CN115775629B - Method for assessing risk of PPCs (point-to-point) based on sub-polar motion test before operation - Google Patents
Method for assessing risk of PPCs (point-to-point) based on sub-polar motion test before operation Download PDFInfo
- Publication number
- CN115775629B CN115775629B CN202310096280.XA CN202310096280A CN115775629B CN 115775629 B CN115775629 B CN 115775629B CN 202310096280 A CN202310096280 A CN 202310096280A CN 115775629 B CN115775629 B CN 115775629B
- Authority
- CN
- China
- Prior art keywords
- characteristic data
- data set
- minute
- test
- sub
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Abstract
The application relates to a method for assessing PPCs risk based on a sub-polar motion test before operation, which comprises the following steps: acquiring continuous respiration signals of an estimated person before, during and after performing a sub-maximum exercise test; extracting a pre-operative respiratory signal characteristic data set of the subject based on the continuous respiratory signal, the pre-operative respiratory signal characteristic data set including at least two of a tidal volume characteristic data set, a ventilation volume characteristic data set, a respiratory rate characteristic data set, and an inspiratory expiratory time characteristic data set; each characteristic data set comprises at least one characteristic data; and constructing a prediction model by taking the characteristic data and the preoperative clinical physiological characteristic data as variables and taking the postoperative lung complication probability as a function. According to the method and the device, the probability of the occurrence of the pulmonary complications after the heart valve operation can be accurately obtained by acquiring the data related to the postoperative pulmonary complications and the established prediction model, so that guidance is provided for the establishment of an operation treatment scheme, and risks are avoided.
Description
Technical Field
The application belongs to the field of lung complication prediction, and particularly relates to a method for assessing PPCs risk based on a sub-maximum movement test before operation.
Background
The method plays an important role in the aspects of postoperative pulmonary complications PPCs related risk assessment, clinical treatment scheme formulation, prognosis judgment, complications prediction, treatment effect evaluation, medical resource demand estimation and the like of patients subjected to heart valve surgery.
In the prior art, prediction or evaluation of postoperative pulmonary complications PPCs before surgery is often obtained based on experience, and a professional medical worker evaluates and judges the probability of postoperative pulmonary complications based on physiological data of an evaluated person based on experience. This approach relies on the experience of the healthcare worker and is also subject to inaccuracy.
Sub-maximal exercise tests, such as the six-Minute Walk Test (6 MWT), have been widely used in the prior art as a simple to operate, easily accepted Test by patients to monitor therapeutic response and prognostic outcome of cardiopulmonary disease. However, the six-Minute Walk Test (6 MWT) is generally only a simple parameter for simply detecting the walking distance, heart rate, etc. of the subject as a reference to the subject's related health data. This assessment is likewise imprecise and only qualitatively gives an approximation of the health of the person being assessed.
Methods have also emerged in the prior art for assessing the risk of certain diseases by modeling the monitoring of physiological data in six-minute walking. However, for other disease types, such as PPCs, there is no teaching in the prior art of what data is available for six-minute walking and what is a strong correlation between PPCs and what is in these data.
Disclosure of Invention
The present application is made in view of the above-mentioned needs of the prior art, and an object of the present application is to provide a method for assessing PPCs risk based on a sub-maximal exercise test before surgery, so as to conveniently and relatively accurately assess the probability of postoperative complications of the lung before surgery.
In order to solve the above problems, the technical solution provided in the present application includes:
acquiring continuous respiratory signals of an evaluated person in a first time period before performing the sub-maximum exercise test, a second time period during the sub-maximum exercise test and a third time period after performing the sub-maximum exercise test; wherein the continuous respiratory signal comprises a continuous chest respiratory signal, a continuous abdominal respiratory signal; extracting a pre-operative respiratory signal characteristic data set of the subject based on the continuous respiratory signal, wherein the pre-operative respiratory signal characteristic data set comprises at least two of a tidal volume characteristic data set, a ventilation volume characteristic data set, a respiratory frequency characteristic data set and an inspiratory expiratory time characteristic data set; each characteristic data set comprises at least one characteristic data; and constructing a prediction model to predict the postoperative pulmonary complications probability by taking the characteristic data and the preoperative clinical physiological characteristic data as variables and the postoperative pulmonary complications probability as a function.
According to the method, the continuous characteristic data related to the postoperative pulmonary complications and the established prediction model are acquired on the basis of the sub-maximum motion test before the operation, so that the probability of the occurrence of the pulmonary complications after the heart valve operation can be accurately obtained, guidance is provided for the establishment of an operation treatment scheme, the effect of operation treatment is improved, and risks are avoided.
Preferably, the pre-operative respiratory signal characteristic data set is a tidal volume characteristic data set and a ventilation volume characteristic data set. Two sets of pre-operative respiratory signal characteristic data that have the greatest impact on predicting post-operative pulmonary complications are determined.
