CN113421656B - Method, system, computer device and storage medium for coagulation real-time early warning - Google Patents
Method, system, computer device and storage medium for coagulation real-time early warning Download PDFInfo
- Publication number
- CN113421656B CN113421656B CN202110964981.1A CN202110964981A CN113421656B CN 113421656 B CN113421656 B CN 113421656B CN 202110964981 A CN202110964981 A CN 202110964981A CN 113421656 B CN113421656 B CN 113421656B
- Authority
- CN
- China
- Prior art keywords
- data
- coagulation
- real
- time
- classification
- 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
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Abstract
The invention belongs to the technical field of medical equipment, and particularly relates to a method, a system, computer equipment and a storage medium for coagulation real-time early warning. The device and the system can realize the following methods: (1) inputting pressure data with time sequence continuously collected by a CRRT (continuous CrRT) machine; (2) preprocessing the pressure data and converting the pressure data into functional data; (3) sliding a sliding window on the functional data, taking out the functional data in the range of the sliding window after each sliding, inputting a functional data classification model for classification, and obtaining a classification result whether coagulation appears in the range of each sliding window; (4) and judging whether to give out real-time early warning for blood coagulation according to the numerical value and/or classification result of the pressure data. The invention can send out the real-time early warning that the patient has the blood coagulation risk in advance, so that medical staff has enough time to process the blood coagulation risk of the patient in advance, and the invention has high application value.
Description
Technical Field
The invention belongs to the technical field of medical equipment, and particularly relates to a method, a system, computer equipment and a storage medium for coagulation real-time early warning.
Background
Continuous Renal Replacement Therapy (CRRT), also known as continuous blood purification, refers to a group of extracorporeal blood purification treatment techniques, and is a generic term for all continuous, slow-clearing water and solute treatment modalities. The main modes include continuous venous-venous hemofiltration (CVVH), continuous venous-venous hemodialysis (CVVHD), continuous venous-venous hemodiafiltration (CVVHDF), and the like. CRRT is the most important treatment technology in the field of severe renal diseases at present, and is widely used for treating critical patients with acute and chronic renal failure caused by various reasons, including severe infection (such as sepsis), crush syndrome (such as earthquake injury), drug or biotoxin poisoning (such as bee sting), renal function damage caused by various reasons after surgery (such as cardiothoracic operation), tumor-related (such as tumor lysis syndrome), autoimmune diseases (such as systemic lupus erythematosus) and the like, and relates to the critical and severe field of multiple disciplines. The CRRT has the advantages of stable hemodynamics, accurate volume balance control, slow and continuous toxin removal, inflammation medium removal, immune function regulation, nutrition supplement guarantee, drug therapy and the like, and is particularly suitable for the support therapy of critical patients.
Ensuring that the extracorporeal circulation pipeline does not generate blood coagulation is one of key core technologies for safe and smooth CRRT treatment. The blood coagulation event can not only directly cause that the CRRT filter can not be used continuously, thereby being forced to interrupt the treatment, but also consume a large amount of body important components such as hemoglobin, platelets, blood coagulation factors and the like of a patient, can obviously increase the related risks of anemia and bleeding of the patient, and directly influences the safety and the effectiveness of the treatment of the patient. In addition, changing filters will also result in increased cost of the treatment consumables and increased care effort.
Therefore, monitoring and prevention of coagulation appears to be very important in CRRT treatment. At present, the mainstream CRRT machine can monitor the cyclic extraction pressure (AOP), the pre-filter pressure (PFP), the waste liquid pressure (EP), and the cyclic feedback pressure (RIP) in real time, and can obtain the transmembrane pressure (TMP) of the filter by calculation: TMP ═ (PFP + RIP)/2-EP, unit mmHg. However, only when a significant clotting event occurs, significant abnormalities in the circuit pressure (primarily significant increases in filter transmembrane pressure and circuit return pressure) can be observed. In the extracorporeal blood purification treatment, the coagulation of extracorporeal circulation lines can be classified into filter coagulation and venous kettle coagulation, and can be classified into three degrees according to the severity. The change in circulation line pressure may also be perturbed by other factors and is not simply correlated with the occurrence of a clotting event. This makes the circulating line pressure sensing function of the CRRT machine unable to monitor all coagulation events. In addition, compared with remedial measures after coagulation, the prevention of coagulation is undoubtedly more important, but the pressure detection through the circulating pipeline can only detect whether the pressure at the current time point exceeds the limit, the trend of the pressure change side cannot be monitored, and therefore early warning on coagulation cannot be given out in advance.
