CN113936801A - Machine learning fusion-based general anesthesia induced contraction compression prediction method and system - Google Patents
Machine learning fusion-based general anesthesia induced contraction compression prediction method and system Download PDFInfo
- Publication number
- CN113936801A CN113936801A CN202111207748.5A CN202111207748A CN113936801A CN 113936801 A CN113936801 A CN 113936801A CN 202111207748 A CN202111207748 A CN 202111207748A CN 113936801 A CN113936801 A CN 113936801A
- Authority
- CN
- China
- Prior art keywords
- prediction
- primary
- learner
- machine learning
- training
- 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.)
- Pending
Links
- 238000010801 machine learning Methods 0.000 title claims abstract description 42
- 238000000034 method Methods 0.000 title claims abstract description 37
- 238000002695 general anesthesia Methods 0.000 title claims abstract description 36
- 230000008602 contraction Effects 0.000 title claims abstract description 31
- 230000006835 compression Effects 0.000 title claims abstract description 24
- 238000007906 compression Methods 0.000 title claims abstract description 24
- 230000004927 fusion Effects 0.000 title claims abstract description 23
- 238000012549 training Methods 0.000 claims abstract description 46
- 238000012360 testing method Methods 0.000 claims abstract description 31
- 238000012795 verification Methods 0.000 claims abstract description 30
- 238000005070 sampling Methods 0.000 claims abstract description 15
- 238000007781 pre-processing Methods 0.000 claims abstract description 7
- 230000003993 interaction Effects 0.000 claims abstract description 5
- 230000008569 process Effects 0.000 claims description 11
- 238000012545 processing Methods 0.000 claims description 11
- 230000009467 reduction Effects 0.000 claims description 11
- 238000003066 decision tree Methods 0.000 claims description 6
- 238000012417 linear regression Methods 0.000 claims description 5
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 244000025254 Cannabis sativa Species 0.000 claims 1
- 235000012766 Cannabis sativa ssp. sativa var. sativa Nutrition 0.000 claims 1
- 235000012765 Cannabis sativa ssp. sativa var. spontanea Nutrition 0.000 claims 1
- 235000009120 camo Nutrition 0.000 claims 1
- 235000005607 chanvre indien Nutrition 0.000 claims 1
- 239000011487 hemp Substances 0.000 claims 1
- 230000010388 wound contraction Effects 0.000 claims 1
- 238000004422 calculation algorithm Methods 0.000 abstract description 9
- 230000007547 defect Effects 0.000 abstract description 4
- 230000035488 systolic blood pressure Effects 0.000 description 8
- 238000012544 monitoring process Methods 0.000 description 4
- 206010002091 Anaesthesia Diseases 0.000 description 3
- 206010020772 Hypertension Diseases 0.000 description 3
- 230000037005 anaesthesia Effects 0.000 description 3
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 description 3
- 239000008280 blood Substances 0.000 description 3
- 210000004369 blood Anatomy 0.000 description 3
- 229910052760 oxygen Inorganic materials 0.000 description 3
- 239000001301 oxygen Substances 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000002360 preparation method Methods 0.000 description 2
- 238000000513 principal component analysis Methods 0.000 description 2
- 238000007637 random forest analysis Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 241000134253 Lanka Species 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 210000001367 artery Anatomy 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000036772 blood pressure Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000035622 drinking Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000000391 smoking effect Effects 0.000 description 1
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
- G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- 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
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Public Health (AREA)
- Data Mining & Analysis (AREA)
- Biomedical Technology (AREA)
- Primary Health Care (AREA)
- Pathology (AREA)
- Epidemiology (AREA)
- General Health & Medical Sciences (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a prediction method and a system of general anesthesia invasive contraction compression based on machine learning fusion, wherein the method comprises the steps of S1, collecting physical characteristic data and vital sign data of a patient, preprocessing each data, and dividing the preprocessed data into a training set, a test set and a verification set; respectively carrying out slice sampling on the training set, the test set and the verification set to obtain n training subsets, test subsets and verification subsets; s2, constructing a prediction model based on machine learning fusion, and predicting the general anesthesia induced contraction pressure by using the prediction model; the prediction model comprises n primary learners and one secondary learner, and the output values of all the primary learners are used as the input of the secondary learner. The system comprises a database module, a data acquisition module, a prediction module and an interaction module; and a prediction program is stored in the prediction module. The invention overcomes the defect of low prediction precision of a single algorithm, optimizes the input of the secondary learner and further improves the prediction precision.
