the other two electrodes are used as grounding ends, and/or a second detection mode, wherein any two electrodes of the six electrodes are used as transmitting ends and receiving ends at the same time, and the remaining four electrodes are used as grounding ends, and/or a third detection mode, wherein any two electrodes of the six electrodes are used as transmitting ends and receiving ends at the same time, and any two electrodes of the remaining four electrodes are grounded, and the other two electrodes are suspended.

In some embodiments, the second prediction module specifically includes:

the first calculation unit is used for calculating the lung tissue impedance spectrum of each healthy subject in each group based on the preset health disturbance coefficient set of each healthy subject group detected in the rotation detection mode;

a second calculation unit for fitting the lung tissue impedance spectrums of each healthy subject in each group by using a least square method to obtain respective standard lung impedance spectrums of each group 。

In some embodiments, the pulmonary edema risk prediction system based on electromagnetic field biological detection further comprises:

The model training module is used for respectively acquiring abnormal disturbance coefficient sets of a plurality of pulmonary edema subjects in sitting postures and prone postures under the preset rotation detection mode to construct a training sample library, and learning the abnormal disturbance coefficient sets in the training sample library by using a machine learning algorithm to obtain the pulmonary water volume prediction model.

In some embodiments, the pulmonary edema risk prediction system based on electromagnetic field biological detection further includes a data preprocessing module for performing data preprocessing on the electrical signals acquired through the six electrode pads, the data preprocessing including data cleaning, denoising and normalization processing.

The method and the system have the beneficial effects that the method and the system can realize noninvasive (or noninvasive), real-time and continuous (for example, 24 hours a day) monitoring and prediction of the lung water volume, are lower in cost and more convenient compared with the traditional auxiliary screening modes such as ultrasound, CT, X-ray and the like, particularly, the CT and the X-ray can not realize real-time monitoring, usually can only realize once a day, namely, can not realize continuous monitoring, and the method and the system avoid the risk that patients receive ionizing radiation, and are particularly suitable for long-term systemic circulation capacity management of heart failure patients. In addition, the noninvasive detection mode is adopted, and the adhesive electrode plates can be used for monitoring at any time beside a bed, so that the infection risk caused by invasive or invasive detection is avoided. And the possibility of triggering systemic inflammatory response, alleviating pain and medical risks for the patient.

The method and the system can timely capture dynamic changes of pulmonary edema, and provide more timely and accurate data support for diagnosis and treatment of doctors, so that the treatment scheme is adjusted, and illness delay caused by untimely examination is avoided.

The method and the system of the invention adopt a six-electrode rotation measurement structure (an electrode arrangement structure shown in fig. 2), break through the limitation of ReDS Pro single pair electrodes, and have the advantages of realizing the acquisition of multi-dimensional signals and effectively reducing the signal offset caused by the anisotropy of thoracic tissues.

The method and the system support the monitoring of the lung water in the prone position, compared with the sitting position, the lung compliance and the lung resistance of the heart failure patient can be changed, and the lung capacity can be changed, so that the acquired data are different. Therefore, the method is very suitable for real-time and continuous monitoring of the pulmonary edema of the prone position of the heart failure patient. Moreover, through the design of the rotation measurement structure, the deviation between the lung water volume measured in the supine position and the lung water volume in the anteverted sitting position is reduced to a certain extent, so that the two are almost the same, real-time and continuous monitoring can be carried out even if an ICU intubated patient (a person who cannot keep sitting posture) is realized, and the applicable crowd is enlarged.

The system has great advantages in the aspect of cost control, the terminal selling price is compressed to 1/3 of the bid, and the ReDS Pro terminal selling price is about 30 ten thousand yuan/station, so that even a primary hospital with less budget can configure the system.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below. Like elements or portions are generally identified by like reference numerals throughout the several figures. In the drawings, elements or portions thereof are not necessarily drawn to scale. It will be apparent to those of ordinary skill in the art that the drawings in the following description are of some embodiments of the invention and that other drawings may be derived from these drawings without inventive faculty.

FIG. 1 is a flow chart of a method for pulmonary edema prediction according to an embodiment of the present invention;

FIG. 2 is a schematic illustration of six electrodes in six designated positions of the chest during testing;

FIG. 3 is a flow chart of another embodiment of a pulmonary edema prediction method of the present invention;

FIG. 4 shows the results of comparing perturbation coefficients R obtained by the method of the present invention for the experimental group (i.e., pulmonary edema subject group) and the control group (i.e., healthy subject group), respectively;

FIG. 5 is a graph showing the results of monitoring 66 healthy subjects by the method of the present invention and statistically analyzing the disturbance coefficients obtained from the monitoring results;

FIG. 6 is a graph showing the results of analysis of perturbation coefficients obtained by repeated monitoring of a portion of 66 healthy subjects twice;

FIG. 7 is a graph showing the results of monitoring 16 pulmonary edema subjects using the method of the present invention and statistically analyzing the disturbance factor based on the monitoring results;

FIG. 8 is a comparison of mean perturbation coefficients for 66 healthy subjects and 16 pulmonary edema subjects.