Preferably, the tidal volume characterization data set includes at least tidal volume characterization data for expiration during a third time period.
Preferably, the ventilation characteristic data set includes at least an inspiratory minute ventilation, an inspiratory minute ventilation maximum, and a slope of the inspiratory minute ventilation rising from baseline to 75% of the maximum over the first time period. Four characteristic data with the greatest influence on postoperative pulmonary complications in the respiratory signal data are determined.
Preferably, the sub-maximum exercise test comprises a six-minute walking test, and the continuous respiration signals of the subject are collected one minute before the six-minute walking test starts, during the six-minute walking test, and one minute after the six-minute walking test. In addition to acquiring physiological information of walking distance, continuous physiological information of each stage of walking test is acquired, so that diversity of six-minute walking test evaluation indexes can be increased.
Preferably, the pre-operative clinical physiological characteristic data at least comprises pre-operative pulmonary artery diameter data, predictive data (FEV 1-predictive) of volume of the first second of maximum exhalate following the pre-operative maximum deep inhalation, and surgical mode data. Three characteristic data with the greatest influence on postoperative pulmonary complications in clinical data are determined.
Preferably, the prediction function of the prediction model is that
Wherein, the liquid crystal display device comprises a liquid crystal display device,
,/>
,/>
the complication probability of the array x formed for the input evaluative subject characteristic data, θ is the vector formed by the characteristic coefficients, ++>
Is constant (I)>
Is->
Characteristic coefficients corresponding to the item characteristic data, x is a vector formed by the characteristic data, ++>
，/>
Is->
Item feature data. Prediction function->
The result of (2) is a value of 0 to 1, with a closer to 1 indicating a greater probability of postoperative pulmonary complications.
Preferably, the prediction function is constructed based on logistic regression, wherein a loss function of the prediction function is expressed as:
wherein, the liquid crystal display device comprises a liquid crystal display device,
is a true tag.
Preferably, the cost function obtained based on the loss function is expressed as:
wherein m is the number of continuous physiological clinical parameter arrays in the data set,
is->
A continuous set of physiological clinical parameters->
Is->
Corresponding tag, < >>
The value of (2) is 0 or 1.
Preferably, the characteristic coefficients are initialized by gradient descent and gradually updated until the characteristic coefficients optimal for the characteristic data in the continuous physiological clinical parameter array are obtained
Expressed as:
wherein, the liquid crystal display device comprises a liquid crystal display device,
represents the current +.>
Parameters of round iteration, ++>
Represents->
Model coefficients for round iterations. And determining the optimal characteristic coefficient vector by a loss function, a cost function and a gradient descent method.
Compared with the prior art, the method and the device have the advantages that the breathing physiological characteristic data closely related to the PPCs are obtained in the sub-maximum exercise test in a characteristic extraction mode, and parameters with large influence on the PPCs are selected through a prediction model by combining the pre-operation clinical physiological characteristic data, so that the probability of postoperative pulmonary complications can be accurately and efficiently determined, and the model prediction with high reliability is established; parameters of the prediction model are simplified, and the accuracy of prediction is improved.
Drawings
In order to more clearly illustrate the embodiments of the present description or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments described in the embodiments of the present description, and other drawings may be obtained according to these drawings for a person having ordinary skill in the art.
FIG. 1 is a flow chart of the steps of a method of preoperatively assessing risk of PPCs based on a sub-polar kinetic test in an embodiment of the present application;
FIG. 2 is a flowchart of method steps for forming a predictive model in an embodiment of the present application.
Detailed Description
For the purposes of making the objects, technical solutions and advantages of the embodiments of the present application more clear, the technical solutions of the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is apparent that the described embodiments are some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the present disclosure, are within the scope of the present disclosure.
In the description of the embodiments of the present application, it should be noted that, unless explicitly specified and limited otherwise, the term "connected" is to be construed broadly, and for example, it may be a fixed connection, a detachable connection, or an integral connection, a mechanical connection, an electrical connection, a direct connection, or an indirect connection via an intermediary. The specific meaning of the terms in this application will be understood by those of ordinary skill in the art as the case may be.
For the purpose of facilitating an understanding of the embodiments of the present application, reference will now be made to the following description of specific embodiments, taken in conjunction with the accompanying drawings, in which the embodiments are not intended to limit the embodiments of the present application.
Before operation, for example, before heart valve operation, postoperative pulmonary complications, namely PPCs (Postoperative pulmonary complications) risk assessment, are carried out on patients, and important effects are played in the aspects of making clinical treatment schemes, judging prognosis and the like. The pulmonary function test and the cardiopulmonary exercise test are standard tests for predicting the risk of PPCs, and the pre-operative respiratory signal characteristic data and the pre-operative clinical physiological characteristic data of the evaluated person are considered as important physiological indexes capable of reflecting the cardiopulmonary function of the evaluated person. However, the pulmonary function test and the cardiopulmonary exercise test equipment are expensive, the occurrence rate of the process contraindications and adverse events is high, and the use is inconvenient.