For the reasons mentioned above, in the prior art, the decision as to whether or not coagulation has occurred during CRRT treatment and the severity of coagulation is made primarily by the nursing operator by observing the extracorporeal circulation circuit. However, relying on manual observation of the filter for signs of possible clotting is still not really effective in preventing the clotting events from occurring in advance. This is because the timely finding and predicting of the occurrence of blood coagulation risk is not only influenced by the degree of busy nursing work of the medical staff, but also is related to the individual actual work experience of the medical staff. Therefore, the manual monitoring method still lacks sufficient timeliness and stable early warning performance.
Although artificial intelligence has been used for preventing and treating diseases, for example, the chinese patent application "CN 201911146419.7 early death risk assessment model building method and apparatus based on ensemble learning" provides a model and apparatus for assessing the risk of patients in intensive care unit. However, the accuracy of the artificial intelligence prediction result has a great relationship with the model construction method, the model type selection, the model parameter selection, the input data processing method, and the like. Thus, the existing techniques of these artificial intelligence to predict risk in the medical field are still not applicable to the process of CRRT treatment. Therefore, there is a need to propose a new artificial intelligence risk prediction scheme for CRRT treatment.
Disclosure of Invention
The device and the system provided by the invention can be used for analyzing based on pressure data acquired by a CRRT (CrRT) machine, obtaining classification results of whether coagulation is about to occur in each small-range time period by using a sliding window, and then judging whether coagulation real-time early warning is given out in real time based on the numerical value of the pressure data and the classification results of whether coagulation is about to occur in a period of time.
A method for real-time early warning of coagulation comprising the steps of:
(1) inputting pressure data with time sequence continuously collected by a CRRT (continuous CrRT) machine;
(2) preprocessing the pressure data and converting the pressure data into functional data;
(3) sliding a sliding window on the functional data, taking out the functional data in the range of the sliding window after each sliding, inputting a functional data classification model for classification, and obtaining a classification result whether coagulation is about to occur in each sliding window range;
(4) and (4) judging whether coagulation real-time early warning is given according to the numerical value of the pressure data obtained in the step (1) and/or the classification result obtained in the step (3).
Preferably, in step (1), the pressure data is discrete data continuously collected by a CRRT machine, and the discrete data is pressure data of one data point per second;
and/or the pressure data is at least one of transmembrane pressure data, recycle draw pressure, pre-filter pressure, waste hydraulic pressure, or recycle return pressure.
Preferably, in the step (2), the pretreatment comprises the following steps: and converting the pressure data into functional data by using a basis function expansion method.
Preferably, in the step (2), in the basis function expansion method, the basis system is a B-sample strip basis system, and the estimation of the coefficient vector is a least square estimation.
Preferably, in the step (3), the classification model performs curve supervised classification prediction on the functional data by using a correction tape depth of the functional data.
Preferably, in the step (3), the classification model adopts a classification rule based on the functional data depth MBD, and the classification rule is a D method or a TAD method;
the D method is to calculate the distance between the newly observed curve of the functional data and the alpha-truncation mean value of the known curve class group and classify the newly observed curve and the known curve class group;
the TAD method calculates the weighted average distance between the newly observed curve of the functional data and each element of the known curve class, and classifies it.
Preferably, in the step (3), when the classification model is trained, the data of the sample includes pressure data in a T time period and a label of whether blood coagulation occurs at the end of the T +. DELTA.t time period; the value of T is 2-4h, and the value of delta T is 10-30 min;
and/or randomly drawing 2/3 of all samples as a training set, 1/3 as a testing set, repeating the random sampling for 500 to 1000 times, and training the classification model by calculating the error rate of the testing set.