The reference numerals are summarized as a first electrode 0, a second electrode 1, a third electrode 3, a fourth electrode 4, a fifth electrode 3 and a sixth electrode 5.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. It will be apparent that the described embodiments are some, but not all, embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

In this document, suffixes such as "module", "component", or "unit" used to represent elements are used only for facilitating the description of the present invention, and have no particular meaning in themselves. Thus, "module," "component," or "unit" may be used in combination.

The terms "upper," "lower," "inner," "outer," "front," "rear," "one end," "the other end," and the like herein refer to an orientation or positional relationship based on that shown in the drawings, merely for convenience of description and to simplify the description, and do not denote or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus should not be construed as limiting the invention. Furthermore, the terms "first," "second," and the like, are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.

The terms "mounted," "configured," "connected," and the like, as used herein, are intended to be interpreted broadly, unless otherwise specifically stated and defined, and they may be fixedly connected, detachably connected, or integrally connected, mechanically connected, directly connected, indirectly connected via an intermediary, or communicate between two elements. The specific meaning of the above terms in the present invention will be understood in specific cases by those of ordinary skill in the art.

Herein, "and/or" includes any and all combinations of one or more of the associated listed items.

Herein, "plurality" means two or more, i.e., it includes two, three, four, five, etc.

As used in this specification, the term "about" is typically expressed as +/-5% of the value, more typically +/-4% of the value, more typically +/-3% of the value, more typically +/-2% of the value, even more typically +/-1% of the value, and even more typically +/-0.5% of the value.

In this specification, certain embodiments may be disclosed in a format that is within a certain range. It should be appreciated that such a description of "within a certain range" is merely for convenience and brevity and should not be construed as a inflexible limitation on the disclosed ranges. Accordingly, the description of a range should be considered to have specifically disclosed all possible sub-ranges and individual numerical values within that range. For example, the description of ranges 1-6 should be considered to have specifically disclosed subranges such as from 1 to 3, from 1 to 4, from 1 to 5, from 2 to 4, from 2 to 6, from 3 to 6, etc., as well as individual numbers within this range, e.g., 1,2,3,4,5, and 6. The above rule applies regardless of the breadth of the range.

The prior art CN 115295139A discloses a bioelectrical impedance-based edema monitoring system which comprises a real-time control system, an upper computing system and a cloud storage system, wherein the real-time control system is in communication connection with the upper computing system, the upper computing system is in communication connection with the cloud storage system, the real-time control system is worn by a patient and used for measuring bioelectrical impedance and sending the bioelectrical impedance to the upper computing system, the upper computing system is arranged on a mobile terminal of the patient and used for extracting bioelectrical impedance characteristics and displaying the bioelectrical impedance characteristics to the patient and simultaneously sending the bioelectrical impedance characteristics to the cloud storage system, the cloud storage system is arranged on a hospital information system and used for storing bioelectrical impedance and characteristics and displaying the bioelectrical impedance characteristics to a doctor, and after the doctor makes evaluation on the edema condition, an evaluation result is returned for the patient to check. The monitoring system employs two excitation electrodes and two receiving electrodes to calculate the bioimpedance from just two voltages. However, the internal tissues and structures of different body parts are different, and in this same way, the pertinence is lacking, and the reliability of the detection result is required to be improved.

Embodiment one referring to fig. 1, a flowchart of a method for predicting (or evaluating) lung water volume based on electromagnetic field biological detection according to the present invention, specifically, the method comprises the steps of:

s101, obtaining a disturbance coefficient in a preset rotation detection mode through six electrode plates arranged at preset six positions on the chest of a user to be detected.

Referring to fig. 2, in some embodiments, the preset six positions are respectively that the first electrode is located between the 3 rd and 4 th ribs of the right collarbone midline in front of the chest of the user to be detected, the second and third electrodes are arranged between the 6 th and 7 th ribs of the right armpit midline in side by side of the chest of the user to be detected, the fourth and fifth electrodes are arranged between the 6 th and 7 th ribs of the left armpit midline in side of the chest of the user to be detected, and the sixth electrode is located between the 3 rd and 4 th ribs of the left collarbone midline in front of the chest of the user to be detected. Accordingly, the above-mentioned disturbance factor is the average value of disturbance factors on the propagation path between all the electrodes as transmitting ends to each electrode as receiving ends.

The difference between the disturbance factor of the chest (i.e., the first electrode and the sixth electrode) and the disturbance factor of the chest side (i.e., the second to fifth electrodes) is larger than the difference between the disturbance factor of the chest back (i.e., the mirror image position on the back of the user corresponding to the 3 rd, 4 th intercostal position of the right collarbone midline of the chest, the mirror image position between the 3 rd, 4 th intercostal position of the left collarbone midline of the chest) and the disturbance factor of the chest side, and therefore, in order to reduce the error, the disturbance factor set of the specified position of the chest and the disturbance factor set of the specified position of the chest side are acquired.

For example, a BORN-BE noninvasive cerebral edema dynamic monitor is used to collect the disturbance factor dataset.