Description
Technical Field
The invention belongs to the technical field of medical surgical operation monitoring, and particularly relates to a general anesthesia created contraction compression prediction method and system based on machine learning fusion.
Background
In general anesthesia operation, the initial systolic pressure is an important basis for doctors to judge the vital sign condition of patients, the smooth operation can be directly influenced, the change of the initial systolic pressure in the general anesthesia operation can be accurately predicted, and the operation risk can be effectively reduced by making a response strategy in advance. At present, a life monitoring instrument is generally used for monitoring the variables of the initial systolic pressure, the blood oxygen, the heart rate and the like in real time in the operation process, doctors are required to artificially predict the initial systolic pressure according to clinical experience, and the result of artificial prediction is not ideal. On one hand, the judgment of the situation is different due to different clinical experience and professional degree of doctors; on the other hand, the human approach has limited prediction accuracy for continuous variables. In the operation process, if the induced contraction and compression after general anesthesia needs to be predicted quickly and accurately, a large amount of professional knowledge is required, and the comprehensive influence of various factors in the operation process is also considered, so that the artificial mode is difficult to predict accurately. The manual prediction mode not only has high requirements on the clinical experience of doctors, but also needs to spend a great deal of time and energy to influence the operation.
With the development of machine learning technology, the machine learning algorithm can not only learn a large amount of knowledge through a mathematical model and a mechanism formula, but also dig out deep association or rules among data, and can quickly and accurately calculate and process body characteristic data, basic illness condition and vital sign data of a patient. In addition, the data volume required by machine learning is small, the prediction of the created contraction pressure after general anesthesia can be realized by establishing an off-line model, the prediction speed is high, the accuracy is high, a large amount of time and energy are saved, and the safety of anesthesia and the safety of operation are guaranteed.
T.h.wu et al (t.h.wu, g.k.pang and e.w.kwong, "Predicting symptomatic Blood Pressure Using Machine Learning" in 7th International Conference on Information and Automation for sustamability, colomobo, Sri Lanka,2014, pp.1-6.) utilize Machine Learning techniques to predict Systolic Blood Pressure by modeling of characteristic variables such as Body Mass Index (BMI), age, motion, drinking conditions, and smoking conditions, utilize a back-propagation neural network and a radial basis function network to construct and verify a prediction system that has a highest prediction accuracy of 69.9% at an absolute error of 15mmHg and a lower prediction accuracy.
In summary, the invention provides a machine learning fusion-based general anesthesia invasive contraction compression prediction method and system, a prediction model is constructed by adopting a machine learning fusion idea, and the prediction precision is obviously improved.
Disclosure of Invention
Aiming at the defects in the prior art, the invention aims to solve the technical problem of providing a machine learning fusion-based general anesthesia created contraction compression prediction method and system.
The technical scheme adopted by the invention for solving the technical problems is as follows:
a machine learning fusion-based general anesthesia created contraction compression prediction method is characterized by comprising the following steps:
s1, collecting body characteristic data and vital sign data of a patient, preprocessing each data, and dividing the preprocessed data into a training set, a test set and a verification set; respectively carrying out slice sampling on the training set, the test set and the verification set to obtain n training subsets, test subsets and verification subsets;
s2, constructing a prediction model based on machine learning fusion, and predicting the general anesthesia induced contraction pressure by using the prediction model; the prediction model comprises n primary learners and one secondary learner, and the output values of all the primary learners are used as the input of the secondary learner.
