Respiratory status warning method, device and computer equipment based on vital sign signals

Abstract

The present invention relates to the field of respiratory warning, and in particular to a respiratory status warning method, device and computer equipment based on vital sign signals, which obtain the vital sign signals of a sample user at a preset first time scale; extract the ballistocardiogram signal and the respiratory signal from the vital sign signals of the respiratory status sample user; determine the feature sequence corresponding to the ballistocardiogram signal and the respiratory status breathing signal of the respiratory status sample user; determine the differential feature sequence of the respiratory status feature sequence at several different second time scales; input the differential feature sequence of the respiratory status sample user into a preset respiratory status classification model to obtain the respiratory status sequence of the respiratory status sample user; update the model parameter data of the respiratory status classification model according to the differential feature sequence and the respiratory status sequence of the respiratory status sample user to obtain the target respiratory status classification model, which can reduce the cost and improve the accuracy of the respiratory status warning.

Description

Technical Field

[0001] The present invention relates to the field of respiratory warning, and in particular to a respiratory status warning method, device, computer equipment and storage medium based on vital sign signals. Background Art

[0002] Chronic obstructive pulmonary disease (COPD) is a preventable and treatable respiratory disease characterized by persistent airflow limitation. Clinically, it can manifest as varying degrees of dyspnea, cough or sputum. Failure to intervene in time can accelerate disease progression and seriously affect the patient's quality of life.

[0003] At present, early warning of COPD can be achieved by detecting the patient's respiratory status. The existing respiratory status warning methods have a short monitoring time and are mostly limited to daytime monitoring, resulting in low accuracy of respiratory status warning and high price. Summary of the invention

[0004] Based on this, the purpose of the present invention is to provide a respiratory status warning method, device, computer equipment and storage medium based on vital sign signals, which can reduce costs and improve the accuracy of respiratory status warning.

[0005] In a first aspect, an embodiment of the present application provides a respiratory status warning method based on a vital sign signal, comprising the following steps:

[0006] Obtaining a vital sign signal of a sample user at a preset first time scale;

[0007] Extracting a ballistocardiogram signal and a respiration signal from the vital sign signals of the sample user;

[0008] Determine a characteristic sequence corresponding to the ballistocardiogram signal and the breathing signal of the sample user;

[0009] Determining differential characteristic sequences of the characteristic sequence at a plurality of different second time scales;

[0010] Inputting the differential feature sequence of the sample user into a preset respiratory state classification model to obtain the respiratory state sequence of the sample user;

[0011] According to the differential feature sequence and the respiratory state sequence of the sample user, the model parameter data of the respiratory state classification model is updated, and according to the updated model parameter data, the differential feature sequence and a preset number of iterations, the model parameter data of the respiratory state classification model is repeatedly updated to obtain a target respiratory state classification model;

[0012] Obtain a ballistocardiogram signal and a respiratory signal of a user to be tested, determine a differential feature sequence corresponding to the ballistocardiogram signal and the respiratory signal of the user to be tested, input the differential feature sequence corresponding to the ballistocardiogram signal and the respiratory signal of the user to be tested into the target respiratory state classification model, and obtain the respiratory state sequence of the user to be tested; and obtain a respiratory state warning result of the user to be tested according to the number of target respiratory state vectors in the respiratory state sequence of the user to be tested.

[0013] In a second aspect, an embodiment of the present application provides a respiratory status warning device based on a vital sign signal, comprising:

[0014] A vital sign signal acquisition module, used to obtain a vital sign signal of a sample user at a preset first time scale;

[0015] A signal extraction module, used to extract a ballistocardiogram signal and a breathing signal from the vital sign signal of the sample user;

[0016] A feature sequence determination module, used to determine the feature sequence corresponding to the ballistocardiogram signal and the breathing signal of the sample user;

[0017] A differential feature sequence determination module, used to determine the differential feature sequences of the feature sequence at a plurality of different second time scales;

[0018] A breathing state sequence acquisition module, used for inputting the differential feature sequence of the sample user into a preset breathing state classification model to obtain the breathing state sequence of the sample user;

[0019] A model updating module, used for updating the model parameter data of the respiratory state classification model according to the differential feature sequence and the respiratory state sequence of the sample user, and repeatedly updating the model parameter data of the respiratory state classification model according to the updated model parameter data, the differential feature sequence and a preset number of iterations to obtain a target respiratory state classification model;

[0020] A respiratory state warning module is used to obtain the ballistocardiogram signal and the respiratory signal of the user to be tested, determine the differential feature sequence corresponding to the ballistocardiogram signal and the respiratory signal of the user to be tested, input the differential feature sequence corresponding to the ballistocardiogram signal and the respiratory signal of the user to be tested into the target respiratory state classification model to obtain the respiratory state sequence of the user to be tested; and obtain the respiratory state warning result of the user to be tested according to the number of target respiratory state vectors in the respiratory state sequence of the user to be tested.

[0021] In a third aspect, an embodiment of the present application provides a computer device comprising: a processor, a memory, and a computer program stored in the memory and executable on the processor; when the computer program is executed by the processor, the steps of the respiratory status warning method based on vital sign signals as described in the first aspect are implemented.

[0022] In a fourth aspect, an embodiment of the present application provides a storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the respiratory status warning method based on vital sign signals as described in the first aspect are implemented.

[0023] In the present embodiment, a respiratory state warning method, apparatus, computer equipment and storage medium based on vital sign signals are provided, which are performed by obtaining the vital sign signals of a sample user at a preset first time scale; extracting a ballistocardiogram signal and a respiratory signal from the vital sign signals of the sample user; determining the feature sequences corresponding to the ballistocardiogram signal and the respiratory signal of the sample user; determining the differential feature sequences of the feature sequences at several different second time scales; inputting the differential feature sequences of the sample user into a preset respiratory state classification model to obtain the respiratory state sequence of the sample user; and according to the differential feature sequences and the respiratory state sequence of the sample user, classifying the model of the respiratory state classification model. The parameter data is updated, and the model parameter data of the respiratory state classification model is repeatedly updated according to the updated model parameter data, the differential feature sequence and the preset number of iterations to obtain the target respiratory state classification model; the ballistocardiogram signal and the respiratory signal of the user to be tested are obtained, and the differential feature sequence corresponding to the ballistocardiogram signal and the respiratory signal of the user to be tested is determined, and the differential feature sequence corresponding to the ballistocardiogram signal and the respiratory signal of the user to be tested is input into the target respiratory state classification model to obtain the respiratory state sequence of the user to be tested; according to the number of target respiratory state vectors in the respiratory state sequence of the user to be tested, the respiratory state warning result of the user to be tested is obtained. It can reduce costs and improve the accuracy of respiratory state warning.