In some embodiments, the preset rotation detection mode refers to detection in a plurality of detection modes within a preset period time (for example, 3min-5 min), and excitation signals in a frequency range of 10KHz-100KHz are sequentially applied in each detection mode, where the plurality of detection modes specifically include:

a first detection mode in which any two of the six electrodes are used as transmitting terminals, any two of the remaining four electrodes are used as receiving terminals, and the other two electrodes are used as grounding terminals (correspondingly, four disturbance coefficients are obtained), and/or,

A second detection mode in which any two of the six electrodes are used as the transmitting end and the receiving end at the same time, and the remaining four electrodes are used as the grounding ends (correspondingly, four disturbance coefficients are obtained), and/or,

In the third detection mode, any two electrodes of six electrodes are used as a transmitting end and a receiving end at the same time, any two electrodes of the remaining four electrodes are grounded, and the other two electrodes are suspended (correspondingly, four disturbance coefficients are obtained). Preferably, the disturbance factor of the final output is the average of the 12 disturbance factors.

Specifically, after an excitation signal generated by an excitation signal generator in the BORN-BE noninvasive cerebral edema dynamic monitor passes through the chest through a transmitting end electrode, the excitation signal flows into a receiving device in the BORN-BE noninvasive cerebral edema dynamic monitor from a receiving end electrode, and data analysis processing is performed to obtain a corresponding disturbance coefficient. Specifically, the preset frequency band is 10 kHz-100 kHz, and the preset frequency adjustment value assumes 10kHz, so that the excitation signal generator sequentially generates a reference electromagnetic wave signal and an adjustment electromagnetic wave signal at preset time intervals. Assuming that the frequency of the reference electromagnetic wave signal is 10kHz (corresponding to the excitation signal I), the frequency of the first adjustment electromagnetic wave signal is 10+10=20 kHz (corresponding to the excitation signal II), the frequency of the second adjustment electromagnetic wave signal is 20+10=30 kHz (corresponding to the excitation signal III.) and so on. I.e. detection under 10 excitation signals is completed within a preset period of time. Preferably, under each excitation signal, the disturbance coefficients in the three detection modes are obtained, so as to obtain a disturbance coefficient set.

Further, to improve the reliability of the data, data over multiple cycle times is acquired to construct the data set.

The low frequency signal is mainly sensitive to extracellular hydrates, while the high frequency signal can penetrate through the cell membrane into the cell, i.e. the high frequency signal is sensitive to both the inside and the outside of the cell, and the impedance of the tissue at the same position is lower at high frequency under the same state. Therefore, when pulmonary edema (e.g., pleural effusion, pericardial effusion) occurs, the pulmonary conductivity increases, and the bioelectrical impedance decreases, and at this time, the impedance decrease of the high-frequency signal is more remarkable, so that the sensitivity of the high-frequency signal to pulmonary edema is higher. Therefore, in the present embodiment, preferably, an excitation signal in the frequency range of 10KHz-100KHz is used.

S102, inputting the disturbance coefficient detected in the rotation detection mode into a pre-trained lung water volume prediction model, and predicting to obtain a lung water volume predicted value of a user to be detected.

Because the disturbance factor is positively correlated with the circumference, it is necessary to perform circumference segmentation in advance and acquire the respective lung water volumes and disturbance factors of healthy subjects and pulmonary edema subjects of different circumferences within each segment for model training. Correspondingly, the step of training the lung water volume prediction model specifically comprises the following steps:

building a training data set:

setting at least three bust segments based on a preset maximum bust threshold and minimum bust threshold;

obtaining a healthy disturbance coefficient training set of healthy subjects (the lung water quantity of the healthy subjects is a normal physiological experience value) with different chest sizes in each chest size section in a preset rotation detection mode;

And model training, namely learning the healthy disturbance coefficient training set and the abnormal disturbance coefficient data set by using a machine learning algorithm to obtain the lung water volume prediction model.

In some embodiments, to reduce errors, the plurality of segments are partitioned. Specifically, segmentation is performed based on a preset maximum chest circumference threshold value and a preset minimum chest circumference threshold value, and a preset chest circumference interval threshold value and the number of segments N, so as to obtain a normal disturbance coefficient set and an abnormal disturbance coefficient set of each segment (i.e., a disturbance coefficient training set of each segment). For example, a minimum frame size of 798mm, a maximum frame size of 986mm, a preset frame size interval threshold of 23mm and a preset number of segments of 8, segments of less than 798mm, segments of 798mm-821mm, & gtof 936mm-959mm, segments of greater than 982mm are obtained.

Further, because the different sexes have different threshold ranges of the chest circumference, different maximum chest circumference threshold values and different minimum chest circumference threshold values are preset for the different sexes respectively, and then segmentation is carried out. For example, for men, 874mm is a segment, 874mm-980mm is a segment, and greater than 980mm is a segment, while for women, 798mm is a segment, 798mm-986mm is a segment, and greater than 986mm is a segment. Correspondingly, in the subsequent process, a health disturbance coefficient training set of the healthy subjects with different chest girth in each chest girth section under different sexes in a preset rotation detection mode is obtained. Similarly, under different sexes, an abnormal disturbance coefficient training set of pulmonary edema subjects with different chest circumference and different pulmonary water volumes in each chest circumference section in a preset rotation detection mode is obtained. The disturbance coefficient training set is obtained by sex separation and chest circumference segmentation, so that errors introduced when the chest circumference segmentation is carried out without sex separation are eliminated.