In this embodiment, the sub-maximum exercise test is a test for health detection by exercise of low intensity, which is commonly used in the medical field. The sub-maximum exercise test can be widely applied to various crowds due to low load intensity, so the sub-maximum exercise test has high convenience as a data source for PPCs prediction.
A common sub-polar exercise test may include a six minute walk test, i.e., 6-MWT (six minute walk test). For ease of understanding, the inventive concepts of the present application will be described in this detailed description with reference to a 6-MWT.
In the traditional health assessment based on six-minute walking test, the exercise capacity of a patient is graded mainly by the distance 6-MWD (six minute walk distance) which walks as much as possible in six minutes, and on one hand, the index of 6-MWD is greatly influenced by the individual difference and the environment of an assessed person; on the other hand, the evaluation index is relatively single and the continuous physiological information of each stage of the walking test is omitted.
While some physiological data changes during the first few hours of adverse events, continuous monitoring of physiological parameters can reflect more physiological information.
In addition, even if intermittent signal detection is performed in the six-minute walk test process, in many cases, the detection is limited to the blood oxygen saturation SpO2 to determine whether or not hypoxia occurs, and parameters commonly used such as Heart Rate (HR), respiratory Rate (RR), blood Pressure (BP), and pulse are used. Intermittent signal monitoring results in relatively single evaluation indexes, and is easy to miss continuous physiological signals of each stage of walking test, so that early deterioration signs are caused, and adverse events occur. Furthermore, these intermittent signal monitoring have limited mining of PPCs-related information that these data may contain, while medical judgment is very specialized and rigorous science, so that prediction of PPCs can stay not only in general qualitative tests but also provide accurate predictive models, and these models have high verification accuracy that is a problem to be considered in this embodiment.
To this end, the present embodiments provide a method for preoperatively assessing risk of PPCs based on a sub-maximal exercise test. The steps of the method can be seen from the description of figures 1 and 2 of the specification.
S1, acquiring continuous respiration signals of an estimated person in a first time period before performing the sub-maximum exercise test, a second time period during the sub-maximum exercise test and a third time period after performing the sub-maximum exercise test; wherein the continuous respiratory signal comprises a continuous chest respiratory signal, a continuous abdominal respiratory signal.
In this embodiment, the wearable physiological parameter monitoring device is used to collect continuous respiration signals of the evaluated person in a predetermined period of time before and after the six-minute walking test. The former predetermined period of time refers to a period of time immediately following the start of the six-minute walk test, immediately after which the six-minute walk test starts, and the latter predetermined period of time refers to a further predetermined period of time immediately following the six-minute walk test, immediately after which the six-minute walk test starts. The preceding predetermined period of time and the following predetermined period of time may be the same or different.
For example, the data processing and analysis can be performed on the continuous respiration signals for a total of 8 minutes each of 1 minute before and after the six-minute walk test by the subject and 6 minutes in the six-minute walk test. The 6MWT is divided into stages according to time: 1 minute before the onset of 6MWT is a resting baseline phase (baseline); the 6MWT Test is a walking stage (WT); the 6MWT end time is the end stage (end); the recovery phase (recovery) is 1 minute after the end of the 6 MWT.
The continuous chest respiratory signal refers to a continuous chest respiratory signal acquired by a respiratory motion sensor in the wearable physiological parameter detection device, for example, a chest respiratory signal with a sampling rate of 25 Hz.
The continuous abdominal respiration signal refers to a continuous abdominal respiration signal acquired by a respiratory motion sensor in the wearable physiological parameter detecting device, for example, an abdominal respiration signal with a sampling rate of 25 Hz.
Further, in this step, the chest-abdomen summation signal, that is, the summation signal formed by the superposition of the chest respiration signal and the abdomen respiration signal peaks and troughs, may also be obtained from the continuous chest respiration signal and the continuous abdomen respiration signal.
Chest breathing is a common breathing mode and can represent the condition of human lung function; the abdominal respiration focuses on diaphragm movement, expands the thoracic range, can effectively exchange gas relative to chest respiration, and helps to enhance the vital capacity, so that the method has great benefit on the lung function of a human body, and reduces the occurrence rate of PPCs. However, the quantitative research on abdominal respiration is limited at present, and the quantitative research on the measurement of the overall chest respiration and abdominal respiration on the pulmonary function is limited, so that the contribution degree of abdominal respiration can be better quantified by continuously collecting the chest and abdomen respiration signals in the specific embodiment, and the overall respiratory condition data of the evaluated person can be better collected.
The wearable physiological parameter monitoring device is not limited in this embodiment, and only needs to acquire the continuous respiratory signal.