Preferably, in the step (3), the sliding window width is 2-4h, the sliding interval is t, the t value is 2-5min, and a classification result of normal or impending blood coagulation is continuously obtained by using a classification model;
in the step (4), when one of the following two conditions is judged to occur, blood coagulation real-time early warning is sent out:
in case one, step (3) obtains k consecutive classification results, wherein more than two thirds of the classification results are about to cause coagulation, and the last data point value of the pressure data obtained in step (1) is greater than 180 mmHg; the value of k is 3-9;
in case two, the pressure data obtained in step (1) has a value of 5 or more continuous data points greater than 250 mmHg.
The invention also provides computer equipment for continuous real-time coagulation warning of kidney substitution treatment, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, and is characterized in that the processor realizes the steps of the method for real-time coagulation warning when executing the program.
The invention also provides a system for coagulation real-time early warning, comprising:
a pressure data acquisition and/or input device for acquiring and/or inputting pressure data having a time sequence;
the computer equipment is used for judging whether coagulation real-time early warning is given or not in real time;
and the early warning device is used for receiving the coagulation real-time early warning and giving an alarm.
The invention also provides a computer-readable storage medium, on which a computer program for implementing the steps of the above-described method for coagulation real-time warning is stored.
In the present invention, the "pressure data" refers to pressure parameters that can be monitored or calculated by the CRRT machine in real time, such as: cyclic extraction pressure (AOP), pre-filter pressure (PFP), waste liquor pressure (EP), cyclic return pressure (RIP), and transmembrane pressure (TMP) of the filter, among others. The "functional data" refers to smooth continuous data, and for a discrete series of data points, the data can be converted into functional data by a non-parametric smoothing technique (such as a basis function expansion method).
The invention realizes the real-time monitoring and prediction of whether a patient will generate blood coagulation in the CRRT treatment process by using a machine learning method for the first time. Because the classification model in the method is based on the functional data, the change trend of the pressure data can be identified and judged, so that early warning is given out in advance before severe blood coagulation occurs, medical personnel have enough time to treat a risk patient in advance, and the safety of CRRT treatment is greatly improved.
Obviously, many modifications, substitutions, and variations are possible in light of the above teachings of the invention, without departing from the basic technical spirit of the invention, as defined by the following claims. Such as using other functional data depth definitions (e.g., FM depth, FSD depth, KFSD depth), other classifiers (e.g., non-parametric KNN, etc.), or ensemble classifiers.
The present invention will be described in further detail with reference to the following examples. This should not be understood as limiting the scope of the above-described subject matter of the present invention to the following examples. All the technologies realized based on the above contents of the present invention belong to the scope of the present invention.
Drawings
FIG. 1 is a schematic flow chart of an embodiment of the present invention.
Detailed Description
It should be noted that, in the embodiment, the algorithm of the steps of data acquisition, transmission, storage, processing, etc. which are not specifically described, as well as the hardware structure, circuit connection, etc. which are not specifically described, can be implemented by the contents disclosed in the prior art.
Embodiment 1A continuous kidney replacement therapy coagulation real-time early warning system
The system of the embodiment comprises: CRRT computer, computer device and early warning device with filter transmembrane pressure (TMP) calculation function. The CRRT machine is connected with the computer device through a data line and a data interface, so that the CRRT machine can transmit pressure data acquired in real time to computer equipment. The early warning device can be an alarm, and also can be a message sending module integrated in computer equipment, and is used for sending the real-time early warning message of blood coagulation to medical personnel in time.
The computer device of this embodiment includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the program, as shown in fig. 1, the following workflow is implemented:
1. training data preprocessing
And in the existing time sequence data set, selecting the TMP value of each patient in the last T plus Delta T minutes for analysis, and judging whether the patient has the blood coagulation risk needing early warning after the label of the sample is T plus Delta T minutes. Intercepting the TMP value of the first T minutes in the Delta T minutes for modeling, and reserving the Delta T time as early warning time. In this embodiment, T takes 3 hours and Δ T takes 10 min.