Step S2 includes S2-1, constructing the primary learner: training various machine learning models by using a training set, and optimizing the trained models according to a test set to obtain optimized models; predicting according to the optimized model by using the verification set, and selecting n machine learning models with high prediction precision to respectively construct primary learners to obtain n primary learners;
respectively training n primary learners by using the n training subsets obtained in the step S1, and performing parameter optimization on the respective corresponding primary learners through the n test subsets to obtain optimized primary learners; predicting according to the n optimized primary learners by using the n verification subsets;
s2-2, constructing a secondary learner: and taking the predicted values of all the primary learners as the input of the secondary learner, and training the secondary learner to obtain the trained secondary learner.
Step S2-2 further includes optimizing the input of the secondary learner; the method specifically comprises the following steps: giving a weight to each primary learner according to the prediction error of the primary learners, wherein the larger the prediction error is in the weight assignment process, the lower the weight is;
let the input matrix of the secondary learner be
h1(x)、h2(x)、…、hn(x) Are respectively the predicted values, omega, of the respective primary learners1、ω2、…、ωnWeights of the predicted values of the primary learners, where1+ω2+…+ωn＝1；
And (3) adjusting the weight by using a coefficient of variation method, wherein the coefficient of variation of each primary learner satisfies the formula (2):
in the formula (2), ViCoefficient of variation, σ, of the ith primary learneriIs the standard deviation of the predicted value and the true value of the ith primary learner, hi' (x) is the true value corresponding to the ith primary learner;
the weight of the ith primary learner predicted value satisfies equation (3).
Preprocessing comprises missing value filling, attribute identification, standardization processing and dimension reduction processing; the specific process of slice sampling is as follows: training the sample data set after dimensionality reduction by using a gradient lifting decision tree to obtain the probability of each feature, namely a feature importance list;
dividing the sample data set after dimensionality reduction into a training set, a test set and a verification set, and respectively copying the training set, the test set and the verification set for n times, wherein n is a positive integer; and respectively carrying out n times of slice sampling on the copied training set, test set and verification set according to the sample sampling proportion, and then carrying out feature selection on the feature importance list according to the feature sampling proportion to obtain n training subsets, test subsets and verification subsets.
The number of the primary learners is three, and the three primary learners are respectively constructed by adopting a support vector regression model, a polynomial regression model and a linear regression model; the secondary learner is constructed using a ridge regression model.
A general anesthesia created contraction compression prediction system based on machine learning fusion is characterized by comprising a database module, a data acquisition module, a prediction module and an interaction module; and a prediction program is stored in the prediction module and used for predicting the initial contraction and compression of the general anesthesia according to the method.
Compared with the prior art, the invention has the beneficial effects that:
1. the prediction model is constructed based on a machine learning fusion idea, the prediction model comprises a plurality of primary learners and a secondary learner, the plurality of primary learners are obtained based on different machine learning algorithms through training, and prediction results of all the primary learners are used as input of the secondary learner, so that the defect of low prediction precision of a single algorithm is effectively overcome; in order to further improve the accuracy of the prediction model, the prediction result of each primary learner is weighted according to the prediction error of the primary learner, the larger the prediction error is, the lower the weight is, the input of the secondary learner is optimized, and the prediction accuracy of the prediction model is further improved.
2. The experimental result shows that the accuracy of the prediction of the invasive systolic contraction pressure of the general anesthesia of the invention is up to 91.5%, compared with the research of T.H.Wu et al, the accuracy is improved by about 20% under the condition of the same absolute error, even under the condition of different research background and characteristic variables, the prediction precision of the invention is higher, the quick and accurate prediction of the invasive systolic contraction pressure of the general anesthesia of the patient can be realized, the doctor can make a response strategy within sufficient time, the safety of anesthesia is ensured, and the risk of the operation is reduced to a great extent.