For the user to be detected, the prone position can be almost realized, but for some users, the sitting position cannot be realized, but the disturbance coefficients detected under different positions have slight differences, so that preferably, the disturbance coefficient set of the healthy subject under the prone position and the disturbance coefficient of the non-edema subject are obtained, and correspondingly, the disturbance coefficient of the user to be detected under the prone position is also obtained in the subsequent prediction process.

Furthermore, in order to eliminate the influence of the body position, the abnormal disturbance coefficients of different body positions of each pulmonary edema subject under the same excitation signal can be obtained respectively. For example, a first abnormal disturbance factor of sitting posture and a second abnormal disturbance factor of prone posture in the same detection mode, and then calculating the average of the two disturbance factors and taking the average as the abnormal disturbance factor of the pulmonary edema subject. And in the subsequent prediction process, the abnormal disturbance coefficient mean value obtained by calculating different body positions based on the same principle is used as the abnormal disturbance coefficient of the user to be detected to be input into a prediction model for prediction.

In other embodiments, data preprocessing including data cleaning, denoising and normalization processes is performed prior to acquiring the disturbance factor data set based on electrical signals acquired by the six electrode pads.

The data cleaning step comprises the following steps:

Specifically, after the disturbance coefficient sets of the chest designated position and the chest designated position are obtained, the first electrode and the sixth electrode are directly switched to the mirror positions, the same detection mode is adopted again for detection, then the disturbance coefficient difference value between the chest back and the chest side (for example, the difference value between the disturbance coefficient when the first electrode of the mirror position on the chest back corresponds to the receiving end and the disturbance coefficient when the second electrode/the third electrode serves as the receiving end) is calculated, and if the difference value is smaller than or equal to the disturbance coefficient difference value between the chest front and the chest side (for example, the difference value between the disturbance coefficient when the first electrode positioned in the chest is used as the receiving end and the disturbance coefficient when the second electrode/the third electrode is used as the receiving end), the training data of the corresponding chest front and chest side are removed.

Based on the above method, the present invention further provides a pulmonary edema risk prediction (or risk assessment) method based on electromagnetic biological detection, referring to fig. 1, specifically, the method includes the steps of, in addition to the steps of the above embodiment:

S103, matching chest circumference data of a user to be detected to a corresponding preset chest circumference section, comparing the lung water volume predicted value with a preset threshold value corresponding to the preset chest circumference section, wherein if the lung water volume predicted value is smaller than a first preset threshold value, the predicted result is risk-free, if the lung water volume predicted value is larger than a second preset threshold value, the predicted result is high risk, if the lung water volume predicted value is larger than or equal to the first preset threshold value and smaller than a third preset threshold value, the predicted result is low risk, and if the lung water volume predicted value is larger than or equal to the third preset threshold value and smaller than the second preset threshold value, the predicted result is risk.

As described above, at least three chest circumference segments are preset, and different preset thresholds are set for each chest circumference segment, and accordingly, when risk determination is performed, the user to be detected is first matched with the corresponding chest circumference segment and the preset threshold corresponding to the chest circumference segment, and then risk evaluation is performed.

Specifically, for each segment, since the abnormal disturbance factor training set of the pulmonary edema subjects with different amounts of lung water under the preset rotation detection mode and the healthy disturbance factor training set of the healthy subjects are obtained in advance, and the oxygenation index is the oxygenation index as the gold standard for judging the severity of the case, the respective oxygenation index of each pulmonary edema subject can be obtained in advance (the oxygenation index is obtained as the prior art, which is not repeated here), and then the user classification is performed based on the oxygenation index, so as to obtain the low risk user, the middle risk user, the high risk user and the no risk user respectively, and the corresponding disturbance factor training set is marked. And then, respectively calculating the average value of the corresponding lung water quantity of the low-risk user, the medium-risk user and the high-risk user as a preset threshold value corresponding to each segment and each user.

Referring to fig. 3, a flowchart of another embodiment of a pulmonary edema risk prediction method based on electromagnetic field biological detection according to the present invention specifically includes the above step S101, except that in this embodiment, risk prediction is performed not based on the lung water amount but based on the cosine Similarity between the lung impedance spectrum Z _b of the user to be detected and the standard lung impedance spectrum Z _a of the corresponding healthy subject, specifically, after performing step S101, the steps are performed:

s104, calculating the lung impedance spectrum of the user to be detected based on the disturbance coefficient detected in the rotation detection mode ,Is vector quantityIs included in the frequency components of the frequency domain.

S105, matching the chest circumference data of the user to be detected to the corresponding preset chest circumference segment and the standard lung impedance spectrum corresponding to the preset chest circumference segment.