S2, extracting a pre-operation respiratory signal characteristic data set of the evaluated person based on the continuous respiratory signal, wherein the pre-operation respiratory signal characteristic data set comprises at least two of a tidal volume characteristic data set, a ventilation volume characteristic data set, a respiratory frequency characteristic data set and an inspiration expiration time characteristic data set; each feature data set includes at least one item of feature data.
Extracting the pre-operation respiratory signal characteristic data set of the evaluated person based on the continuous respiratory signal in the step comprises, optionally, smoothing and filtering the original respiratory signal to remove abnormal values and obtain a clean respiratory signal, so as to extract the related characteristics of different time domains of respiratory physiological parameters including respiratory frequency characteristic data sets from the clean chest respiratory signal and the abdominal respiratory signal; using prior art techniques such as those described in khodad, d., et al, optimized breath detection algorithm in electrical imdaancetomograph, physiol Meas, 2018.39 (9): p. 094001, continuous chest and abdomen summation signal peak-trough were detected and the inspiration phase (peak-trough) was selected to calculate tidal volume, minute ventilation, inspiration expiration time related feature data sets. Wherein, tidal volume = inspiration time x supply flow rate, the longer the inspiration time, the greater the tidal volume; minute ventilation = respiratory rate x tidal volume, refers to the total amount of gas that enters or exits the lungs per minute.
The pre-operative respiratory signal characteristic data set obtained through the calculation in the step comprises at least two sets of a tidal volume characteristic data set, a ventilation volume characteristic data set, a respiratory frequency characteristic data set and an inspiration and expiration time characteristic data set; each feature data set includes at least one item of feature data.
The respiratory rate refers to the number of breaths per minute, and one fluctuation of the chest is one breath. Respiratory rate is an important indicator of vital signs. The normal adult respiratory rate is 12-20 times per minute, and relatively lower respiratory rate in the normal range means that the respiratory depth is increased, and the lung function is reflected to a certain degree. Collecting and extracting respiratory rate data can facilitate analysis of the relationship between respiratory rate and lung function.
Tidal volume refers to the volume of gas inhaled or exhaled each time during calm breathing, is an indicator showing lung volume, and is mainly used for ventilation function examination in lung function examination. When the respiratory function is not complete, the tidal volume can be reduced, and the tidal volume index can help to check whether the respiratory system diseases such as interstitial pneumonia, pulmonary fibrosis, pulmonary edema and the like exist, so that the tidal volume characteristic data set is also an important index data for measuring the PPCs.
Ventilation refers to the total amount of gas exhaled and inhaled into the lungs in a unit time and can comprehensively reflect the ventilation function of the lungs, so that the ventilation characteristic data set is also an important index data for measuring PPCs.
Tidal volume, respiratory rate, minute ventilation, and inspiratory expiratory time of the subject are generally indicative of lung function, and the incidence of postoperative PPCs should be reduced in patients with good pre-operative lung function. The extracted tidal volume, respiratory rate, minute ventilation, and inspiratory expiration time may be used to determine quantitative relevant lung function data for the subject to be assessed to provide a data basis for the quantitative model to predict PPCs odds.
Wherein the respiratory rate characteristic data set may comprise: a resting baseline stage respiratory rate average value (br_base), a respiratory rate value at the last end of the walking stage (br_end), a recovery stage respiratory rate average value (br_recovery), a respiratory rate value at the time of recovery 1 minute after the end of walking (brr), a walking stage maximum respiratory rate value (br_max), a time taken for the respiratory rate to increase from the start to the maximum value (br_acc_time), a slope at which the respiratory rate of the walking stage increases from the value at the start to 75% of the maximum value (br_slope), a resting baseline stage respiratory rate standard deviation (br_base_std), a walking stage respiratory rate standard deviation (br_wt_std), and a recovery stage respiratory rate standard deviation (br_recovery_std).
The respiratory rate characteristic data can be embodied in the respiratory rate characteristics of each stage of the 6MWT test, and the real-time state of the respiratory system before, during and after the exercise of the person to be evaluated can be detected, so that the respiratory inhibition degree of the person to be evaluated can be judged.
The tidal volume characterization data set may include: the time it takes for the tidal volume to rise from the base line to a maximum value (vt_in_acc_time), the slope of the tidal volume from the base line to 75% of the maximum value (vt_in_slope), the resting base line phase tidal volume average (vt_in_base), the walking phase maximum tidal volume value (vt_in_max), the recovery phase tidal volume average (vt_in_recovery), the resting base line phase tidal volume standard deviation (vt_base_std), the walking phase tidal volume standard deviation (vt_wt_std), the recovery phase tidal volume standard deviation (vt_recovery_std), the recovery phase expiratory tidal volume (vt_ex_re).
The tidal volume characteristic data can be embodied in the tidal volume characteristics of each stage of the 6MWT test, the ventilation function of the lungs of the evaluated person can be detected, and the respiratory depression degree of the evaluated person can be judged.