[0024] For better understanding and implementation, the present invention is described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A schematic diagram of a flow chart of a respiratory status warning method based on a vital sign signal provided in one embodiment of the present application;

[0026] Figure 2 A schematic diagram of the flow of S2 in a respiratory status warning method based on a vital sign signal provided in one embodiment of the present application;

[0027] Figure 3A schematic diagram of the flow of S3 in a respiratory status warning method based on a vital sign signal provided in one embodiment of the present application;

[0028] Figure 4 A schematic diagram of the flow of S4 in a respiratory status warning method based on a vital sign signal provided in one embodiment of the present application;

[0029] Figure 5 A schematic flow chart of a respiratory status warning method based on a vital sign signal provided in another embodiment of the present application;

[0030] Figure 6 A schematic diagram of the flow of S6 in a respiratory status warning method based on a vital sign signal provided in one embodiment of the present application;

[0031] Figure 7 A schematic diagram of the flow of S61 in a respiratory status warning method based on a vital sign signal provided in one embodiment of the present application;

[0032] Figure 8 A schematic diagram of the structure of a respiratory status warning device based on a vital sign signal provided in one embodiment of the present application;

[0033] Fig. 9 A schematic diagram of the structure of a computer device provided for one embodiment of the present application. DETAILED DESCRIPTION

[0034] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0035] The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit this application. The singular forms of "a", "said" and "the" used in this application and the appended claims are also intended to include plural forms unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0036] It should be understood that, although the terms first, second, third, etc. may be used in the present application to describe various information, these information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the words "if" / "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determination".

[0037] See also Figure 1 , Figure 1 A flowchart of a respiratory status warning method based on a vital sign signal provided in an embodiment of the present application is provided, and the method comprises the following steps:

[0038] S1: Obtaining a vital sign signal of a sample user at a preset first time scale.

[0039] The executor of the respiratory state warning method based on vital sign signals of the present application is a warning device of the respiratory state warning method based on vital sign signals (hereinafter referred to as the warning device). In an optional embodiment, the warning device can be a computer device, a server, or a server cluster composed of multiple computer devices.

[0040] In this embodiment, the early warning device obtains a multi-dimensional data stream to be processed. Specifically, the multi-dimensional data stream is a real-time aviation data stream, including aviation parameters of several dimensions to reflect the real-time flight status of the aircraft.

[0041] The early warning device can obtain the vital signs signals of the sample users at several preset different time scales by querying in a preset database. The early warning device can also use a non-intrusive sensor to collect the vital signs signals of the sample users at several preset different time scales in real time. Among them, the non-intrusive sensor includes a signal acquisition module and a data storage module.

[0042] Specifically, the signal acquisition module is placed under the user's pillow. When the user sleeps on the pillow, the user's body will vibrate slightly due to cardiac activity, respiratory activity, etc., and the center of gravity will shift, generating a force signal. The signal acquisition module can convert the force signal into an analog electrical signal, and then filter, amplify, and A / D convert the analog electrical signal into a digital signal with a sampling rate of 1000Hz through the built-in filter circuit, amplifier circuit, and A / D conversion circuit, which is the vital sign signal.

[0043] The preset first time scale is the average duration of time that the user uses the detection device. For example, the preset first time scale is 100 days.

[0044] S2: Extracting a ballistocardiogram signal and a breathing signal from the vital sign signal of the sample user.

[0045] The vital sign signal is a mixed signal of the ballistocardiogram signal and the respiratory signal. In this embodiment, the early warning device extracts the ballistocardiogram signal and the respiratory signal from the vital sign signal of the sample user.

[0046] See also Figure 2 , Figure 2 The flowchart of S2 in the respiratory state warning method based on vital sign signals provided in one embodiment of the present application includes steps S21 to S25, which are as follows:

[0047] S21: Filter the vital sign signal using a filter to obtain a filtered vital sign signal.

[0048] In this embodiment, the early warning device uses a second-order Butterworth bandpass filter with a cutoff frequency of 2 to 20 Hz to remove noise from several vital sign signals of different time scales. Furthermore, the power frequency interference and Gaussian additive noise are removed by a fourth-order Butterworth bandpass filter with a cutoff frequency of 20 Hz. Furthermore, with 15s as a fixed time window and a sliding scale of 1s, the peak-to-average ratio of the time window signal is detected to detect body movement signals. When the signal peak-to-average ratio is greater than the threshold, it is considered that there is a body movement signal in the time window, and the signal in the time window is cleared and re-incorporated into the new queue signal.

[0049] S22: performing an opening operation of first corroding and then dilating the filtered vital sign signal to obtain a first vital sign signal.

[0050] S23: performing a closing operation of first dilation and then erosion on the filtered vital sign signal to obtain a second vital sign signal.

[0051] S24: Calculate an average of the first vital sign signal and the second vital sign signal, and use the average result as a breathing signal.

[0052] S25: Subtract the filtered vital sign signal from the respiratory signal to obtain a ballistocardiogram signal.

[0053] In this embodiment, on the basis of excluding the body motion signal, due to the difference in the main frequency bands of the respiratory signal and the cardiac shock signal, the early warning device filters out the contribution of the respiratory signal to the cardiac shock signal through morphological filtering, and then extracts the respiratory signal. Since the respiratory signal and the cardiac shock signal are superimposed, the cardiac shock signal can be obtained by subtracting the respiratory signal from the filtered vital sign signal.

[0054] S3: Determine the characteristic sequence corresponding to the ballistocardiogram signal and the breathing signal of the sample user.

[0055] Heart rate (HR) refers to the number of heartbeats per minute in a normal person at rest. Heart rate variability (HRV) refers to the variation in the difference between successive heartbeats. Respiratory rate (RR) refers to the number of breaths per minute.

[0056] In this embodiment, the early warning device determines the characteristic sequences corresponding to the ballistocardiogram signal and the breathing signal of the sample user, wherein the characteristic sequences include a heart rate characteristic sequence, a heart rate variability characteristic sequence, a breathing frequency characteristic sequence, and a sleep characteristic sequence;

[0057] See also Figure 3 , Figure 3 The flowchart of S3 in the respiratory state warning method based on vital sign signals provided in one embodiment of the present application includes steps S31 to S36, which are specifically as follows:

[0058] S31: performing a timing bidirectional detection on the ballistocardiogram signal according to a preset ballistocardiogram signal template, determining the peak point coordinates of the ballistocardiogram signal, and obtaining a plurality of cardiac intervals according to the peak point coordinates.