In some embodiments, the step of calculating a standard lung impedance spectrum for the respective chest circumference segment specifically comprises the steps of:

setting at least three bust segments based on a preset maximum bust threshold and minimum bust threshold;

Screening based on each chest circumference segment to obtain corresponding healthy subject groups, and obtaining at least three healthy subject groups;

acquiring a health disturbance coefficient set of all healthy subjects in a preset rotation detection mode, and calculating the respective lung tissue impedance spectrum of each healthy subject based on the health disturbance coefficient detected by each healthy subject in the rotation detection mode;

Fitting the lung tissue impedance spectra of all healthy subjects in each group by using a least square method to obtain a standard lung impedance spectrum corresponding to each group, namely each chest circumference segment ;

Wherein, the ,Is vector quantityN is the total number of frequency components.

In this embodiment, by performing chest circumference segmentation, and the duty ratio of the disturbance coefficient set of each segment is the same, the chest circumference distribution of healthy subjects is more reasonable, and the influence caused by the larger chest circumference difference is reduced as much as possible.

Referring to the above embodiment, when chest circumference segmentation is performed without discriminating the sex, a plurality of segments are divided in order to reduce errors. Specifically, segmentation is performed based on a preset maximum chest circumference threshold value and a preset minimum chest circumference threshold value, and a preset chest circumference interval threshold value and the number of segments N, so as to obtain a normal disturbance coefficient set and an abnormal disturbance coefficient set (i.e., a disturbance coefficient training set) of each segment. For example, a minimum frame size of 798mm, a maximum frame size of 986mm, a preset frame size interval threshold of 23mm and a preset number of segments of 8, segments of less than 798mm, segments of 798mm-821mm, & gtof 936mm-959mm, segments of greater than 982mm are obtained.

In other embodiments, different maximum and minimum bust thresholds are preset and segmented for different sexes, respectively. For example, for men, 874mm is a segment, 874mm-980mm is a segment, and greater than 980mm is a segment, while for women, 798mm is a segment, 798mm-986mm is a segment, and greater than 986mm is a segment. Correspondingly, in the subsequent process, a health disturbance coefficient training set of the healthy subjects in the different chest circumference sections under different polarities in a preset rotation detection mode is obtained. Similarly, an abnormal disturbance coefficient training set of pulmonary edema subjects with different lung water volumes in different chest circumference sections under different sexes in a preset rotation detection mode is obtained. The training set is acquired by separating the gender and the segmentation, so that errors introduced when the segmentation is performed without distinguishing the gender are eliminated.

S106, calculating the lung impedance spectrum of the user to be detected based on cosine similarity theoryStandard lung impedance spectrum segmented with corresponding chest circumferenceCosine Similarity between them.

In some embodiments of the present invention, in some embodiments,,For pulmonary impedance spectroscopyIs the i-th of the frequency component,Is a standard pulmonary impedance spectrumN is the total number of frequency components, where,To vector the vectorProjection to vectorIn the time-course of which the first and second contact surfaces, the inner product of these two vectors; the angle between the two vectors.

S107, predicting the pulmonary edema of the user to be detected based on the cosine Similarity, wherein the predicted result is risk-free if the cosine Similarity is 1, high risk if the cosine Similarity is-1, low risk if the cosine Similarity is more than or equal to 0 and less than 1, and risk if the cosine Similarity is less than 0 and less than 1.

The disturbance coefficient set of nearly 200 healthy people and pulmonary edema patients is obtained by the method, and analysis is carried out, so that the disturbance coefficient of the pulmonary edema group is obviously lower than that of a control group (namely the healthy people group), and therefore, the risk prediction of the pulmonary edema can be carried out based on the disturbance coefficient. Compared with the traditional screening mode, the method has good stability and higher primary screening accuracy and monitoring accuracy for the pulmonary edema evaluation of the patient.

Referring to fig. 5, it is known that the perturbation coefficients of 66 healthy subjects (39 of them are men and 27 of them are women) all satisfy the normal distribution (p=0.69 > 0.05). The maximum perturbation coefficient Max for these 66 healthy subjects was 330 and the minimum perturbation coefficient was 105.

To verify the reliability of the method, 8 healthy subjects were repeatedly monitored using the method of the present invention, 1 of which was invalid data. See fig. 6, where the two measurements of 7 subjects in effect were not significantly different (t=0.374, p=0.715 > 0.05). Wherein P, T is a statistically common parameter.

Referring to fig. 7, the perturbation coefficients of 16 pulmonary edema subjects (12 men, 4 women) in which the effective cases were counted did not satisfy the normal distribution (p=0.04 < 0.05).

Referring to fig. 8, the perturbation coefficients of the healthy human group satisfy the normal distribution, but the perturbation coefficients of the case group do not satisfy the normal distribution, and the comparison of the differences of the perturbation coefficients of the healthy human group and the case group by rank sum test shows that the differences of the perturbation coefficient values of the two groups have statistical significance (p < 0.001).

Based on the method, the invention also provides a pulmonary edema risk prediction system based on electromagnetic field biological detection, which comprises the following steps:

The device comprises six electrode plates, a first electrode, a second electrode, a third electrode, a fourth electrode, a fifth electrode and a sixth electrode, wherein the six electrode plates are used for constructing an excitation signal propagation path between six specified positions on the chest of a user to be detected and obtaining a disturbance coefficient of each propagation path in a preset rotation detection mode;

The data processing module is used for calculating the mean value based on the disturbance coefficients on all propagation paths in a preset rotation detection mode and taking the mean value as the disturbance coefficient of the user to be detected;

the excitation generator is used for generating electric stimulation signals (namely excitation signals) to the six electrode plates respectively based on a preset rotation detection mode;

The first prediction module is used for inputting the disturbance coefficient detected in the rotation detection mode into a pre-trained lung water volume prediction model to predict a lung water volume predicted value of a user to be detected, predicting a predicted result of lung edema of the user to be detected based on the lung water volume predicted value, wherein the predicted result is risk-free if the lung water volume predicted value is smaller than a first preset threshold value, the predicted result is high if the lung water volume predicted value is larger than a second preset threshold value, the predicted result is low if the lung water volume predicted value is larger than or equal to the first preset threshold value and smaller than a third preset threshold value, the predicted result is medium risk if the lung water volume predicted value is larger than or equal to the third preset threshold value and smaller than the second preset threshold value, or

A second prediction module for calculating the lung impedance spectrum of the user to be detected based on the disturbance coefficient detected in the rotation detection modeAnd calculating the lung impedance spectrum of the user to be detected based on cosine similarity theoryWith a pre-calculated lung impedance spectrum of a healthy subjectAnd finally, predicting the occurrence of pulmonary edema of the user to be detected based on the cosine Similarity, wherein the prediction result is risk-free if the cosine Similarity is 1, the prediction result is high if the cosine Similarity is-1, the prediction result is low if the cosine Similarity is more than or equal to 0 and less than 1, and the prediction result is risk-free if the cosine Similarity is less than 0 and less than-1.

In some embodiments, the preset rotation detection mode refers to detection in a plurality of detection modes within a preset period time, and excitation signals in a frequency range of 10K-100K are sequentially applied to each detection mode, where the plurality of detection modes specifically include:

a first detection mode, in which any two electrodes of six electrodes are used as transmitting ends, any two electrodes of the remaining four electrodes are used as receiving ends, and the other two electrodes are used as grounding ends, and/or,

A second detection mode in which any two electrodes of the six electrodes are used as a transmitting end and a receiving end at the same time, and the remaining four electrodes are used as grounding ends, and/or,

In the third detection mode, any two electrodes in the six electrodes are used as a transmitting end and a receiving end at the same time, any two electrodes in the remaining four electrodes are grounded, and the other two electrodes are suspended.

Compared with the low frequency band below 10KHz and the high frequency band above 50KHz, the disturbance coefficient mean value obtained in the frequency band of 10KHz-100KHz is more stable, so in order to reduce errors, in the embodiment, the excitation signal in the frequency band of 10KHz-100KHz is mainly adopted for excitation.

In some embodiments, at least three chest circumference segments are preset based on a preset maximum chest circumference threshold and a preset minimum chest circumference threshold, then corresponding groups of healthy subjects are obtained by screening based on each chest circumference segment, at least three groups of healthy subjects are obtained, and finally the respective lung tissue impedance spectrums of each group are calculated respectively. The second prediction module specifically comprises a first calculation unit, a second calculation unit and a third calculation unit, wherein the first calculation unit is used for calculating the lung tissue impedance spectrum of each healthy subject in each group based on the respective health disturbance coefficient set of each healthy subject group detected in the rotation detection mode, and the second calculation unit is used for fitting the lung tissue impedance spectrum of each healthy subject in each group by using a least square method to obtain the respective standard lung impedance spectrum of each group。

In some embodiments, where the system includes a first prediction module, the system further includes:

The system comprises a data set construction module, a data set generation module and a data set generation module, wherein the data set construction module is used for acquiring health disturbance coefficient training sets of healthy subjects in different chest circumference sections in a preset rotation detection mode;

And model training, namely learning the healthy disturbance coefficient training set and the abnormal disturbance coefficient data set by using a machine learning algorithm to obtain the lung water volume prediction model.

In some embodiments, when obtaining abnormal disturbance coefficient training sets of pulmonary edema subjects with different chest circumference segments and different amounts of lung water in a preset rotation detection mode, for the same pulmonary edema subject, respectively obtaining a first abnormal disturbance coefficient and a second abnormal disturbance coefficient of the pulmonary edema subject in a sitting posture, and taking a mean value of the first abnormal disturbance coefficient of the sitting posture and the second abnormal disturbance coefficient corresponding to a prone posture under the same excitation signal as the abnormal disturbance coefficient of the pulmonary edema subject, thereby obtaining an abnormal disturbance coefficient set of each group.

In some embodiments, the system further comprises a data preprocessing module for performing data preprocessing on the electrical signals acquired through the six electrode pads, wherein the data preprocessing comprises data cleaning, denoising and normalizing.

It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM, magnetic disk, optical disk) comprising several instructions for causing a computer terminal (which may be a mobile phone, a computer, a server, or a network device, etc.) to perform the method according to the embodiments of the present invention.

The embodiments of the present invention have been described above with reference to the accompanying drawings, but the present invention is not limited to the above-described embodiments, which are merely illustrative and not restrictive, and many forms may be made by those having ordinary skill in the art without departing from the spirit of the present invention and the scope of the claims, which are to be protected by the present invention.