Wherein the minute ventilation feature data set may comprise: the rest baseline phase inspiration minute ventilation (MV in base), inspiration minute ventilation maximum (MV in max), and the slope of inspiration minute ventilation from baseline up to 75% of maximum (MV in slope).
The minute ventilation characteristic data described above can be expressed as minute ventilation characteristics at each stage of the 6MWT test, and the total amount of air inhaled into the lungs per minute of the subject can be detected, thereby determining the respiratory depression degree of the subject.
Wherein the inspiration-expiration time feature data set may comprise: a preparation phase inspiration time average (ti_base), a preparation phase expiration time average (te_base), a walking phase inspiration time average (ti_wt), a walking phase expiration time average (te_wt), a recovery phase inspiration time average (ti_recovery), a recovery phase expiration time average (te_recovery), a preparation phase inspiration time duty cycle average (ti_ratio_base), a walking phase inspiration time duty cycle average (ti_ratio_wt), a recovery phase inspiration time duty cycle average (ti_ratio_recovery).
In addition, PPCs-related risk assessment is closely related to the formulation of clinical treatment schemes, the evaluation of treatment effects, the estimation of medical resource requirements, and the like, so that the characteristic data of the postoperative lung complication prediction probability also includes a plurality of preoperative clinical physiological parameters closely related to PPCs.
The pre-operative clinical physiological characteristic data set closely related to PPCs may include: surgical procedure, cardiac function NYHA, risk Score Euro Score, left chamber internal diameter, left chamber enlargement, left Fang Najing, left chamber enlargement, right chamber internal diameter, right chamber enlargement, left chamber end diastole volume, left chamber end diastole structural change, left chamber end systole volume, left chamber end systole structural change, left chamber ejection fraction, left chamber end systole functional change, pulmonary artery diameter, pulmonary hypertension, hemoglobin, glomerular filtration rate, FEV 1-actual measurement, predictive data of volume of maximum exhalation first second after maximum pre-operative deep inhalation and surgical procedure data (FEV 1-prediction), FVC-actual measurement, FVC-prediction, admission 6MWD.
The clinical physiological characteristic data are important indexes for detecting heart and lung diseases, and are closely related to heart valve operation and postoperative lung complications PPCs of an evaluated person.
There are 26 feature data in the preoperative clinical physiological feature dataset closely related to PPCs, namely: surgical procedure, cardiac function NYHA, risk Score Euro Score, left chamber internal diameter, left chamber enlargement, left Fang Najing, left chamber enlargement, right chamber internal diameter, right chamber enlargement, left chamber end diastole volume, left chamber end diastole structural change, left chamber end systole volume, left chamber end systole structural change, left chamber ejection fraction, left chamber end systole functional change, pulmonary artery diameter, pulmonary hypertension, hemoglobin, glomerular filtration rate, FEV 1-actual measurement, predictive data of volume of maximum exhalation first second after maximum pre-operative deep inhalation and surgical procedure data (FEV 1-prediction), FVC-actual measurement, FVC-prediction, admission 6MWD.
Preferably, the characteristic data in the tidal volume characteristic data set may be only the expiratory tidal volume (vt_ex_re) during the third time period.
Preferably, the characteristic data in the minute ventilation characteristic data set may be only: the inspiratory minute ventilation (mv_in_base), the inspiratory minute ventilation maximum (mv_in_max), and the slope of the inspiratory minute ventilation from baseline up to 75% of the maximum (mv_in_slope) for the first time period.
Preferably, the feature data in the pre-operative clinical physiological feature data may be only: pre-operative pulmonary artery diameter data, predictive data of volume of the first second of maximum exhalation taken after maximum pre-operative deep inhalation (FEV 1-prediction), and surgical mode data.
The above features are feature data which has the strongest correlation with PPCs and is selected from the above feature data set by simulation calculation, and will be described in detail later.
In summary, the physiological characteristic data set related to the respiratory signal has 31 characteristic data.
The 57 items of characteristic data may be related to PPCs for each item, but such a large amount of data as input may cause the speed of calculation to be affected, and which of these characteristic data is strongly related to PPCs, and which data is not strongly related to the risk probability of occurrence of PPCs needs to be accurately evaluated. Therefore, in the specific embodiment, the logistic regression prediction model is constructed for the purposes of screening the 57 item of characteristic data strong correlation data and improving the calculation speed of the prediction model.
S3, constructing a prediction model to predict the postoperative lung complication probability by taking the postoperative lung complication probability as a function based on the physiological feature data related to the respiratory signals and the preoperative clinical physiological feature data as variables.