The equal time interval based on TMP of the invention(once per second) time series data for each patient's TMP observation { y ] assuming that the actual observation (i.e., pressure data) was generated by a potentially continuous function processij-is seen as a whole, represented by a smooth curve or continuous function. Wherein, yijRepresenting the jth observed pressure value for the ith object.
Due to the influence of observation errors or other random factors, the finally obtained observation value is recorded as:
wherein n isiRefers to the number of data points of TMP data of the ith observation object. Thus, the data set { (t)ij,yij):i＝1,2,…,n；j＝1,2,…,niThe following model is satisfied:
yij＝xi(tij)+εij
wherein n represents the number of observation targets, i.e., the number of TMP curves of the patient, t represents the observation timeijThe j observation time, x, of the i observation objectiFor the pressure curve of the ith observation object, ε represents the noise, εijThe noise is the noise of the jth observation time of the ith observation object. Including random errors in the observed data, and assuming that its mean is zero and the variance is constant, it is recorded as:
Eεij＝0,varεij＝σ2
since the observation data is discrete time series data, it is first necessary to convert the discrete observation data into smooth functional data, where the continuity and smoothness of the data are realized by the basis function expansion method in the non-parametric smoothing technique.
A basis function expansion method: first, a base system is selected, the base function is phik(t) (K ═ 1,2, …, K), the linear combination of basis functions in the basis system is then used to estimate the function x (t) (for each x (t))i(t) all perform the same fitting estimation, so the subscript i) is omitted from the following expression, i.e.
Wherein the coefficient vector is a basis function vector and the coefficient vector is
Next is selecting a base system and estimating a coefficient vector.
(1) B-like strip-based System (refer to the prior art James Ramsay, Giles Hooker, Spencer Graves. functional Data Analysis with R and MATLAB.2009, [1])
The choice of basis functions is important for data analysis, with fourier basis functions, which are often used to analyze data with periodicity, and B-spline basis functions, which are the two most commonly used types of basis functions for data without significant periodicity. The invention selects the commonly used B-spline basis function suitable for the non-periodic function to carry out curve fitting. Setting:
first, all curves are represented using the same base;
second, the basis function is defined as an equidistant node B spline basis function.
(2) Estimation of coefficient vectors
After the basis functions are selected, the coefficient vectors in the basis function expansions need to be estimated. Suppose the observed data of the function curve x (t) is { (t)j，yj): j 1,2,.. n }, the model obeyed is yj＝x(tj)+εjCombining the expansion of x (t) to obtain a model
y＝φc+ε
Wherein y ═ y1，...，yn)′，c＝(c1，...，ck)′，ε＝(ε1，...，εn) ' is a random error term, satisfies independent distribution assumptions, and E ∈ 0, ν α r ∈ ═ σ2In(ii) a Phi is an element of phik(tj) N rows and k columns of matrix. Estimating coefficientsThe most common method is the least squares method, i.e. minimizing the sum of the squared residuals of the fit:
writing in matrix form:
the least squares estimate of the coefficient vector is:
similarly, a weighted least squares estimate considering the case where the random error term ε is an heteroscedastic or correlated term is:
where W is a positive definite weight matrix.
To this end, each pressure data xi(t) the fit may be developed by a B-spline basis function.
2. Depth calculation of each group of truncated mean values based on functional data
The statistical depth of the functional data represents a measure of "centrality" or "externality" of the observations within a set of data, providing an index that orders the observations from the centre to the outside. This embodiment uses the correction tape (MBD) depth, which is a method based on defining a tape graphically on a plane, is easy to calculate, and can flexibly handle irregular curves.
The curve depth calculation is carried out with reference to the prior art (Journal of the American Statistical Association.2009,104, 718-734).
Based on the function type data depth, the alpha-truncation mean value can be obtained, let x(1),x(2),…,x(n)Is based on the MBD center-out ordering of samples, then the α -truncated mean is:
wherein m isαIs an alpha-truncated mean, [ n α ]]Is an integer part of n α.