3. The characteristics collected by the invention comprise physical characteristic data and vital sign data of a patient, the composite influence of multiple factors on the invasive systolic pressure of general anesthesia is considered to the maximum extent, most of the factors influencing the result are considered from the perspective of medical professional knowledge and the actual situation of an operating room site, and the data is deeply mined by utilizing a machine learning algorithm, so that the defect that medical staff judge the invasive systolic pressure of the general anesthesia according to the vital sign data is overcome.
Drawings
FIG. 1 is an overall flow diagram of the present invention;
FIG. 2 is a flow chart of the present invention for constructing a predictive model;
fig. 3 is a schematic diagram of the system of the present invention.
Detailed Description
The technical solution of the present invention will be clearly and completely described below with reference to the specific embodiments and the accompanying drawings, but the scope of the present invention is not limited thereto.
The invention relates to a machine learning fusion-based general anesthesia created contraction compression prediction method (a method for short, see figures 1-2), which comprises the following steps:
s1, collecting body characteristic data and vital sign data of a patient, preprocessing each data, and dividing the preprocessed data into a training set, a test set and a verification set; respectively carrying out slice sampling on the training set, the test set and the verification set to obtain n training subsets, test subsets and verification subsets;
step S1-1, the physical characteristic data comprises the age, sex, height, weight, ASA grade and whether the patient suffers from hypertension; the body characteristic data and the basic diseased condition are timely input into a database before the operation, so that the data can be timely extracted from the database when the patient needs to be predicted; the prediction of the general anesthesia induced contraction compression is influenced by various factors, and the physical characteristic data and the basic diseased condition of the patient are not enough to accurately predict the general anesthesia induced contraction compression of the patient;
vital sign data includes three characteristics of heart rate, invasive systolic pressure, and blood oxygen saturation (SaO 2); the vital sign data acquisition time period is an operation preparation stage when a patient just enters an operating room, and the data acquisition device is a vital monitor equipped for the operating room and has an acquisition time interval of 60 s; in the operation preparation stage, the patient has no interference of other factors except the operation of the artery needle and the anesthesia, so that the authenticity and the accuracy of prediction are guaranteed;
step S1-2, preprocessing comprises missing value filling, attribute identification, standardization processing and dimension reduction processing; if the body characteristic data is missing, deleting the sample; for vital sign data, filling individual missing values by adopting a nearest neighbor interpolation method; performing attribute identification on the characteristics filled with the missing values, namely performing 0-1 assignment processing on the gender and whether the patients suffer from hypertension respectively, processing the ASA grade according to discrete variables, dividing the ASA grade into five grades, and correspondingly assigning 1-5 values respectively; processing other characteristics according to continuous variable;
the physical characteristic data and the vital sign data of the patient are complete and consistent in unit, but the data are abnormal due to misoperation, instrument failure or other reasons; the data lack one or more corresponding data, the data is defined as missing data, and the sample is deleted when the physical characteristic data of the patient are missing; when the vital sign data of the patient is missing, filling the missing data with the adjacent data;
carrying out standardization processing on the characteristic after attribute identification through an equation (1), wherein the standardization processing is to scale characteristic values in proportion so that each characteristic value falls into a small specific interval;
wherein, X is the original characteristic value,
is the mean value of the original feature sequence, S2Is the variance of the signature sequence;
carrying out dimensionality reduction on all standardized features by adopting a Principal Component Analysis (PCA), screening out features with high correlation with the yield of general anesthesia, and obtaining a sample data set after dimensionality reduction;
step S1-3, training the sample data set after dimensionality reduction by using a Gradient Boosting Decision Tree (GBDT), and obtaining the probability of each feature, wherein the more important the feature is, the greater the probability is, so that a probability list reflecting the importance of each feature, namely a feature importance list is obtained; selecting features according to the feature importance list, filtering redundant features, and constructing a prediction model with higher efficiency and lower consumption;