[0059] In this embodiment, the early warning device establishes a preset ballistocardiogram signal template according to the morphological characteristics of the ballistocardiogram signal, and then performs a timing bidirectional detection on the ballistocardiogram signal according to the preset ballistocardiogram signal template to locate the "J" peak position. The located "J" peak position is further refined and calibrated by setting the refractory period and the interval stability to obtain the final "J" peak position. The corresponding "JJ" interval value can be obtained by differential calculation of several front and rear "J" peaks, that is, several cardiac intervals are obtained.

[0060] S32: obtaining a plurality of heart rates according to the plurality of cardiac intervals; obtaining heart rate data parameters corresponding to the plurality of heart rates, and using the heart rate data parameters as a heart rate feature sequence.

[0061] In this embodiment, the early warning device can calculate several heart rates through several cardiac intervals. The calculation formula of the heart rate per minute is as follows:

[0062]

[0063] in, Indicates heart rate, Indicates the cardiac interval.

[0064] After obtaining several heart rates, the early warning device can calculate the mode, mean, maximum and minimum of several heart rates to obtain a heart rate feature sequence. Specifically, taking a night corresponding to 5 hours of sleep as an example, 5 minutes is a unit time, and 5 hours are divided into 60 unit time. An average cardiac interval can be obtained every 5 minutes, so that 60 average cardiac intervals can be obtained throughout the night. The corresponding heart rate is calculated for the 60 average cardiac intervals, and then the mode, mean, maximum and minimum of the heart rate are obtained. The mode, mean, maximum and minimum of the heart rate are combined to obtain the heart rate data parameters.

[0065] S33: Calculate the standard deviation of the plurality of cardiac intervals, and obtain a plurality of power spectra according to the standard deviation and a preset autoregressive spectrum estimation method; compare the plurality of power spectra with a preset spectrum range to determine the spectrum range in which each power spectrum is located; obtain the heart rate deceleration force according to the plurality of cardiac intervals; obtain the heart rate variability characteristic sequence according to the average value of all power spectra in the same spectrum range and the heart rate deceleration force;

[0066] In this embodiment, the early warning device can calculate the corresponding standard deviation based on a number of cardiac intervals, and the calculation formula is as follows:

[0067]

[0068] in, N represents the total number of cardiac intervals, express N The average cardiac interval, Indicates i Specifically, the average value of 60 average cardiac intervals is calculated, and then the standard deviation is calculated.

[0069] The autoregressive spectrum estimation method is an important spectrum estimation method with the advantages of high frequency resolution and improved spectrum line splitting and frequency offset. The early warning device performs frequency domain analysis on each standard deviation through the preset autoregressive spectrum estimation method to obtain several power spectra. According to the preset spectrum range, several power spectra are divided into four frequency bands, including high frequency band (High Frequency, HF): 0.15Hz~0.4Hz, low frequency band (Low Frequency, LF): 0.04Hz~0.15Hz, very low frequency band (VLF): 0.003Hz~0.04Hz, ultra low frequency band (ULF): <0.003Hz. The power of the above four frequency bands is used as the HRV frequency domain features.

[0070] The early warning device obtains the heart rate deceleration capacity (DC) based on several cardiac intervals. Specifically, the heart rate deceleration capacity calculation formula is as follows:

[0071] DC=[X(0)+X(1)-X(-1)-X(-2)]*1 / 4

[0072] Among them, X(0) is the mean of all cardiac intervals of the deceleration center point; X(1) is the mean of the first cardiac interval immediately to the right of the deceleration center point; X(-1) is the mean of the first cardiac interval adjacent to the left of the deceleration center point; X(-2) is the mean of all second cardiac intervals adjacent to the left of the deceleration center point.

[0073] The early warning device uses the frequency band powers LF, HF, ULF, VLF, the ratio of LF to HF, and DC as the characteristic sequence of heart rate variability.

[0074] S34: Determine a number of respiratory intervals of the respiratory signal according to a preset zero-crossing point detection method.

[0075] In this embodiment, the early warning device adopts a zero-crossing detection method, by detecting the zero-crossing points of the respiratory signal in two adjacent rising edges as a respiratory interval, that is, the end point of one respiratory signal is regarded as the starting point of the next respiratory signal.

[0076] S35: obtaining a plurality of respiratory frequencies according to the plurality of respiratory intervals; obtaining respiratory data parameters corresponding to the plurality of respiratory frequencies, and using the respiratory data parameters as a respiratory frequency feature sequence.

[0077] The respiratory data parameters include the mode of respiratory frequency, the mean of respiratory frequency, the maximum value of respiratory frequency, the minimum value of respiratory frequency, the mean of respiratory frequency greater than or equal to a first preset threshold, and the mean of respiratory frequency less than a second preset threshold.

[0078] In this embodiment, the calculation formula of the respiratory rate per minute is as follows:

[0079]

[0080] in, Respiratory rate, Indicates the breathing interval.

[0081] After obtaining several respiratory frequencies, the early warning device can calculate the mode RR1, mean RR2, maximum RR3, and minimum RR4 of the respiratory frequencies. At the same time, several respiratory frequencies are compared with the first preset threshold and the second preset threshold to determine all respiratory frequencies greater than or equal to the first preset threshold and all respiratory frequencies less than the second preset threshold. All respiratory frequencies greater than or equal to the first preset threshold are averaged to obtain an average result RR5. Among them, the first preset threshold is 20 times / minute, and RR5 is used to indicate that the respiratory frequency is too fast. All respiratory frequencies less than the second preset threshold are averaged to obtain an average result RR6. Among them, the second preset threshold is 9 times / minute, and RR6 is used to indicate that the respiratory frequency is too slow. RR1, RR2, RR3, RR4, RR5 and RR6 are used as respiratory frequency feature sequences.

[0082] S36: Obtain a sleep characteristic sequence according to a plurality of cardiac intervals, a plurality of respiratory intervals and a preset cardiopulmonary coupling method.

[0083] In this embodiment, the sleep characteristic sequence includes the time of falling asleep, the time of waking up, the total sleep duration, and the effective sleep duration (i.e., the total sleep duration minus the sleep waking duration). With the help of cardiac intervals and respiratory intervals, the early warning device uses a preset cardiopulmonary coupling method to calculate the user's sleep stages at different time scales, thereby obtaining the sleep characteristic sequence at the corresponding time scale.

[0084] S4: Determine differential feature sequences of the heart rate feature sequence, the heart rate variability feature sequence, the respiratory frequency feature sequence, and the sleep feature sequence of the sample user at several different second time scales.

[0085] The second time scale is a continuous time scale, and the several different second time scales may be 3 consecutive days, 5 consecutive days, 7 consecutive days, 10 consecutive days, 14 consecutive days and 30 consecutive days.

[0086] In this embodiment, after obtaining the heart rate feature sequence, heart rate variability feature sequence, respiratory rate feature sequence and sleep feature sequence at a preset first time scale, the early warning device can further obtain differential feature sequences at several different second time scales, wherein the differential feature sequence includes differential feature vectors of several time points, and the differential feature vectors include first-order differential feature vectors and second-order differential feature vectors.

[0087] See also Figure 4 , Figure 4 The flowchart of S4 in the respiratory state warning method based on vital sign signals provided in one embodiment of the present application includes step S41, which is as follows:

[0088] S41: traverse each second time scale, calculate the first-order difference original value and the second-order difference original value of the feature sequence at the current second time scale, perform ordinal variable encoding conversion on the first-order difference original value and the second-order difference original value, obtain the first-order difference encoding value and the second-order difference encoding value, respectively as the first-order difference feature vector and the second-order difference feature vector at the current second time scale, until the first-order difference feature vector and the second-order difference feature vector corresponding to the feature sequence at all second time scales are determined, and the differential feature sequence of the feature sequence at several different second time scales is determined.

[0089] In this embodiment, for the cases where the second time scale is 3 consecutive days, 5 consecutive days, 7 consecutive days, 10 consecutive days, 14 consecutive days and 30 consecutive days, the early warning device calculates the first-order differential feature vectors and the second-order differential feature vectors in the corresponding differential feature sequences.

[0090] Take the first-order difference feature vector in the differential feature sequence corresponding to HR1 in the heart rate feature sequence for the second time scale of 5 consecutive days as an example for explanation. Under the preset first time scale, the HR1 of each day is known, and the average value is calculated after removing the maximum and minimum values ​​according to the HR1 of 3 consecutive days, and the first-order difference calculation is performed on the average value obtained to obtain the first-order difference original value, and the first-order difference original value is subjected to the ordinal variable coding conversion to obtain the first-order difference coding value. Specifically, if the first-order difference original value is negative, the first-order difference coding value of the first-order difference original value subjected to the ordinal variable coding conversion is 0, indicating that the indicator shows a downward trend; if the first-order difference original value is 0, the first-order difference coding value of the first-order difference original value subjected to the ordinal variable coding conversion is 1, indicating that the indicator remains stable and has no significant changes; if the first-order difference original value is positive, the first-order difference coding value of the first-order difference original value subjected to the ordinal variable coding conversion is 2, indicating that the indicator shows an upward trend, as the value of each day in the first-order difference feature sequence.

[0091] For the value of the first day in the first-order difference feature sequence, only the HR1 of the first day is known, and there is no HR1 of the second and third days. Therefore, the first-order difference encoding value of the first day in the first-order difference feature sequence is based on 0. For the value of the second day in the first-order difference feature sequence, only the HR1 of the first and second days is known, and there is no HR1 of the third day. Therefore, the value of the second day in the first-order difference feature sequence is 0. For the value of the third day in the first-order difference feature sequence, since the HR1 of the first, second and third days are all known, the first-order difference original value is calculated based on the HR1 of the first, second and third days. The ordinal variable coding conversion is obtained to obtain the first-order difference coding value X1 of the third day in the first-order difference feature sequence. The first-order difference original value is calculated based on the HR1 of the second, third and fourth days. The ordinal variable coding conversion is obtained to obtain the first-order difference coding value X2 of the fourth day in the first-order difference feature sequence, ..., the first-order difference original value is calculated based on the HR1 of the 98th, 99th and 100th days. The ordinal variable coding conversion is obtained to obtain the first-order difference coding value X98 of the 100th day in the first-order difference feature sequence, which are respectively used as the first-order difference feature vectors, thereby obtaining the first-order difference feature sequence of HR1 of the heart rate feature sequence for 3 consecutive days.

[0092] Repeat the process at each second time scale until the first-order differential feature sequences and second-order differential feature sequences corresponding to the heart rate feature sequence, the heart rate variability feature sequence, the respiratory frequency feature sequence, and the sleep feature sequence at all second time scales are determined, and the differential feature sequences of the feature sequence at several different second time scales are determined. The solution process for the first-order differential feature sequences and second-order differential feature sequences corresponding to the heart rate feature sequence, the heart rate variability feature sequence, the respiratory frequency feature sequence, and the sleep feature sequence at different second time scales is consistent with the solution process of the above example, and will not be repeated here.

[0093] See also Figure 5 , Figure 5 The flowchart of the respiratory state warning method based on the vital sign signal provided in another embodiment of the present application further includes step S8, which is before step S4 and is specifically as follows:

[0094] S8: adopting recursive feature elimination and combining with a stratified cross-validation method, feature selection is performed on the differential feature sequences of the sample users at each second time scale, and the differential feature sequences of the sample users at several different second time scales after feature selection are obtained.

[0095] Recursive Feature Elimination (RFE) is a feature selection method that streamlines the feature set by building multiple models and gradually removing the features that contribute the least to model performance. Stratified Cross-Validation (SCV) is a model evaluation technique that ensures the accuracy of model evaluation by maintaining the original distribution of data categories in each validation fold, and is particularly suitable for processing datasets with imbalanced categories.

[0096] In this embodiment, the early warning device adopts recursive feature elimination and combines the stratified cross-validation method to construct a pre-trained feature selection model, inputs the differential feature sequences of the sample users at several different second time scales into the pre-trained feature selection model, performs feature selection on the differential feature sequences of the sample users at each second time scale, and obtains the differential feature sequences of the sample users at several different second time scales after feature selection.

[0097] S5: Inputting the differential feature sequence of the sample user into a preset respiratory state classification model to obtain the respiratory state sequence of the sample user.

[0098] The respiratory state classification model adopts the Hidden Markov Model (HMM), which is a statistical model used to describe a Markov process with hidden unknown parameters. In the hidden Markov model, the state is not directly visible, but the output depends on the state and is visible. Each state has a possible probability distribution through possible output symbols. The hidden layer state transfer and symbol output form a dual random process relationship.

[0099] In this embodiment, the early warning device inputs the differential feature sequence of the sample user into a preset respiratory state classification model, and uses the Viterbi algorithm based on the model parameter data of the respiratory state classification model to be trained and the differential feature sequence to obtain the respiratory state sequence of the sample user, wherein the respiratory state sequence includes respiratory state vectors of the respiratory states at several time points.