3. Classification rules
Classification rules based on functional data depth MBD: d and TAD processes. The two methods are to judge the attribution of a new sample of an unknown class on the basis of training data of the known class. In the known curve class group A1,…,AGAny new observation is assigned to one of the G groups. The flow of two classification methods is described below:
d: the distance of the new observed value to the alpha-truncated mean of the known class group is calculated, as follows,
(1) calculate the average value of each group of α -truncations (given α, the value is between 10% and 30%, and the specific value in this embodiment is 10%):
(2) calculating a new observed value x and
the distance of (c):
wherein is defined as [0, 1]]L between the upper two curves x and y1Distance d:
(3) classification rule of x: classified as k when
Where D has a special value, when α is 0, the distance is calculated as the new viewThe distance between the measured value and the mean of all the curves in the group of known classes.
TAD classification method: a weighted average distance of the new observation to each element in the set of known curve classes is calculated. Let Ag＝{x1,…,xngThe new observed value x and the g-th group of curves AgThe weighted average distance of (d) is:
where S is the depth of each observation, ngIs AgThe significance of weighting is to use the depth of each observation itself, with deeper depths giving greater weight and hence distance affected by depth. If the observed values in each group are different greatly, the classification effect is unstable. Therefore, the following modified AD classification method, TAD, is employed: calculating the distance by considering m deepest observed values in each group, and fixing m to be less than or equal to n1,…,ngAnd then:
4. estimating error rate distribution
In order to estimate the error rate distribution, 2/3 of all samples is randomly sampled to be used as a training set, 1/3 is used as a testing set, the random sampling is repeated for multiple times (1000 times can be taken), the error rate e is calculated for different classification methods each time, and the model is trained and optimized by using the error rate e. The error rate e is the number of samples with classification errors in the test set T to the total number of samples nTIn a ratio of (i) to (ii)
5. Real-time early warning process after model building
When a patient is subjected to continuous renal replacement therapy, the real-time early warning comprises the following steps:
(1) inputting pressure data with time sequence continuously collected by a CRRT (continuous CrRT) machine;
(2) preprocessing the pressure data in the same way as the training data and converting the pressure data into functional data;
(3) sliding a sliding window on the functional data, taking out the functional data in the range of the sliding window after each sliding, inputting a functional data classification model for classification, and obtaining a classification result whether coagulation appears in the range of each sliding window; in this embodiment, the sliding time range (window width) is equal to the value of T, the sliding interval is T, and the classification result of normal or impending blood coagulation is continuously obtained by using the classification model. In this embodiment, t is 5 min.
(4) And (4) judging whether coagulation real-time early warning is given according to the numerical value of the pressure data obtained in the step (1) and/or the classification result obtained in the step (3).
In this embodiment, when it is determined that one of the following two conditions occurs, a blood coagulation real-time warning is issued:
in case one, of the k (k value is 6 in this embodiment) consecutive classification results obtained in step (3), more than two thirds of the classification results are about to cause coagulation, and the last data point value of the pressure data obtained in step (1) is greater than 180 mmHg;
in case two, the pressure data obtained in step (1) has a value of 5 or more continuous data points greater than 250 mmHg.
When the method of the embodiment is adopted to carry out real-time early warning on blood coagulation of 89 blood coagulation patients and a functional data D classification method based on depth (MBD) is used for judging whether blood coagulation is about to occur, the average error rate is 15.8%. The preset alarm range of the current most dialysis machines is TMP >250mm Hg, and the early warning error rate of 89 blood coagulation patients is 45.07 percent according to the existing dialysis machine alarm mechanism by intercepting the TMP value of the 89 blood coagulation patients in a half hour before the blood coagulation. The above results show that: the method of the embodiment has the advantage that the alarm accuracy is remarkably improved.
The embodiment shows that the method can analyze the pressure data which is continuously collected by the CRRT and has time sequence, and can early warn the risk of the impending blood coagulation of the patient in advance in real time, so that medical staff have enough time to process the blood coagulation risk of the patient in advance. The invention can obviously improve the safety and effectiveness of continuous renal replacement therapy and has high application value.
Claims (8)
1. A method for real-time early warning of coagulation comprising the steps of:
(1) inputting pressure data with time sequence continuously collected by a CRRT (continuous CrRT) machine;
(2) preprocessing the pressure data and converting the pressure data into functional data;
(3) sliding a sliding window on the functional data, taking out the functional data in the range of the sliding window after each sliding, inputting a functional data classification model for classification, and obtaining a classification result whether coagulation is about to occur in each sliding window range;
(4) judging whether coagulation is sent out in real time according to the numerical value of the pressure data obtained in the step (1) and the classification result obtained in the step (3);
in the step (3), the classification model adopts a classification rule based on functional data depth MBD, and the classification rule is a D method or a TAD method;
the D method is to calculate the distance between the newly observed curve of the functional data and the alpha-truncation mean value of the known curve class group and classify the newly observed curve and the known curve class group;
the TAD method is to calculate the weighted average distance between the newly observed curve of the functional data and each element of the known curve class and classify the curve;
when the classification model is trained, the data of the sample comprisesTPressure data over time period andT +△Ta label of whether clotting has occurred at the end of the time period; the above-mentionedTThe value is 2-4h, and the deltaTThe value is 10-30 min;
the sliding window width is 2-4h, the sliding interval is t, the t value is 2-5min, and a classification result of normal or impending blood coagulation is continuously obtained by utilizing a classification model;
in the step (4), when one of the following two conditions is judged to occur, blood coagulation real-time early warning is sent out:
in case one, step (3) obtains k consecutive classification results, wherein more than two thirds of the classification results are about to cause coagulation, and the last data point value of the pressure data obtained in step (1) is greater than 180 mmHg; the value of k is 3-9;
in case two, the pressure data obtained in step (1) has a value of 5 or more continuous data points greater than 250 mmHg.
2. The method for coagulation real-time warning according to claim 1, wherein: in the step (1), the pressure data is discrete data continuously collected by a CRRT (continuous CrRT) machine, and the discrete data is pressure data of one data point per second;
the pressure data is at least one of transmembrane pressure data, circulating extraction pressure, pre-filter pressure, waste hydraulic pressure or circulating feedback pressure.
3. The method for coagulation real-time warning according to claim 1, wherein: in the step (2), the pretreatment is as follows: converting the pressure data into functional data by using a basis function expansion method; in the basis function expansion method, a B sample strip basis system is selected as a basis system, and the estimation of a coefficient vector is least square estimation.
4. The method for coagulation real-time warning according to claim 1, wherein: in the step (3), the classification model performs curve supervision, classification and prediction on the functional data by using the correction tape depth of the functional data.
5. The method for coagulation real-time warning according to claim 1, wherein:
in the step (3), when the classification model is trained, 2/3 of all samples is randomly extracted as a training set, 1/3 of all samples is used as a test set, random sampling is repeated for 500 to 1000 times, and the classification model is trained by calculating the error rate of the test set.
6. A computer device for continuous real-time pre-warning of coagulation in renal replacement therapy, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the steps of the method for real-time pre-warning of coagulation as claimed in any one of claims 1 to 5 when executing the program.
7. A system for real-time early warning of coagulation, comprising:
a pressure data acquisition and/or input device for acquiring and/or inputting pressure data having a time sequence;
the computer device of claim 6, configured to determine in real time whether to issue a coagulation real-time alert;
and the early warning device is used for receiving the coagulation real-time early warning and giving an alarm.
8. A computer-readable storage medium characterized by: stored thereon a computer program for carrying out the steps of the method for real-time pre-warning of coagulation as claimed in any one of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110964981.1A CN113421656B (en) | 2021-08-23 | 2021-08-23 | Method, system, computer device and storage medium for coagulation real-time early warning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110964981.1A CN113421656B (en) | 2021-08-23 | 2021-08-23 | Method, system, computer device and storage medium for coagulation real-time early warning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113421656A CN113421656A (en) | 2021-09-21 |
| CN113421656B true CN113421656B (en) | 2021-11-26 |
Family
ID=77719130
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110964981.1A Active CN113421656B (en) | 2021-08-23 | 2021-08-23 | Method, system, computer device and storage medium for coagulation real-time early warning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113421656B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104305960A (en) * | 2010-09-30 | 2015-01-28 | 泰科保健集团有限合伙公司 | Monitoring compliance using venous refill detection |
| CN104363938A (en) * | 2012-12-31 | 2015-02-18 | 甘布罗伦迪亚肾脏产品公司 | Occlusion detection in delivery of fluids |
-
2021
- 2021-08-23 CN CN202110964981.1A patent/CN113421656B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104305960A (en) * | 2010-09-30 | 2015-01-28 | 泰科保健集团有限合伙公司 | Monitoring compliance using venous refill detection |
| CN104363938A (en) * | 2012-12-31 | 2015-02-18 | 甘布罗伦迪亚肾脏产品公司 | Occlusion detection in delivery of fluids |
Non-Patent Citations (4)
| Title |
|---|
| Aadil Kakajiwala等.Membrane pressures predict clotting of pediatric continuous renal replacement therapy circuits.《Pediatr Nephrol》.2017,第32卷(第7期), * |
| DEPTH-BASED CLASSIFICATION FOR FUNCTIONAL DATA;Sara Lopez等;《Researchgate(https://www.researchgate.net/publication/4849658)》;20051130;第1-16页 * |
| Detection of Anomalies in Water Networks by Functional Data Analysis;Laura Millan-Roures等;《Mathematical Problems in Engineering》;20180621;第1-14页 * |
| Membrane pressures predict clotting of pediatric continuous renal replacement therapy circuits;Aadil Kakajiwala等;《Pediatr Nephrol》;20170731;第32卷(第7期);第1-18页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113421656A (en) | 2021-09-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP0495412B1 (en) | Hemodialysis method | |
| US10368757B2 (en) | Methods and systems for adverse event prediction using pump operating data | |
| US20090149724A1 (en) | System and method for predicting septic shock | |
| CN112365978A (en) | Method and device for establishing early risk assessment model of tachycardia event | |
| CN107252313A (en) | The monitoring method and system of a kind of safe driving, automobile, readable storage medium storing program for executing | |
| Campbell et al. | An application of Lomax distributions in receiver operating characteristic (ROC) curve analysis | |
| CN111553478B (en) | Community old people cardiovascular disease prediction system and method based on big data | |
| CN115083604B (en) | Method, system and prediction device for constructing postoperative adverse event prediction model | |
| CN113421656B (en) | Method, system, computer device and storage medium for coagulation real-time early warning | |
| Gospodinova et al. | Automatic analysis of ECG signals based on their fractal and multifractal properties | |
| CN117312997B (en) | Intelligent diagnosis method and system for power management system | |
| CN111317458A (en) | Blood pressure detection system based on deep learning | |
| Acharya et al. | A systems approach to cardiac health diagnosis | |
| CN116831546A (en) | Method, device, equipment and storage medium for predicting abnormal degree of fetal heart rate curve | |
| WO2020220403A1 (en) | Fatigue state detection method, apparatus and device, and storage medium | |
| Sang et al. | Study on survival prediction of patients with heart failure based on support vector machine algorithm | |
| CN112905463B (en) | Software test monitoring method and device, electronic equipment and readable storage medium | |
| Wong et al. | Probabilistic detection of vital sign abnormality with Gaussian process regression | |
| Akbilgic et al. | Categorizing atrial fibrillation via symbolic pattern recognition | |
| Baig et al. | Detection and classification of hypovolaemia during anaesthesia | |
| Clifton et al. | Novelty detection for identifying deterioration in emergency department patients | |
| Gilani et al. | Automated classification of congestive heart failure severity using time domain, frequency domain and non-linear heart rate variability measures | |
| CN111489828A (en) | Method for screening thrombus forming factors after fracture | |
| Slaughter et al. | Prediction of mortality | |
| JP7333549B2 (en) | Disease risk assessment method, disease risk assessment system, and health information processing device |
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 |