dividing the sample data set after dimensionality reduction into a training set, a test set and a verification set, and respectively copying the training set, the test set and the verification set for n times, wherein n is a positive integer; respectively carrying out n times of slicing sampling on the copied training set, test set and verification set according to the sample sampling proportion, and then carrying out feature selection on the feature importance list according to the feature sampling proportion to obtain n training subsets, test subsets and verification subsets;
step S2, constructing a prediction model based on machine learning fusion, wherein the prediction model comprises n primary learners and a secondary learner;
step S2-1, constructing a primary learner;
training a plurality of machine learning models by using a training set, optimizing the trained models according to a test set, and optimizing parameters of various algorithms by adopting a grid search mode to obtain optimized models; the principle of grid search is that all parameters or hyper-parameters of a model are arranged and combined, all the arrangement and combination form a two-dimensional grid, a plurality of hyper-parameters are combined pairwise to be regarded as a grid of a higher-dimensional space, all nodes in the grid are traversed, and the optimal solution in the grid is used as the final parameter of the model;
the plurality of machine learning models include polynomial regression (Lasso), Linear Regression (LR), Support Vector Regression (SVR), K-nearest neighbor algorithm (KNN), extreme gradient boosting tree (XGBoost), multi-level perceptron (MLP), decision tree (DecisionTree), Random Forest (RF), Adaboost, Gradient Boosting Decision Tree (GBDT), and Ridge regression (Ridge); obtaining a prediction result according to the optimized model by using the verification set, calculating the prediction precision of each model, and calculating the initial accuracy and the optimized accuracy of each model by taking the absolute error of 15mmHg as a prediction precision evaluation index to obtain the result shown in the table 1;
TABLE 1 prediction results of various machine learning models
Selecting n machine learning models with high prediction precision to respectively construct primary learners to obtain n primary learners; as can be seen from table 1, within ± 15mmHg of absolute error, the prediction accuracy of the three machine learning models, SVR, Lasso, and LR, is high, so the three models are selected to construct three primary learners in this embodiment;
respectively training n primary learners by using the n training subsets obtained in the step S1-3, and performing parameter optimization on the respective corresponding primary learners through the n test subsets to obtain optimized primary learners; predicting the general anesthesia induced contraction pressure according to the n optimized primary learners by using the n verification subsets;
step S2-2, constructing a secondary learner;
taking the predicted values of all the primary learners as the input of a secondary learner, and training the secondary learner to obtain a trained secondary learner;
in theory, a secondary learner integrating machine learning can use various models, and in order to prevent overfitting, the complexity of a prediction model needs to be limited, and a machine learning model with too high complexity is not suitable to be selected; in the embodiment, four common machine learning models are selected to predict the initial contraction and compression of general anesthesia, and the prediction precision of various algorithms is shown in table 2; selecting a model with the highest prediction precision as a secondary learner; as can be seen from table 2, within the absolute error ± 15mmHg, the Ridge regression has the highest accuracy, i.e., the highest prediction accuracy, so this embodiment uses the Ridge regression (Ridge) as the secondary learner;
TABLE 2 prediction accuracy of Secondary learner for different algorithms
Optimizing the input of the secondary learner; predicting the n test subsets through the n optimized primary learners to obtain a predicted value of the primary learner; taking the predicted value of the primary learner as the input of a secondary learner, wherein the predicted value of the secondary learner is the predicted value of the prediction model; giving a weight to each primary learner according to the prediction error of the primary learner, so as to optimize the input of the secondary learner and complete the construction of the secondary learner; in the weight assignment process, the larger the prediction error is, the lower the weight is;
let the input matrix of the secondary learner be
h1(x)、h2(x)、…、hn(x) Are respectively the predicted values, omega, of the respective primary learners1、ω2、…、ωnWeights of the predicted values of the primary learners, where1+ω2+…+ωn＝1；
The weight is adjusted by using a variation coefficient method, so that the precision of a prediction model is improved, and the prediction error is further reduced; the coefficient of variation of each primary learner satisfies equation (2):
in the formula (2), ViCoefficient of variation, σ, of the ith primary learneriIs the standard deviation of the predicted value and the true value of the ith primary learner, hi' (x) is the true value corresponding to the ith primary learner;
the weight of the ith primary learner predictor is:
in conclusion, the prediction model comprises n primary learners and one secondary learner, and is used for real-time prediction of the general anesthesia induced contraction pressure.
The invention also provides a general anesthesia created contraction compression prediction system (a short system, see fig. 3) based on machine learning fusion, which comprises a database module, a data acquisition module, a prediction module and an interaction module; the interaction module is used for manually inputting physical characteristic data of the patient and displaying a prediction result, wherein the physical characteristic data comprises the age, the sex, the height, the weight, the ASA grade and whether the patient suffers from hypertension; the database module is used for storing the physical characteristic data of the patient; the data acquisition module is used for acquiring vital sign data of a patient, the vital sign data are acquired by a vital monitoring instrument equipped in an operating room of a hospital and are transmitted to the data acquisition module, and the vital sign data comprise heart rate, invasive systolic compression and blood oxygen saturation (SaO 2); and a prediction program is stored in the prediction module, and the prediction program predicts the general anesthesia induced contraction compression according to the general anesthesia induced contraction compression prediction method based on machine learning fusion.
The prediction system periodically updates the prediction model in the use process, for example, when the number of predicted patients reaches a certain number, the system adds the new samples into the original training set, expands the original training set, and retrains the prediction model by using the expanded training set, so as to iteratively update the prediction model; iterative updating is a basic method for optimizing a prediction model, and the error between a predicted value and a true value is gradually reduced by continuously iterating the model, so that the optimal model parameter is solved; with the updating of the samples, the whole iteration process is continued all the time, the prediction model is continuously fitted with newly input sample data, the prediction error is reduced, and therefore the periodic updating of the prediction model is achieved.
Nothing in this specification is said to apply to the prior art.
Claims (6)
1. A machine learning fusion-based general anesthesia created contraction compression prediction method is characterized by comprising the following steps:
s1, collecting body characteristic data and vital sign data of a patient, preprocessing each data, and dividing the preprocessed data into a training set, a test set and a verification set; respectively carrying out slice sampling on the training set, the test set and the verification set to obtain n training subsets, test subsets and verification subsets;
s2, constructing a prediction model based on machine learning fusion, and predicting the general anesthesia induced contraction pressure by using the prediction model; the prediction model comprises n primary learners and one secondary learner, and the output values of all the primary learners are used as the input of the secondary learner.
2. The method for predicting machine learning fusion-based general anesthesia invasive contraction compression according to claim 1, wherein step S2 comprises:
s2-1, constructing a primary learner: training various machine learning models by using a training set, and optimizing the trained models according to a test set to obtain optimized models; predicting according to the optimized model by using the verification set, and selecting n machine learning models with high prediction precision to respectively construct primary learners to obtain n primary learners;
respectively training n primary learners by using the n training subsets obtained in the step S1, and performing parameter optimization on the respective corresponding primary learners through the n test subsets to obtain optimized primary learners; predicting according to the n optimized primary learners by using the n verification subsets;
s2-2, constructing a secondary learner: and taking the predicted values of all the primary learners as the input of the secondary learner, and training the secondary learner to obtain the trained secondary learner.