In the present embodimentBased on at least one characteristic data of each of the tidal volume characteristic data set, minute ventilation characteristic data set, and pre-operative clinical physiological characteristic data set, to predict a function
Obtaining the probability of postoperative complications of the lung; wherein, the liquid crystal display device comprises a liquid crystal display device,
，/>
，/>
array formed for input evaluative subject characteristic dataxIs used for the treatment of the complications of the patients,θvectors formed for characteristic coefficients, ++>
Is constant (I)>
Is->
The feature coefficients corresponding to the item feature data,xvectors formed for feature data, ++>
，/>
Is->
Item feature data.
And constructing a data set based on the tidal volume characteristic data set, the minute ventilation characteristic data set and the preoperative clinical physiological characteristic data set, and establishing a prediction model based on the data set to obtain final characteristic data forming the prediction function and final characteristic coefficients corresponding to the final characteristic data.
A first predictive model is built based on a dataset having an array of M-term feature data.
Based on the result of the complications, the prediction model is required to output a numerical value between 0 and 1, and the closer the numerical value is to 1, the greater the probability of the postoperative pulmonary complications is. Setting the first prediction model as a logistic regression model and simultaneously predicting
As a Logistic function, expressed as:
wherein, the liquid crystal display device comprises a liquid crystal display device,
1->
For the number of features>
Is->
Personal characteristics (I)>
Is->
The coefficients corresponding to the individual features.
Representing the current sample->
Is a numerical value of the output probability of (a). />
In this embodiment, the dataset includes 57 items of characteristic data, namely
=47, initially the feature coefficients are randomly assigned.
The loss function of the dataset based on the logistic regression model is expressed as:
wherein, the liquid crystal display device comprises a liquid crystal display device,
is a true tag if->
=1, and predicted +.>
Also equal to 1, the loss is minimal; if->
The losses are very large when predicted to be 0. Likewise, if the tag is authentic->
When 0, add>
Predicting to be 0 and losing to be 0; />
In the case of prediction 1, the loss is very large.
The cost function obtained based on the loss function is expressed as:
wherein m is the number of continuous physiological clinical parameter arrays in the data set,
is->
A continuous set of physiological clinical parameters->
Is->
Corresponding tag, < >>
The value of (2) is 0 or 1. The aim is to make the value of the coefficient theta as small as possible, m being the number of training samples. The final objective is to find the appropriate j parameters θ, minimize +.>
。
Initializing the characteristic coefficient through gradient descent and updating gradually until obtaining the characteristic coefficient optimal for the characteristic data in the continuous physiological clinical parameter array
Expressed as:
wherein, the liquid crystal display device comprises a liquid crystal display device,
represents the current +.>
Parameters of round iteration, ++>
Represents->
Model coefficients for round iterations. Until the difference between the parameters of the two iterative updates is very small, the iteration is stopped and the iteration is stopped>
Can be determined.
Taking the above embodiment as an example, by means of gradient descent, iteration is continuously performed to approach the feature coefficient that minimizes the cost function, i.e. the optimal feature coefficient under the first prediction model.
After the feature coefficients are obtained through the optimization process, the feature coefficients are arranged according to the size sequence, and the minimum feature coefficient and the feature data corresponding to the minimum feature coefficient are deleted. Specifically, the optimal characteristic coefficients under the obtained first prediction model are arranged from large to small, characteristic data corresponding to the 57 th characteristic coefficient is removed, namely, characteristic data with minimum weight relative to postoperative pulmonary complications is removed, namely, the influence of the characteristic data on the postoperative complications is minimum.
And reconstructing the data set with the characteristic data deleted to a second prediction model. Specifically, the above procedure is performed again with the data set of the feature data from which the feature data corresponding to the minimum feature coefficient is removed as a new data set, that is, the second prediction model is reformed based on the data set having 56 pieces of feature data.
The prediction function of the second prediction model is also a Logistic function, and the construction process of the second prediction model is the same as that of the first prediction model, and the two are different in that the feature data deleted after the first iterative sequencing is not performed in the data set of the second prediction model, so that the optimal feature coefficient finally obtained by the second prediction model is different from the optimal feature coefficient obtained by the first prediction model. And then, sorting the optimal characteristic coefficients according to the order of magnitude, and eliminating the characteristic data corresponding to the smallest characteristic coefficient in the data set. The elimination process is also used for deleting the factor which has the smallest capability of affecting the postoperative pulmonary complications so as to discharge the data which does not need to be acquired, so that the acquisition process is simplified, the operation is more convenient and rapid, and the data volume can be effectively reduced.
And circularly carrying out the steps of reconstructing an Nth prediction model based on the residual characteristic data, arranging and eliminating the characteristic data until the complications predicted value of the Q-th prediction model is reduced by a first threshold value after deleting certain characteristic data. The first threshold comprises 60% or more.
And reserving the M-Q item characteristic data and characteristic coefficients corresponding to the reserved characteristic data as final characteristic data and final characteristic coefficients respectively to obtain the prediction function.