[0100] S6: According to the differential feature sequence and the respiratory state sequence of the sample user, the model parameter data of the respiratory state classification model is updated; according to the updated model parameter data, the differential feature sequence and a preset number of iterations, the model parameter data of the respiratory state classification model is repeatedly updated to obtain a target respiratory state classification model.

[0101] In this embodiment, the early warning device updates the model parameter data of the respiratory state classification model according to the differential feature sequence and the respiratory state sequence of the sample user, wherein the model parameter data includes an observation probability matrix, a state transition probability matrix and an initial state probability matrix, and the observation probability matrix includes the observation probability of the respiratory state at several time points, and the observation probability is i The expected number of times a particular observation is observed is divided by the respiratory state i The total expected time of the breathing state is the state that cannot be directly observed in the model. They represent the internal state that the system may be in, but these states cannot be directly observed. Each breathing state i A period corresponding to a time point in a sequence, including a stable period and a prodromal period of acute exacerbation.

[0102] The state transition probability matrix is ​​the transition probability between the respiratory states at several time points, and the transition probability is the transition probability between the respiratory states at a given time point. i Under the condition of j The expected number of times divided by the breathing state i The total expected time.

[0103] The initial state probability matrix includes initial state probabilities of respiratory states, wherein the initial state probabilities are used to indicate the probability of the respiratory state classification model making observations at an initial time point under the corresponding respiratory state at the initial time point.

[0104] The early warning device repeatedly updates the model parameter data of the respiratory state classification model according to the updated model parameter data, the differential feature sequence and a preset number of iterations to obtain a target respiratory state classification model.

[0105] See also Figure 6 , Figure 6 The flowchart of S6 in the respiratory state warning method based on vital sign signals provided in one embodiment of the present application includes steps S61 to S65, which are specifically as follows:

[0106] S61: Calculate the expected probability of the respiratory state at several time points of the differential feature sequence according to the initial state probability matrix, the differential feature sequence and the respiratory state sequence.

[0107] In this embodiment, the early warning device calculates the expected probability of the respiratory state at several time points of the differential feature sequence based on the initial state probability matrix, the differential feature sequence and the respiratory state sequence, wherein the expected probability is used to indicate the probability of being in the corresponding respiratory state at the current time point.

[0108] See also Figure 7 , Figure 7 The flowchart of S61 in the respiratory state warning method based on the vital sign signal provided in one embodiment of the present application includes steps S611 to S613, which are as follows:

[0109] S611: Calculate the observation probability based on the first-order difference feature sequence and the second-order difference feature sequence and the respiratory state sequence, respectively, to obtain the first-order difference observation probability and the second-order difference observation probability of the respiratory state at several time points.

[0110] In this embodiment, the early warning device obtains the first-order differential indication vectors of several time points according to the first-order differential feature sequence and the preset indicator function; performs observation probability calculation according to the first-order differential feature sequence and the respiratory state sequence to obtain the first-order sub-difference observation probability of the respiratory state at several time points; multiplies the first-order differential indication vector of the same time point with the first-order sub-difference observation probability of the respiratory state at several time points to obtain the first-order differential observation probability of the respiratory state at several time points.

[0111] The early warning device obtains the second-order differential indicator vectors of several time points according to the second-order differential feature sequence and the preset indicator function; performs observation probability calculation according to the second-order differential feature sequence and the respiratory state sequence to obtain the second-order sub-difference observation probability of the respiratory state at several time points; multiplies the second-order differential indicator vector of the same time point with the second-order sub-difference observation probability of the respiratory state at several time points to obtain the second-order differential observation probability of the respiratory state at several time points.

[0112] S612: Add the first-order difference observation probability and the second-order difference observation probability of the respiratory state at the same time point to obtain the joint observation probability of the respiratory state at several time points, and obtain the total observation probability based on the joint observation probability of the respiratory state at several time points.

[0113] In this embodiment, the early warning device adds the first-order difference observation probability and the second-order difference observation probability of the respiratory state at the same time point to obtain the joint observation probability of the respiratory state at several time points, wherein the joint observation probability is:

[0114]

[0115] In the formula, For the t The joint observation probability of the respiratory state at each time point is For the t The first-order difference eigenvector of time points, For the t The second-order difference eigenvector at each time point, fort The respiratory state vector at each time point is are the preset hyperparameters; is the indicator function, For the t The first-order difference indicator vector at each time point, For the t The second-order difference indicator vector at each time point, For the t The first-order sub-difference observation probability of the respiratory state at a time point, For the t The second-order sub-difference observation probability of the respiratory state at a time point.

[0116] The early warning device multiplies and accumulates the joint observation probabilities of the respiratory states at several time points according to the joint observation probabilities of the respiratory states at several time points to obtain a total observation probability, wherein the total observation probability is:

[0117]

[0118] In the formula, is the total observation probability, O is the differential feature sequence; Z is the respiratory state sequence, T is the total number of time points.

[0119] S613: Obtain the observation probability of the differential feature sequence, and obtain the expected probability of the respiratory state at several time points according to the observation probability of the differential feature sequence, the initial state probability matrix and the total observation probability.

[0120] In this embodiment, the early warning device obtains the observation probability of the differential feature sequence, and obtains the expected probability of the respiratory state at several time points according to the observation probability of the differential feature sequence, the initial state probability matrix and the total observation probability, wherein the expected probability is:

[0121]

[0122] In the formula, For the t Respiratory status at each time point j The expected probability of For a given breathing state j and hyperparameters The differential feature sequence is observed under O The observation probability of is the breathing state given in the initial state probability matrix j and hyperparameters The initial state probability vector under .

[0123] S62: According to the differential feature vectors at several time points in the differential feature sequence and the expected probability of the respiratory state at several time points, the observation probability of the differential feature vectors of the respiratory state at several time points is obtained, and the observation probability of the differential feature vectors of the respiratory state at the corresponding time points in the observation probability matrix is ​​updated.

[0124] In this embodiment, the early warning device obtains the differential feature vector observation probability of the respiratory state at several time points according to the differential feature vectors at several time points in the differential feature sequence, the expected probability of the respiratory state at several time points and the preset differential feature vector observation probability calculation algorithm. The early warning device updates the differential feature vector observation probability of the respiratory state at the corresponding time point in the observation probability matrix according to the differential feature vector observation probability of the respiratory state at several time points, wherein the differential feature vector observation probability calculation algorithm is:

[0125]

[0126] In the formula, For breathing state j The observation probability of the difference eigenvector of .

[0127] S63: Obtain forward probabilities and backward probabilities of the respiratory states at several time points, and estimate the conditional probabilities between the respiratory states at several time points based on the product of the forward probabilities of the respiratory states at the corresponding time points.

[0128] In this embodiment, the early warning device obtains the forward probability and backward probability of the respiratory state at several time points. Specifically, the early warning device pre-sets the forward probability of the respiratory state at the initial time point and the backward probability of the respiratory state at the final time point. The early warning device obtains the emission probability of the respiratory state at several time points. According to the emission probability, the state transition probability matrix before updating and the preset forward probability calculation algorithm, the emission probability of the respiratory state at the current time point, the forward probability of the respiratory state at the previous time point of the current time point and the transition probability between several respiratory states at the current time point in the state transition probability matrix before updating are multiplied and summed to obtain the forward probability of the respiratory state at several time points. The forward probability calculation algorithm is:

[0129]

[0130] In the formula, For the t Respiratory status at each time point i The forward probability of For the t - Respiratory status at 1 time pointj The forward probability of is the breathing state at the current time point in the state transition probability matrix before updating i Respiratory status j The transition probability between For the t Respiratory status at each time point i The probability of emission.

[0131] The early warning device multiplies and sums the emission probability of the respiratory state at a time point after the current time point, the backward probability of the respiratory state at two time points after the current time point, and the transition probability between several respiratory states at a time point after the current time point in the state transition probability matrix before updating, according to the emission probability, the state transition probability matrix before updating, and the preset backward probability calculation algorithm, to obtain the backward probability of the respiratory state at several time points, wherein the backward probability calculation algorithm is:

[0132]

[0133] In the formula, For the t +1 time point of breathing status j The backward probability of For the t + Respiratory status at 2 time points i The backward probability of For the t +1 time point of breathing status i The probability of emission.

[0134] The early warning device estimates the product of the forward probability of the respiratory state at several time points and the backward probability of the respiratory state at the corresponding time points to obtain the conditional probability between the respiratory states at several time points. Specifically, the early warning device estimates the product of the forward probability of the respiratory state at the current time point and the backward probability of the respiratory state at the next time point to obtain the conditional probability between the respiratory state at the current time point and the next time point, wherein the conditional probability is:

[0135]

[0136] In the formula, For the t +1 time point of breathing status i Respiratory status j The conditional probability between is the estimated symbol.

[0137] S64: obtaining state transition probabilities between the several breathing states according to the expected probabilities of the breathing states at the several time points and the conditional probabilities between the breathing states at the corresponding time points, and updating the state transition probabilities between the several breathing states in the state transition probability matrix.

[0138] In this embodiment, the early warning device obtains the state transition probability between several breathing states according to the expected probability of the breathing state at several time points, the conditional probability between the breathing states at the corresponding time points, and the preset state transition probability calculation algorithm, and updates the state transition probability between several breathing states in the state transition probability matrix, wherein the state transition probability calculation algorithm is:

[0139]

[0140] In the formula, For breathing state i Respiratory status j The state transition probability between .

[0141] S65: Based on the expected probability of the respiratory state at the first time point of the differential feature sequence as the initial state probability, the initial state probability of the respiratory state in the initial state probability matrix is ​​updated.

[0142] In this embodiment, the early warning device updates the initial state probability of the respiratory state in the initial state probability matrix according to the expected probability of the respiratory state at the first time point of the differential feature sequence as the initial state probability, wherein the initial state probability is:

[0143]

[0144] In the formula, For breathing state i The initial state probability, is the respiratory status at the first time point i The expected probability of .

[0145] S7: Obtain the ballistocardiogram signal and the breathing signal of the user to be tested, determine the differential feature sequence corresponding to the ballistocardiogram signal and the breathing signal of the user to be tested, input the differential feature sequence corresponding to the ballistocardiogram signal and the breathing signal of the user to be tested into the target respiratory state classification model, and obtain the respiratory state sequence of the user to be tested; obtain the respiratory state warning result of the user to be tested according to the number of target respiratory state vectors in the respiratory state sequence of the user to be tested.

[0146] In this embodiment, the early warning device obtains the ballistocardiogram signal and the breathing signal of the user to be tested, and determines the differential feature sequence corresponding to the ballistocardiogram signal and the breathing signal of the user to be tested.

[0147] The early warning device inputs the differential feature sequence corresponding to the ballistocardiogram signal and the breathing signal of the user to be tested into the target breathing state classification model to obtain the breathing state sequence of the user to be tested; the early warning device obtains the breathing state early warning result of the user to be tested according to the number of target breathing state vectors in the breathing state sequence of the user to be tested. Specifically, the early warning device counts and accumulates the number of breathing state vectors of the target breathing state in the breathing state sequence to obtain the total target breathing state number, and obtains the breathing state early warning result of the user to be tested according to the total target breathing state number and a preset early warning number threshold.

[0148] Please refer to Figure 8 , Figure 8 This is a schematic diagram of the structure of a respiratory state warning device based on a vital sign signal provided by an embodiment of the present application. The device can implement all or part of the respiratory state warning device based on a vital sign signal through software, hardware, or a combination of both. The device 8 includes:

[0149] A vital sign signal acquisition module 81 is used to obtain a vital sign signal of a sample user at a preset first time scale;

[0150] A signal extraction module 82, used to extract a ballistocardiogram signal and a respiratory signal from the vital sign signal of the sample user;

[0151] A feature sequence determination module 83, used to determine the feature sequence corresponding to the ballistocardiogram signal and the breathing signal of the sample user;

[0152] A differential feature sequence determination module 84 is used to determine the differential feature sequences of the feature sequence at a plurality of different second time scales;

[0153] A breathing state sequence acquisition module 85 is used to input the differential feature sequence of the sample user into a preset breathing state classification model to obtain the breathing state sequence of the sample user;

[0154] A model training module 86 is used to update the model parameter data of the respiratory state classification model according to the differential feature sequence and the respiratory state sequence of the sample user, and repeatedly update the model parameter data of the respiratory state classification model according to the updated model parameter data, the differential feature sequence and a preset number of iterations to obtain a target respiratory state classification model;

[0155] The respiratory state warning module 87 is used to obtain the ballistocardiogram signal and the respiratory signal of the user to be tested, determine the differential feature sequence corresponding to the ballistocardiogram signal and the respiratory signal of the user to be tested, input the differential feature sequence corresponding to the ballistocardiogram signal and the respiratory signal of the user to be tested into the target respiratory state classification model, and obtain the respiratory state sequence of the user to be tested; and obtain the respiratory state warning result of the user to be tested according to the number of target respiratory state vectors in the respiratory state sequence of the user to be tested.