3. The machine learning fusion-based general anesthesia invasive contraction compression prediction method of claim 2, wherein the step S2-2 further comprises optimizing the input of the secondary learner; the method specifically comprises the following steps: giving a weight to each primary learner according to the prediction error of the primary learners, wherein the larger the prediction error is in the weight assignment process, the lower the weight is;
let the input matrix of the secondary learner be
h1(x)、h2(x)、…、hn(x) Are respectively the predicted values, omega, of the respective primary learners1、ω2、…、ωnWeights of the predicted values of the primary learners, where1+ω2+…+ωn＝1；
And (3) adjusting the weight by using a coefficient of variation method, wherein the coefficient of variation of each primary learner satisfies the formula (2):
in the formula (2), ViCoefficient of variation, σ, of the ith primary learneriIs the standard deviation of the predicted value and the true value of the ith primary learner, hi' (x) is the true value corresponding to the ith primary learner;
the weight of the ith primary learner predicted value satisfies equation (3).
4. The machine learning fusion-based general anesthesia invasive contraction compression prediction method according to claim 1, wherein the preprocessing comprises missing value filling, attribute identification, standardization processing and dimension reduction processing; the specific process of slice sampling is as follows: training the sample data set after dimensionality reduction by using a gradient lifting decision tree to obtain the probability of each feature, namely a feature importance list;
dividing the sample data set after dimensionality reduction into a training set, a test set and a verification set, and respectively copying the training set, the test set and the verification set for n times, wherein n is a positive integer; and respectively carrying out n times of slice sampling on the copied training set, test set and verification set according to the sample sampling proportion, and then carrying out feature selection on the feature importance list according to the feature sampling proportion to obtain n training subsets, test subsets and verification subsets.
5. The machine learning fusion-based general anesthesia invasive contraction compression prediction method according to claim 1, wherein the number of the primary learners is three, and the three primary learners are respectively constructed by a support vector regression model, a polynomial regression model and a linear regression model; the secondary learner is constructed using a ridge regression model.
6. A general anesthesia created contraction compression prediction system based on machine learning fusion is characterized by comprising a database module, a data acquisition module, a prediction module and an interaction module; the prediction module stores a prediction program, and the prediction program predicts the total hemp wound contraction pressure according to the method of any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111207748.5A CN113936801A (en) | 2021-10-18 | 2021-10-18 | Machine learning fusion-based general anesthesia induced contraction compression prediction method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111207748.5A CN113936801A (en) | 2021-10-18 | 2021-10-18 | Machine learning fusion-based general anesthesia induced contraction compression prediction method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113936801A true CN113936801A (en) | 2022-01-14 |
Family
ID=79280002
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111207748.5A Pending CN113936801A (en) | 2021-10-18 | 2021-10-18 | Machine learning fusion-based general anesthesia induced contraction compression prediction method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113936801A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115099374A (en) * | 2022-08-26 | 2022-09-23 | 江南大学 | Milling cutting force prediction system and method based on 5G multi-element mixed structure network |
| WO2023137530A1 (en) * | 2022-01-20 | 2023-07-27 | Annuo Medical Technology Solutions Pty Ltd | An artificially intelligent healthcare management system |
| CN116681291A (en) * | 2023-08-02 | 2023-09-01 | 杭州小策科技有限公司 | Wind control prediction method and system based on integrated model |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111145912A (en) * | 2019-12-23 | 2020-05-12 | 浙江大学 | Machine learning-based prediction device for personalized ovulation promotion scheme |
| CN112700325A (en) * | 2021-01-08 | 2021-04-23 | 北京工业大学 | Method for predicting online credit return customers based on Stacking ensemble learning |
| CN113096773A (en) * | 2021-03-31 | 2021-07-09 | 上海暖禾临床心理诊所有限公司 | System and method for collecting and processing cognitive information |
| CN113178261A (en) * | 2021-06-04 | 2021-07-27 | 福州大学 | Diabetes prediction model construction method and system based on machine learning |
| CN113436684A (en) * | 2021-07-02 | 2021-09-24 | 南昌大学 | Cancer classification and characteristic gene selection method |
-
2021
- 2021-10-18 CN CN202111207748.5A patent/CN113936801A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111145912A (en) * | 2019-12-23 | 2020-05-12 | 浙江大学 | Machine learning-based prediction device for personalized ovulation promotion scheme |
| CN112700325A (en) * | 2021-01-08 | 2021-04-23 | 北京工业大学 | Method for predicting online credit return customers based on Stacking ensemble learning |
| CN113096773A (en) * | 2021-03-31 | 2021-07-09 | 上海暖禾临床心理诊所有限公司 | System and method for collecting and processing cognitive information |
| CN113178261A (en) * | 2021-06-04 | 2021-07-27 | 福州大学 | Diabetes prediction model construction method and system based on machine learning |
| CN113436684A (en) * | 2021-07-02 | 2021-09-24 | 南昌大学 | Cancer classification and characteristic gene selection method |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023137530A1 (en) * | 2022-01-20 | 2023-07-27 | Annuo Medical Technology Solutions Pty Ltd | An artificially intelligent healthcare management system |
| CN115099374A (en) * | 2022-08-26 | 2022-09-23 | 江南大学 | Milling cutting force prediction system and method based on 5G multi-element mixed structure network |
| CN116681291A (en) * | 2023-08-02 | 2023-09-01 | 杭州小策科技有限公司 | Wind control prediction method and system based on integrated model |
| CN116681291B (en) * | 2023-08-02 | 2023-11-07 | 杭州小策科技有限公司 | Wind control prediction method and system based on integrated model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113936801A (en) | Machine learning fusion-based general anesthesia induced contraction compression prediction method and system | |
| WO2020010668A1 (en) | Human body health assessment method and system based on sleep big data | |
| Silva et al. | Rating organ failure via adverse events using data mining in the intensive care unit | |
| CN110111886A (en) | A kind of intelligent interrogation system and method based on XGBoost disease forecasting | |
| CN108511057A (en) | Transfusion volume model foundation and prediction technique, device, equipment and its storage medium | |
| CN110051324A (en) | A kind of acute respiratory distress syndrome anticipated mortality method and system | |
| US20210089965A1 (en) | Data Conversion/Symptom Scoring | |
| CN117854739A (en) | Intelligent internal medicine nursing monitoring system | |
| CN117936072B (en) | Vital sign data monitoring-based lung lobectomy PPCs risk assessment method | |
| Buldakova et al. | Hierarchy of human operator models for digital twin | |
| CN114927230B (en) | Prognosis decision support system and method for severe heart failure patient based on machine learning | |
| CN118280579B (en) | Sepsis patient condition assessment method and system based on multi-mode data fusion | |
| CN118035870B (en) | Method and system for optimizing pension service based on big data | |
| CN118280562A (en) | Knowledge graph driven medical large model diagnosis method | |
| CN117045889A (en) | End stage renal disease hemodialysis blood pressure mode identification and intervention system and related device | |
| CN109545374A (en) | A kind of disease based on body gait feature big data automatically tracks, treats and supplying system | |
| CN116019429B (en) | Health monitoring method, device, equipment and storage medium based on physiological index | |
| CN113990502A (en) | ICU heart failure prediction system based on heterogeneous graph neural network | |
| AU2021363110A1 (en) | Method and system for personalized prediction of infection and sepsis | |
| CN114864084A (en) | Health intervention path evaluation system, method and device and readable storage medium | |
| Vilas-Boas et al. | Hourly prediction of organ failure and outcome in intensive care based on data mining techniques | |
| CN116739037A (en) | Personality model construction method and device with personality characteristics | |
| CN113782221A (en) | Disease prediction device, equipment and storage medium based on self-training learning | |
| Vijayakumar et al. | An Intelligent stacking Ensemble-Based Machine Learning Model for Heart abnormality | |
| Alotaibi et al. | Predictive model for transferring stroke in-patients to intensive care unit |
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 |