Compared with the neural network and other models, the prediction result is obtained more quickly and intuitively by the logistic regression model on the premise of ensuring the probability of predicting the complications.
Based on the implementation manner of the embodiment, the input of the finally obtained prediction model is 7 items of characteristic data, which is: expiration tidal volume during the recovery phase, inspiration minute ventilation during the resting baseline phase, inspiration minute ventilation maximum, slope of inspiration minute ventilation rising from baseline to 75% of maximum, pre-operative pulmonary artery diameter data, predictive data (FEV 1-prediction) of volume of ventilation for the first second of maximum expiration following the maximum pre-operative deep inspiration, surgical mode data.
I.e. the characteristic data in the tidal volume characteristic data set is at least the recovery phase tidal volume of the expired breath.
The characteristic data in the minute ventilation characteristic data set are: the rest baseline phase inspiration minute ventilation, inspiration minute ventilation maximum and slope of inspiration minute ventilation rise from baseline to 75% of maximum.
The characteristic data of the preoperative clinical physiological characteristic data are as follows: pre-operative pulmonary artery diameter data, predictive data of volume of the first second of maximum exhalation taken after maximum pre-operative deep inhalation (FEV 1-prediction), and surgical mode data.
Obtaining a prediction model according to the finally determined 7 items of characteristic data and final characteristic coefficients corresponding to the final characteristic data, wherein a prediction function of the prediction model is expressed as follows:
In the foregoing implementation manner of the present embodiment, the prediction function of the prediction model is expressed as: ppc=0.0033×operation+0.01284×pulmonary-0.011087×fev-0.000001×mv_in_slope-0.000052×mv_in_base-0.00001×mv_in_max+0.000829×vt_ex_re+0.000229.
After 7 features in the prediction function are subjected to a logistic regression model, the output result is a percentage value, and the represented meaning is: postoperative lung complications were scored as the incidence of the criteria Melbourne.
Although the prediction model is obtained by means of logistic regression in the present embodiment, in the actual prediction process, only the prediction function may be adopted, 7 pieces of characteristic data in the prediction function may be collected, and the probability of PPCs may be obtained by the prediction function.
In addition, due to the difference of the adopted feature extraction method and the acquisition equipment, the variables and coefficients of the prediction functions may also be changed compared with those of the 7 features, and the specific variables and coefficients of the variables can be selected by the person skilled in the art according to the actual situation.
Compared with the prior art, the probability of the physiological characteristic data and postoperative pulmonary complications in the six-minute walking test is determined through theoretical analysis, experience summarization and model calculation, the probability result of the postoperative pulmonary complications of the heart valve with high accuracy can be obtained through the data test in the six-minute walking test, and an evaluator can conveniently obtain accurate evaluation, so that guidance is provided for the formulation of an operation treatment scheme, the effect of operation treatment is improved, and risks are avoided.
The foregoing embodiments have been provided for the purpose of illustrating the general principles of the present application, and are not meant to limit the scope of the invention, but to limit the scope of the invention.
Claims (4)
1. A method of preoperatively assessing risk of PPCs based on a sub-maximal exercise test, the method comprising:
acquiring continuous respiration signals of an evaluated person in a first time period before performing the sub-maximum exercise test, a second time period during the sub-maximum exercise test and a third time period after performing the sub-maximum exercise test; wherein the continuous respiratory signal comprises a continuous chest respiratory signal and a continuous abdomen respiratory signal; the sub-maximum exercise test comprises a six-minute walk test, the first time period comprises a predetermined time period prior to the six-minute walk test, the second time period comprises a six-minute walk process time period, and the third time period comprises a further predetermined time period immediately after the six-minute walk test; extracting a preoperative respiratory signal characteristic data set of the evaluated person based on the continuous respiratory signal, wherein the method comprises the steps of detecting peaks and troughs of continuous chest and abdomen summation signals, selecting the peaks and troughs of the continuous chest and abdomen summation signals in an inspiration stage, and calculating at least two of a tidal volume characteristic data set, a ventilation volume characteristic data set, a respiratory frequency characteristic data set and an inspiration and expiration time characteristic data set; each characteristic data set comprises at least one characteristic data;
predicting the postoperative lung complication probability by taking the characteristic data and the preoperative clinical physiological characteristic data as variables and taking the postoperative lung complication probability as a prediction function, wherein the prediction function is expressed as:
wherein PPC represents prediction of postoperative pulmonary complications probability, operation is surgery, pulmoniy is preoperative pulmonary artery diameter data, FEV is prediction data of volume of maximum exhaled first second after maximum deep inhalation before surgery, mv_in_slope is slope of 75% of inhalation minute ventilation rising from baseline to maximum, mv_in_base is inhalation minute ventilation in a first time period, mv_in_max is inhalation minute ventilation maximum, vt_ex_re is exhalation tidal volume in a recovery stage.