Claims (10)

1. A pulmonary edema risk prediction method based on electromagnetic field biological detection, comprising:

The method comprises the steps of acquiring disturbance coefficients in a preset rotation detection mode through six electrode plates arranged on the chest of a user to be detected at six specified positions, wherein the preset six positions are respectively that a first electrode is positioned between 3 rd ribs and 4 th ribs of a right collarbone midline in front of the chest, a second electrode and a third electrode are arranged between 6 th ribs and 7 th ribs of right armpit midline in the chest side by side, a fourth electrode and a fifth electrode are arranged between 6 th ribs and 7 th ribs of left armpit midline in the chest side by side, and a sixth electrode is positioned between 3 rd ribs and 4 th ribs of the left collarbone midline in front of the chest, and the disturbance coefficients are the average value of the disturbance coefficients on the propagation path from all electrodes serving as transmitting ends to each electrode serving as a receiving end;

Inputting the disturbance coefficient obtained by detection in the rotation detection mode into a pre-trained lung water volume prediction model to predict a lung water volume predicted value of the user to be detected, comparing the lung water volume predicted value with a preset threshold corresponding to the preset chest circumference section based on the chest circumference data of the user to be detected which is obtained in advance, if the lung water volume predicted value is smaller than a first preset threshold, the predicted result is risk-free, if the lung water volume predicted value is larger than a second preset threshold, the predicted result is high, if the lung water volume predicted value is larger than the first preset threshold and smaller than the third preset threshold, the predicted result is low, and if the lung water volume predicted value is larger than or equal to the third preset threshold and smaller than the second preset threshold, the predicted result is risk of stroke;

Or calculating the lung impedance spectrum of the user to be detected based on the disturbance coefficient detected in the rotation detection mode And matching the chest circumference data of the user to be detected to the corresponding preset chest circumference segment based on the pre-acquired chest circumference data, and calculating the lung impedance spectrum of the user to be detected based on cosine similarity theoryStandard lung impedance spectrum corresponding to the pre-calculated preset chest circumference segmentThe method comprises the steps of obtaining a prediction result of pulmonary edema of a user to be detected based on cosine Similarity, obtaining the prediction result of pulmonary edema of the user to be detected based on the cosine Similarity, wherein the prediction result is risk-free if the cosine Similarity is 1, high risk if the cosine Similarity is-1, low risk if the cosine Similarity is more than or equal to 0 and less than 1, and medium risk if the cosine Similarity is less than 0 and less than 1;

Wherein, the ,Is thatIs used for the frequency component of the (i) th frequency component,Is thatN is the total number of frequency components.

2. The pulmonary edema risk prediction method based on electromagnetic field biological detection according to claim 1, wherein the preset rotation detection mode is detection in a plurality of detection modes within a preset period time, and excitation signals in a frequency band of 10KHz-100KHz are sequentially applied in each detection mode, wherein the plurality of detection modes specifically include:

a first detection mode, in which any two electrodes of six electrodes are used as transmitting ends, any two electrodes of the remaining four electrodes are used as receiving ends, and the other two electrodes are used as grounding ends, and/or,

A second detection mode in which any two electrodes of the six electrodes are used as a transmitting end and a receiving end at the same time, and the remaining four electrodes are used as grounding ends, and/or,

In the third detection mode, any two electrodes in the six electrodes are used as a transmitting end and a receiving end at the same time, any two electrodes in the remaining four electrodes are grounded, and the other two electrodes are suspended.

3. The pulmonary edema risk prediction method based on electromagnetic field biological detection according to claim 2, wherein the step of calculating the standard pulmonary impedance spectrum Za corresponding to the preset chest circumference segment specifically comprises the steps of:

setting at least three bust segments based on a preset maximum bust threshold and minimum bust threshold;

Screening based on each chest circumference segment to obtain corresponding healthy subject groups, and obtaining at least three healthy subject groups;

Acquiring a health disturbance coefficient set of all healthy subjects in a preset rotation detection mode, and calculating the lung tissue impedance spectrum of each healthy subject based on the respective health disturbance coefficient of each healthy subject;

Fitting the lung tissue impedance spectrums of all healthy subjects in each group by using a least square method to obtain respective standard lung impedance spectrums of each group 。

4. A pulmonary edema risk prediction method based on electromagnetic field biological detection as claimed in claim 1, wherein the step of training the pulmonary water volume prediction model specifically comprises:

The method comprises the steps of constructing a training data set, setting at least three chest circumference sections based on a preset maximum chest circumference threshold value and a preset minimum chest circumference threshold value, acquiring a healthy disturbance coefficient training set of healthy subjects with different chest circumferences in each chest circumference section in a preset rotation detection mode, and acquiring an abnormal disturbance coefficient training set of pulmonary edema subjects with different chest circumferences and different pulmonary water volumes in each chest circumference section in the preset rotation detection mode;

And model training, namely learning the healthy disturbance coefficient training set and the abnormal disturbance coefficient data set by using a machine learning algorithm to obtain the lung water volume prediction model.

5. The method for predicting pulmonary edema risk based on electromagnetic field biological detection of claim 4, wherein when obtaining an abnormal disturbance coefficient training set of pulmonary edema subjects with different chest circumference and different lung water volumes in each chest circumference section in a preset rotation detection mode, respectively obtaining a first abnormal disturbance coefficient and a second abnormal disturbance coefficient of the pulmonary edema subjects in sitting postures and prone postures for the same pulmonary edema subject, and taking the average value of the first abnormal disturbance coefficient of the sitting postures and the second abnormal disturbance coefficient of the prone postures under the same excitation signals as the abnormal disturbance coefficient of the pulmonary edema subjects to obtain the abnormal disturbance coefficient training set corresponding to each chest circumference section.

6. A pulmonary edema risk prediction system based on electromagnetic field biological detection, comprising:

the six preset positions are respectively that a first electrode is positioned between 3 rd and 4 th ribs of a right collarbone midline in front of the chest of the user, a second electrode and a third electrode are arranged between 6 th and 7 th ribs of right armpit midline in side by side of the chest of the user, a fourth electrode and a fifth electrode are arranged between 6 th and 7 th ribs of left armpit midline in side by side of the chest of the user, and a sixth electrode is positioned between 3 rd and 4 th ribs of left collarbone midline in front of the chest of the user;

The data processing module is used for calculating the mean value based on disturbance coefficients on all propagation paths in a preset rotation detection mode and taking the mean value as a disturbance coefficient of a user to be detected;

The excitation generator is used for respectively sending electric stimulation signals to any two electrodes of the six electrode plates based on a preset rotation detection mode;

A first prediction module, configured to input the disturbance coefficient detected in the rotation detection mode into a pre-trained lung water volume prediction model to predict a lung water volume predicted value of a user to be detected, calculate a comparison between the lung water volume predicted value and a preset threshold corresponding to the preset chest circumference segment based on the chest circumference data of the user to be detected being matched to the corresponding preset chest circumference segment, if the lung water volume predicted value is smaller than the first preset threshold, the predicted result is risk-free, if the lung water volume predicted value is larger than the second preset threshold, the predicted result is high risk, if the lung water volume predicted value is larger than the second preset threshold, the predicted result is low risk, if the lung water volume predicted value is larger than or equal to the first preset threshold and smaller than the third preset threshold, the predicted result is stroke risk, or,

A second prediction module for calculating the lung impedance spectrum of the user to be detected based on the disturbance coefficient detected in the rotation detection modeAnd matching chest circumference data of the user to be detected to corresponding preset chest circumference segments, and calculating the lung impedance spectrum of the user to be detected based on cosine similarity theoryAnd the pre-calculated standard lung impedance spectrum of the preset chest circumference sectionAnd finally, predicting the occurrence of pulmonary edema of the user to be detected based on the cosine Similarity, wherein the prediction result is risk-free if the cosine Similarity is 1, the prediction result is high if the cosine Similarity is-1, the prediction result is low if the cosine Similarity is more than or equal to 0 and less than 1, and the prediction result is risk-free if the cosine Similarity is less than 0 and less than-1;

Wherein, the , Is thatIs used for the frequency component of the (i) th frequency component,Is thatN is the total number of frequency components.

7. The pulmonary edema risk prediction system based on electromagnetic field biological detection of claim 6, wherein the preset rotation detection mode is detection in a plurality of detection modes within a preset cycle time, and excitation signals in a frequency band of 10KHz-100KHz are sequentially applied in each detection mode, wherein the plurality of detection modes specifically include:

a first detection mode, in which any two electrodes of six electrodes are used as transmitting ends, any two electrodes of the remaining four electrodes are used as receiving ends, and the other two electrodes are used as grounding ends, and/or,

A second detection mode in which any two electrodes of the six electrodes are used as a transmitting end and a receiving end at the same time, and the remaining four electrodes are used as grounding ends, and/or,

In the third detection mode, any two electrodes in the six electrodes are used as a transmitting end and a receiving end at the same time, any two electrodes in the remaining four electrodes are grounded, and the other two electrodes are suspended.

8. The pulmonary edema risk prediction system based on electromagnetic field biological detection of claim 6, wherein the second prediction module specifically comprises a first calculation unit for calculating a pulmonary tissue impedance spectrum of each healthy subject in each group based on a preset health disturbance coefficient set of each healthy subject group detected in the rotation detection mode, a second calculation unit for fitting the pulmonary tissue impedance spectrum of each healthy subject in each group by using a least square method to obtain a standard pulmonary impedance spectrum of each group。

9. The pulmonary edema risk prediction system based on electromagnetic field biological detection of claim 6, further comprising:

The model training module is used for respectively acquiring abnormal disturbance coefficient sets of a plurality of pulmonary edema subjects in sitting postures and prone postures under the preset rotation detection mode to construct a training sample library, and learning the abnormal disturbance coefficient sets in the training sample library by using a machine learning algorithm to obtain the pulmonary water volume prediction model.

10. The pulmonary edema risk prediction system based on electromagnetic field biological detection of claim 6, further comprising a data preprocessing module for performing data preprocessing on the electrical signals acquired through the six electrode pads, the data preprocessing including data cleaning, denoising, and normalization processing.