[0156] In this embodiment, a vital sign signal of a sample user at a preset first time scale is obtained through a vital sign signal acquisition module; a ballistocardiogram signal and a respiratory signal are extracted from the vital sign signal of the sample user through a signal extraction module; a feature sequence determination module is used to determine a feature sequence corresponding to the ballistocardiogram signal and the respiratory signal of the sample user; a differential feature sequence determination module is used to determine a differential feature sequence of the feature sequence at several different second time scales; a respiratory state sequence acquisition module is used to input the differential feature sequence of the sample user into a preset respiratory state classification model to obtain the respiratory state sequence of the sample user; and a model training module is used to train the respiratory state sequence according to the differential feature sequence and the respiratory state sequence of the sample user. The model parameter data of the state classification model is updated, and the model parameter data of the respiratory state classification model is repeatedly updated according to the updated model parameter data, the differential feature sequence and the preset number of iterations to obtain the target respiratory state classification model; the ballistocardiogram signal and the respiratory signal of the user to be tested are obtained through the respiratory state warning module, the differential feature sequence corresponding to the ballistocardiogram signal and the respiratory signal of the user to be tested is determined, and the differential feature sequence corresponding to the ballistocardiogram signal and the respiratory signal of the user to be tested is input into the target respiratory state classification model to obtain the respiratory state sequence of the user to be tested; according to the number of target respiratory state vectors in the respiratory state sequence of the user to be tested, the respiratory state warning result of the user to be tested is obtained. It can reduce costs and improve the accuracy of respiratory state warning.

[0157] Please refer to Fig. 9 , Fig. 9 The computer device 9 is a schematic diagram of a structure of a computer device provided in an embodiment of the present application. The computer device 9 includes: a processor 91, a memory 92, and a computer program 93 stored in the memory 92 and executable on the processor 91; the computer device may store a plurality of instructions, which are suitable for being loaded and executed by the processor 91 Figures 1 to 7 For the specific implementation process, please refer to Figures 1 to 7 The specific description is not repeated here.

[0158] Among them, the processor 91 may include one or more processing cores. The processor 91 uses various interfaces and lines to connect various parts in the server, and executes various functions and processes data of the respiratory state warning device 8 based on the vital sign signal by running or executing instructions, programs, code sets or instruction sets stored in the memory 92, and calling the data in the memory 92. Optionally, the processor 91 can be implemented in at least one hardware form of digital signal processing (Digital Signal Processing, DSP), field programmable gate array (Field-Programmable Gate Array, FPGA), and programmable logic array (Programble Logic Array, PLA). The processor 91 can integrate one or more combinations of a central processing unit 91 (Central Processing Unit, CPU), an image processor 91 (Graphics Processing Unit, GPU) and a modem. Among them, the CPU mainly processes the operating system, user interface and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the touch display screen; and the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 91, but implemented separately through a chip.

[0159] The memory 92 may include a random access memory 92 (RAM) or a read-only memory 92 (Read-Only Memory). Optionally, the memory 92 includes a non-transitory computer-readable storage medium. The memory 92 may be used to store instructions, programs, codes, code sets or instruction sets. The memory 92 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch instructions, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 92 may also be optionally at least one storage device located away from the aforementioned processor 91.

[0160] The present application also provides a storage medium that can store multiple instructions, which are suitable for the processor to load and execute the above-mentioned Figures 1 to 7 For the specific implementation process, please refer to Figures 1 to 7 The specific description is not repeated here.

[0161] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0162] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0163] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraint algorithm of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0164] In the embodiments provided by the present invention, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are only schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0165] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0166] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0167] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by a processor, the steps of the above-mentioned method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc.

[0168] The present invention is not limited to the above-mentioned embodiments. If various changes or modifications to the present invention do not depart from the spirit and scope of the present invention, and if these changes and modifications fall within the scope of the claims and equivalent technologies of the present invention, the present invention is also intended to include these changes and modifications.

Claims

1. A respiratory status early warning method based on vital sign signals, characterized in that: The following steps are involved: Obtaining a vital sign signal of a sample user at a preset first time scale; Extracting a ballistocardiogram signal and a respiration signal from the vital sign signals of the sample user; Determine a characteristic sequence corresponding to the ballistocardiogram signal and the breathing signal of the sample user; Determine a differential feature sequence of the feature sequence at a plurality of different second time scales, wherein the differential feature sequence includes differential feature vectors at a plurality of time points; Inputting the differential feature sequence of the sample user into a preset respiratory state classification model to obtain a respiratory state sequence of the sample user, wherein the respiratory state sequence includes respiratory state vectors of respiratory states at several time points; According to the differential feature sequence and the respiratory state sequence of the sample user, the model parameter data of the respiratory state classification model is updated, and according to the updated model parameter data, the differential feature sequence and a preset number of iterations, the model parameter data of the respiratory state classification model is repeatedly updated to obtain a target respiratory state classification model; Obtain a ballistocardiogram signal and a respiratory signal of a user to be tested, determine a differential feature sequence corresponding to the ballistocardiogram signal and the respiratory signal of the user to be tested, input the differential feature sequence corresponding to the ballistocardiogram signal and the respiratory signal of the user to be tested into the target respiratory state classification model, and obtain the respiratory state sequence of the user to be tested; and obtain a respiratory state warning result of the user to be tested according to the number of target respiratory state vectors in the respiratory state sequence of the user to be tested.

2. The respiratory status early warning method based on vital sign signals according to claim 1, characterized in that: The method of extracting a ballistocardiogram signal and a respiratory signal from the vital sign signal of the sample user comprises the steps of: Using a filter to filter the vital sign signal to obtain a filtered vital sign signal; Performing an opening operation of first corroding and then dilating the filtered vital sign signal to obtain a first vital sign signal; Performing a closing operation of first dilation and then corrosion on the filtered vital sign signal to obtain a second vital sign signal; averaging the first vital sign signal and the second vital sign signal, and using the average result as a respiratory signal; The filtered vital sign signal is subtracted from the respiratory signal to obtain a ballistocardiogram signal.

3. The respiratory status early warning method based on vital sign signals according to claim 1, characterized in that: The feature sequences include a heart rate feature sequence, a heart rate variability feature sequence, a respiratory frequency feature sequence, and a sleep feature sequence; The step of determining the characteristic sequence corresponding to the ballistocardiogram signal and the breathing signal of the sample user comprises the steps of: Performing a timing bidirectional detection on the ballistocardiogram signal according to a preset ballistocardiogram signal template, determining the peak point coordinates of the ballistocardiogram signal, and obtaining a plurality of cardiac intervals according to the peak point coordinates; According to the plurality of cardiac intervals, a plurality of heart rates are obtained; and the heart rate data parameters corresponding to the plurality of heart rates are used as a heart rate characteristic sequence; Calculate the standard deviation of the plurality of cardiac intervals, and obtain a plurality of power spectra according to the standard deviation and a preset autoregressive spectrum estimation method; compare the plurality of power spectra with a preset spectrum range to determine the spectrum range in which each power spectrum is located; obtain the heart rate deceleration force according to the plurality of cardiac intervals; obtain the heart rate variability characteristic sequence according to the average value of all power spectra in the same spectrum range and the heart rate deceleration force; According to the several respiratory intervals of the respiratory signal, several respiratory frequencies are obtained; and the respiratory data parameters corresponding to the several respiratory frequencies are used as a respiratory frequency feature sequence; A sleep characteristic sequence is obtained according to a number of cardiac intervals, a number of respiratory intervals and a preset cardiopulmonary coupling method.

4. The respiratory status early warning method based on vital sign signals according to claim 1, characterized in that: The differential feature sequence includes differential feature vectors of several time points, and the differential feature vectors include first-order differential feature vectors and second-order differential feature vectors; The step of determining a differential characteristic sequence of the characteristic sequence at a plurality of different second time scales comprises the steps of: Traverse each second time scale, calculate the first-order difference original value and the second-order difference original value of the feature sequence at the current second time scale, perform ordinal variable encoding conversion on the first-order difference original value and the second-order difference original value, obtain the first-order difference encoding value and the second-order difference encoding value, respectively serve as the first-order difference feature vector and the second-order difference feature vector at the current second time scale, until the first-order difference feature vector and the second-order difference feature vector corresponding to the feature sequence at all second time scales are determined, and the differential feature sequence of the feature sequence at several different second time scales is determined.

5. The respiratory status early warning method based on vital sign signals according to claim 4, characterized in that: Before inputting the differential feature sequence of the sample user into a preset respiratory state classification model to obtain the respiratory state sequence of the sample user, the method further includes the following steps: Recursive feature elimination and a stratified cross-validation method are used to perform feature selection on the differential feature sequences of the sample users at each second time scale, so as to obtain the differential feature sequences of the sample users at several different second time scales after feature selection.

6. The respiratory status early warning method based on vital sign signals according to claim 5, characterized in that: The model parameter data includes an observation probability matrix, a state transition probability matrix and an initial state probability matrix; The updating of the model parameter data of the respiratory state classification model according to the differential feature sequence and the respiratory state sequence of the sample user comprises the steps of: Calculating the expected probability of the respiratory state at several time points of the differential feature sequence according to the initial state probability matrix, the differential feature sequence and the respiratory state sequence; According to the differential feature vectors at several time points in the differential feature sequence and the expected probabilities of the respiratory states at several time points, the differential feature vector observation probabilities of the respiratory states at several time points are obtained, and the differential feature vector observation probabilities of the respiratory states at corresponding time points in the observation probability matrix are updated; Obtaining forward probabilities and backward probabilities of the respiratory states at several time points, and estimating the conditional probabilities between the respiratory states at several time points based on the product of the forward probabilities of the respiratory states at the corresponding time points; According to the expected probabilities of the respiratory states at the several time points and the conditional probabilities between the respiratory states at the corresponding time points, the state transition probabilities between the several respiratory states are obtained, and the state transition probabilities between the several respiratory states in the state transition probability matrix are updated; The expected probability of the respiratory state at the first time point of the differential feature sequence is used as the initial state probability to update the initial state probability of the respiratory state in the initial state probability matrix.

7. The respiratory status early warning method based on vital sign signals according to claim 6, characterized in that: The method of calculating the expected probability data of several time points of the differential feature sequence according to the model parameter data, the differential feature sequence and the respiratory state sequence comprises the steps of: According to the first-order difference feature sequence and the second-order difference feature sequence, respectively, the observation probability calculation is performed with the respiratory state sequence to obtain the first-order difference observation probability and the second-order difference observation probability of the respiratory state at several time points; The first-order difference observation probability and the second-order difference observation probability of the respiratory state at the same time point are added to obtain the joint observation probability of the respiratory state at several time points, and the total observation probability is obtained according to the joint observation probability of the respiratory state at several time points; The observation probability of the differential feature sequence is obtained, and the expected probability of the respiratory state at several time points is obtained according to the observation probability of the differential feature sequence, the initial state probability matrix and the total observation probability.

8. A respiratory status warning device based on vital sign signals, characterized in that: include: A vital sign signal acquisition module, used to obtain a vital sign signal of a sample user at a preset first time scale; A signal extraction module, used to extract a ballistocardiogram signal and a breathing signal from the vital sign signal of the sample user; A feature sequence determination module, used to determine the feature sequence corresponding to the ballistocardiogram signal and the breathing signal of the sample user; A differential feature sequence determination module, used to determine the differential feature sequence of the feature sequence at a plurality of different second time scales, wherein the differential feature sequence includes differential feature vectors at a plurality of time points; A breathing state sequence acquisition module, used for inputting the differential feature sequence of the sample user into a preset breathing state classification model to obtain the breathing state sequence of the sample user, wherein the breathing state sequence includes breathing state vectors of breathing states at several time points; A model updating module, used for updating the model parameter data of the respiratory state classification model according to the differential feature sequence and the respiratory state sequence of the sample user, and repeatedly updating the model parameter data of the respiratory state classification model according to the updated model parameter data, the differential feature sequence and a preset number of iterations to obtain a target respiratory state classification model; A respiratory state warning module is used to obtain the ballistocardiogram signal and the respiratory signal of the user to be tested, determine the differential feature sequence corresponding to the ballistocardiogram signal and the respiratory signal of the user to be tested, input the differential feature sequence corresponding to the ballistocardiogram signal and the respiratory signal of the user to be tested into the target respiratory state classification model to obtain the respiratory state sequence of the user to be tested; and obtain the respiratory state warning result of the user to be tested according to the number of target respiratory state vectors in the respiratory state sequence of the user to be tested.

9. A computer device, characterized in that: The method comprises a processor, a memory and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the respiratory status warning method based on vital sign signals as described in any one of claims 1 to 7 are implemented.

10. A storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the respiratory state warning method based on vital sign signals as described in any one of claims 1 to 7 are implemented.

Images