2. A method of preoperative assessment of PPCs risk based on a sub-polar motion test according to claim 1, wherein the preoperative respiratory signal profile data set is a tidal volume profile data set and a ventilation volume profile data set.
3. A method of preoperatively assessing risk of PPCs based on a sub-polar motion test according to claim 1, wherein the tidal volume characterization data set comprises at least tidal volume characterization data for an expiration of a third period of time.
4. A method of preoperatively assessing risk of PPCs based on a sub-maximal exercise test according to claim 1, wherein said six-minute walking test collects continuous respiration signals of the subject one minute before the start of the six-minute walking test, during the six-minute walking test, and one minute after the six-minute walking test.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310096280.XA CN115775629B (en) | 2023-02-10 | 2023-02-10 | Method for assessing risk of PPCs (point-to-point) based on sub-polar motion test before operation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310096280.XA CN115775629B (en) | 2023-02-10 | 2023-02-10 | Method for assessing risk of PPCs (point-to-point) based on sub-polar motion test before operation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115775629A CN115775629A (en) | 2023-03-10 |
| CN115775629B true CN115775629B (en) | 2023-05-02 |
Family
ID=85393466
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310096280.XA Active CN115775629B (en) | 2023-02-10 | 2023-02-10 | Method for assessing risk of PPCs (point-to-point) based on sub-polar motion test before operation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115775629B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116130107A (en) * | 2023-04-14 | 2023-05-16 | 淄博市中心医院 | Valve replacement early clinical evaluation method and system based on time domain analysis |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106999097B (en) * | 2014-11-12 | 2021-05-11 | 皇家飞利浦有限公司 | Device and method for assessing the severity of chronic obstructive pulmonary disease, COPD, in a subject |
| CN112138361B (en) * | 2020-10-14 | 2022-02-11 | 中国科学院合肥物质科学研究院 | Cardio-pulmonary endurance measurement method and system based on oxygen uptake calculation |
| CN115223679A (en) * | 2022-08-05 | 2022-10-21 | 华中科技大学同济医学院附属同济医院 | Perioperative period risk early warning method based on machine learning |
-
2023
- 2023-02-10 CN CN202310096280.XA patent/CN115775629B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN115775629A (en) | 2023-03-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11395634B2 (en) | Estimating physiological states based on changes in CRI | |
| RU2621393C2 (en) | Automated spirogram analysis and interpretation | |
| JP5866335B2 (en) | Measurement of elements of total carbon dioxide excreted by patients | |
| Mann et al. | Simultaneous activity and respiratory monitoring using an accelerometer | |
| CN115775629B (en) | Method for assessing risk of PPCs (point-to-point) based on sub-polar motion test before operation | |
| CN111407262A (en) | Sleep staging method based on B L STM and device for sleep staging based on B L STM | |
| CN112675393A (en) | Ventilator removing management system and method | |
| JP5710168B2 (en) | Biometric apparatus, biometric method, biometric apparatus control program, and recording medium recording the control program | |
| US20070078357A1 (en) | Method and System for Diagnosing Post-Surgical Pulmonary Vascular Occlusions | |
| JP2018531067A6 (en) | Enhanced acute care management combined with imaging and physiological monitoring | |
| JP2018531067A (en) | Enhanced acute care management combined with imaging and physiological monitoring | |
| Paraschiv et al. | Machine learning approaches based on wearable devices for respiratory diseases diagnosis | |
| JP2008212686A (en) | Methodology and display instrument to noninvasively determine pulmonary characteristics through breath analysis and arterial blood measurement | |
| EP2034886A2 (en) | Methods and systems for assessing metabolic transition points | |
| EP3203900A1 (en) | Weaning readiness indicator, sleeping status recording device, and air providing system applying nonlinear time-frequency analysis | |
| CN115775630A (en) | Postoperative lung complication probability prediction method based on sleep stage data before operation | |
| Crognier et al. | Diaphragmatic speckle tracking imaging for 2D-strain assessment in mechanical ventilation weaning test | |
| CN113066547A (en) | ARDS early dynamic early warning method and system based on conventional noninvasive parameters | |
| Akshay et al. | Design & implementation of real time bio-signal acquisition system for quality health care services for the population of rural India | |
| CN105960198A (en) | Intelligent medical monitoring of patient | |
| Ionescu et al. | Analysis of the respiratory dynamics during normal breathing by means of pseudophase plots and pressure–volume loops | |
| Kazemi et al. | Assessment on methods used to detect asthmatic and non-asthmatic conditions using capnogram signal | |
| US20150005591A1 (en) | Tissue to end tidal co2 monitor | |
| EP4079221A1 (en) | System for identifying a condition of a user | |
| Ravelo-García et al. | Application of RR series and oximetry to a statistical classifier for the detection of sleep apnoea/hypopnoea |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |