CN113782210B - Method for predicting treatment failure probability of noninvasive ventilator - Google Patents
Method for predicting treatment failure probability of noninvasive ventilator Download PDFInfo
- Publication number
- CN113782210B CN113782210B CN202111072623.6A CN202111072623A CN113782210B CN 113782210 B CN113782210 B CN 113782210B CN 202111072623 A CN202111072623 A CN 202111072623A CN 113782210 B CN113782210 B CN 113782210B
- Authority
- CN
- China
- Prior art keywords
- model
- training
- training data
- time sequence
- predicting
- 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
- 238000000034 method Methods 0.000 title claims abstract description 32
- 206010066901 Treatment failure Diseases 0.000 title claims abstract description 22
- 238000012549 training Methods 0.000 claims abstract description 75
- 238000013145 classification model Methods 0.000 claims abstract description 22
- 239000008280 blood Substances 0.000 claims description 11
- 210000004369 blood Anatomy 0.000 claims description 11
- 238000012360 testing method Methods 0.000 claims description 11
- 238000012544 monitoring process Methods 0.000 claims description 9
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 claims description 8
- 229910052760 oxygen Inorganic materials 0.000 claims description 8
- 239000001301 oxygen Substances 0.000 claims description 8
- 201000010099 disease Diseases 0.000 claims description 7
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 claims description 7
- 230000000241 respiratory effect Effects 0.000 claims description 6
- 238000003066 decision tree Methods 0.000 claims description 5
- 206010020772 Hypertension Diseases 0.000 claims description 3
- 230000036760 body temperature Effects 0.000 claims description 3
- 208000026106 cerebrovascular disease Diseases 0.000 claims description 3
- 206010012601 diabetes mellitus Diseases 0.000 claims description 3
- 238000003745 diagnosis Methods 0.000 claims description 3
- 230000035487 diastolic blood pressure Effects 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 claims description 3
- 238000007477 logistic regression Methods 0.000 claims description 3
- 238000010801 machine learning Methods 0.000 claims description 3
- 238000003062 neural network model Methods 0.000 claims description 3
- 210000002569 neuron Anatomy 0.000 claims description 3
- 230000036387 respiratory rate Effects 0.000 claims description 3
- 230000035488 systolic blood pressure Effects 0.000 claims description 3
- 238000012706 support-vector machine Methods 0.000 claims 1
- 230000000694 effects Effects 0.000 abstract description 2
- 230000008569 process Effects 0.000 description 5
- CURLTUGMZLYLDI-UHFFFAOYSA-N Carbon dioxide Chemical compound O=C=O CURLTUGMZLYLDI-UHFFFAOYSA-N 0.000 description 4
- 238000002644 respiratory therapy Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 3
- 238000009825 accumulation Methods 0.000 description 2
- 230000004913 activation Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 229910002092 carbon dioxide Inorganic materials 0.000 description 2
- 239000001569 carbon dioxide Substances 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000003631 expected effect Effects 0.000 description 2
- 208000024891 symptom Diseases 0.000 description 2
- 208000004756 Respiratory Insufficiency Diseases 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 229940079593 drug Drugs 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000010837 poor prognosis Methods 0.000 description 1
- 201000004193 respiratory failure Diseases 0.000 description 1
- 208000023504 respiratory system disease Diseases 0.000 description 1
- 238000002627 tracheal intubation Methods 0.000 description 1
- 238000009423 ventilation Methods 0.000 description 1
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
- G16H50/50—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for simulation or modelling of medical disorders
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
- A61M16/00—Devices for influencing the respiratory system of patients by gas treatment, e.g. mouth-to-mouth respiration; Tracheal tubes
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
- A61M16/00—Devices for influencing the respiratory system of patients by gas treatment, e.g. mouth-to-mouth respiration; Tracheal tubes
- A61M16/0003—Accessories therefor, e.g. sensors, vibrators, negative pressure
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
- A61M2230/00—Measuring parameters of the user
- A61M2230/04—Heartbeat characteristics, e.g. ECG, blood pressure modulation
- A61M2230/06—Heartbeat rate only
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
- A61M2230/00—Measuring parameters of the user
- A61M2230/20—Blood composition characteristics
- A61M2230/202—Blood composition characteristics partial carbon oxide pressure, e.g. partial dioxide pressure (P-CO2)
- A61M2230/204—Blood composition characteristics partial carbon oxide pressure, e.g. partial dioxide pressure (P-CO2) partial carbon monoxide pressure (P-CO)
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
- A61M2230/00—Measuring parameters of the user
- A61M2230/20—Blood composition characteristics
- A61M2230/205—Blood composition characteristics partial oxygen pressure (P-O2)
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
- A61M2230/00—Measuring parameters of the user
- A61M2230/20—Blood composition characteristics
- A61M2230/208—Blood composition characteristics pH-value
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
- A61M2230/00—Measuring parameters of the user
- A61M2230/30—Blood pressure
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
- A61M2230/00—Measuring parameters of the user
- A61M2230/40—Respiratory characteristics
- A61M2230/42—Rate
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
- A61M2230/00—Measuring parameters of the user
- A61M2230/50—Temperature
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- Biomedical Technology (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Public Health (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Pulmonology (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Anesthesiology (AREA)
- Heart & Thoracic Surgery (AREA)
- Hematology (AREA)
- Animal Behavior & Ethology (AREA)
- Emergency Medicine (AREA)
- Veterinary Medicine (AREA)
- Biophysics (AREA)
- Databases & Information Systems (AREA)
- Pathology (AREA)
- Epidemiology (AREA)
- Primary Health Care (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The invention is applicable to the technical field of medical equipment, and relates to a method for predicting the treatment failure probability of a noninvasive ventilator, which comprises the following steps: s10, collecting training data of a user and recording a final treatment result of the user; s20, inputting training data into a time sequence model and a classification model and calibrating labels; s30, training the training data by using an LSTM model, and predicting future physiological parameters by using a time sequence model with relatively small MSE as a time sequence master model; s40, training future physiological parameters by using a LightGBM model, analyzing the final treatment effect of noninvasive breath according to the distribution of training samples, adjusting the parameters so as to determine an optimal classification model, and predicting the probability of treatment failure by using the optimal classification model; s50, training data of the user and training data weight in the time sequence master model are distributed, and the time sequence master model is retrained. The invention can predict future physiological parameters and predict the probability of treatment failure according to the future physiological parameters.
Description
Technical Field
The invention belongs to the technical field of medical equipment, and particularly relates to a method for predicting the treatment failure probability of a noninvasive ventilator.
Background
Noninvasive ventilators although the invention is relatively late with respect to invasive ventilators, the use of noninvasive ventilators for treatment of subjects with less symptoms of respiratory disease is a preferred method. However, as the symptoms of patients become worse gradually, how to switch to invasive treatment at proper time is always a key topic of clinical interest. At present, medical staff analyzes the concerned indexes according to the past experience based on the related monitoring indexes (such as PaCO2 indexes and the like) so as to judge the invasive treatment time. The prior art has some defects: the subjectivity is too strong, the factors of noninvasive treatment failure are many, the age, basic diseases, clinical monitoring indexes and the like of a patient can influence noninvasive treatment results, medical staff analyze according to past experience, the effect is difficult to judge due to the fact that a plurality of factors jointly influence, and the indexes of subjective attention of the medical staff are different, so that the judging results are possibly different; the accumulation time is long, and the method requires years of clinical experience accumulation; and the physiological parameters used are current, so that the physiological parameter change of the patient at the future moment cannot be obtained, and the failure probability of using the noninvasive ventilation of the patient at the future moment cannot be predicted.
Patent document with application number CN202010961481.8 discloses a method for determining expected effect by a respiratory support device and a respiratory support device, and the technical scheme adopted by the method comprises the following steps: s1, acquiring basic sign data of a user, wherein the basic sign data at least comprises basic characteristics, medical history characteristics and detection data; s2, comprehensively judging the basic sign data by using a judging network model, and determining the expected effect of the noninvasive treatment. This application also fails to address the problems associated with the prior art.
Therefore, how to provide a method capable of predicting the physiological parameters of a patient at a future time and predicting the failure probability of a noninvasive ventilator based on the future physiological parameters is a problem to be solved by those skilled in the art.
Disclosure of Invention
Aiming at the defects of the prior art, the invention aims to provide a method for predicting the treatment failure probability of a noninvasive ventilator, so as to solve the problems that the future physiological parameters cannot be predicted and the treatment failure probability cannot be predicted through the future physiological parameters in the prior art.
In order to solve the technical problems, the invention adopts the following technical scheme:
the invention provides a method for predicting the treatment failure probability of a noninvasive ventilator, which comprises the following steps:
s10, collecting training data of a user and recording a final treatment result of the user, wherein the training data comprise basic characteristics, illness states and monitoring data of the user;
s20, inputting the training data into a time sequence model and a classification model and calibrating labels;
s30, selecting an LSTM model to train the training data, taking a time sequence model with relatively small MSE as a time sequence master model, and predicting future physiological parameters through the time sequence master model;
s40, selecting a LightGBM model to train the future physiological parameters, analyzing a final noninvasive respiratory treatment result according to the distribution of training samples, adjusting the parameters so as to determine an optimal classification model, and predicting the probability of treatment failure through the optimal classification model;
s50, distributing the training data of the user and the training data weight in the time sequence master model, and retraining the time sequence master model.
Further, in the step S40, the training data is divided into a training set and a test set according to a predetermined ratio, the training set is used for training the LightGBM model, and the test set is used for test evaluation.
Further, the ratio of the training set to the test set is 4:1.
Further, in the step S50, the weight ratio of the training data of the user to the training data in the time-series master model is 9:1 or 4:1.
Further, in step S30, the machine learning training framework adopted for the LSTM model training is a TensorFlow, and the relevant parameters include the number of training lots, the learning rate, the number of neurons, the number of output results, and the number of iterations.
Further, in the step S40, the model algorithm includes a logistic regression model, a support vector classification model, a neural network model, and a decision tree model.
Further, in the step S40, the parameters related to the LightGBM model include learning rate, tree depth, number of leaf nodes, and number of tool boxes.
Further, in the step S10, the basic characteristics include gender and age, the disease condition includes diagnosis of disease, whether or not hypertension, diabetes and cerebrovascular disease have occurred, and the monitoring data includes body temperature, consciousness index GCS, diastolic blood pressure, systolic blood pressure, respiratory rate, heart rate, CO 2 Partial pressure, O 2 Partial pressure, blood pH, blood oxygen concentration, arterial blood carbon dioxide partial pressure PaCO 2 Partial pressure of arterial blood oxygen PaO 2 Oxygen concentration fraction FiO in inhaled air 2 。
Further, in the step S10, the final treatment result includes non-invasive treatment failure, non-invasive treatment success and indistinguishable.
Further, in the step S10, the basic features of the user described in the foregoing need to be numerically converted.
Compared with the prior art, the method for predicting the treatment failure probability of the noninvasive ventilator has at least the following beneficial effects:
the invention can predict the future physiological parameters of the patient through the time sequence model, and predict the future physical condition of the patient; the probability of non-invasive respiratory therapy failure can be predicted on the basis of future physiological parameters through a classification model; and retraining the time series model by assigning data weights of the patient data being used and the time series mother model to improve the accuracy of predicting future physiological parameters.
Drawings
In order to more clearly illustrate the solution of the invention, a brief description will be given below of the drawings required for the description of the embodiments, it being apparent that the drawings in the following description are some embodiments of the invention and that other drawings may be obtained from these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method for predicting the failure probability of a noninvasive ventilator treatment according to one embodiment of the present invention;
FIG. 2 is a schematic diagram of a conventional RNN model;
fig. 3 is a schematic diagram of a first unit structure of an LSTM model of a method for predicting a failure probability of treatment of a noninvasive ventilator according to an embodiment of the present invention.
Detailed Description
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs; the terms used in the specification are used herein for the purpose of describing particular embodiments only and are not intended to limit the present invention, for example, the orientations or positions indicated by the terms "length", "width", "upper", "lower", "left", "right", "front", "rear", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. are orientations or positions based on the drawings, which are merely for convenience of description and are not to be construed as limiting the present invention.
The terms "comprising" and "having" and any variations thereof in the description of the invention and the claims and the description of the drawings above are intended to cover a non-exclusive inclusion; the terms first, second and the like in the description and in the claims or in the above-described figures, are used for distinguishing between different objects and not necessarily for describing a sequential or chronological order. In the description of the invention and the claims and the above figures, when an element is referred to as being "fixed" or "mounted" or "disposed" or "connected" to another element, it can be directly or indirectly on the other element. For example, when an element is referred to as being "connected to" another element, it can be directly or indirectly connected to the other element.
Furthermore, references herein to "an embodiment" mean that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the present invention. The appearances of such phrases in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. Those of skill in the art will explicitly and implicitly appreciate that the embodiments described herein may be combined with other embodiments.
The invention provides a method for predicting the treatment failure probability of a noninvasive ventilator, which is applied to the noninvasive ventilator, and comprises the following steps: s10, collecting training data of a user and recording the final treatment result of the user, wherein the training data comprise basic characteristics, illness states and monitoring data of the user; s20, inputting training data into a time sequence model and a classification model and calibrating labels; s30, selecting an LSTM model to train training data, taking a time sequence model with relatively small MSE as a time sequence master model, and predicting future physiological parameters through the time sequence master model; s40, selecting a LightGBM model to train future physiological parameters, analyzing a final noninvasive respiratory treatment result according to the distribution of training samples, adjusting the parameters to determine an optimal classification model, and predicting the probability of treatment failure through the optimal classification model; s50, training data of the user and training data weight in the time sequence master model are distributed, and the time sequence master model is retrained.
The invention can predict the future physiological parameters of the patient through the time sequence model, and predict the future physical condition of the patient; the probability of non-invasive respiratory therapy failure can be predicted on the basis of future physiological parameters through a classification model; and retraining the time series model by assigning data weights of the patient data being used and the time series mother model to improve the accuracy of predicting future physiological parameters.
In order to make the person skilled in the art better understand the solution of the present invention, the technical solution of the embodiment of the present invention will be clearly and completely described below with reference to the accompanying drawings.
The invention provides a method for predicting the treatment failure probability of a noninvasive ventilator, which is applied to the noninvasive ventilator, as shown in figure 1, and comprises the following steps:
s10, collecting training data of a user and recording the final treatment result of the user, wherein the training data comprise basic characteristics, illness states and monitoring data of the user;
specifically, training data of the user may be collected clinically including, but not limited to: basic characteristics include gender, age, etc.; disease conditions include diagnosis of the disease, whether or not it has been suffering from hypertension, diabetes mellitus, cerebrovascular disease, etc.; the monitoring data include body temperature, consciousness index GCS, diastolic pressure, systolic pressure, respiratory rate, heart rate and CO 2 Partial pressure, O 2 Partial pressure, blood pH, blood oxygen concentration, arterial blood carbon dioxide partial pressure PaCO 2 Partial pressure of arterial blood oxygen PaO 2 Oxygen concentration fraction FiO in inhaled air 2 And the like, the training data are collected at intervals of 1 hour (optimally 0.5 to 2 hours); the final treatment results of the user comprise non-invasive treatment failure (the conditions of direct death of the patient, intubation after the patient is deteriorated, discharge of the patient but poor prognosis, and the like), non-invasive treatment success (the conditions of smooth discharge of the patient, and the like), incapacity of judgment (the condition of forced discharge of the patient but incapacity of medical staff to know the follow-up condition, which is a very small number and can be deleted in the follow-up analysis), recording the final treatment results of the patient, and carrying out numerical conversion according to the final treatment results, wherein the non-invasive treatment failure can be 0, the non-invasive treatment success can be 1 and the incapacity of judgment can be 2;
s20, inputting training data into a time sequence model and a classification model and calibrating labels;
specifically, the Chinese description features need to be labeled, such as gender (1 for male and 0 for female), the time sequence model takes every 12 groups of parameters as a time sequence training unit, the next group of parameters of each time sequence training unit is taken as a time sequence training label, and so on, so that one patient can obtain a plurality of groups of time sequence training units; the classification model takes the parameter at the last moment as a classification training unit, and the final treatment result is a classification training label;
s30, selecting an LSTM model to train training data, taking a time sequence model with relatively small MSE as a time sequence master model, and predicting future physiological parameters through the time sequence master model;
specifically, as shown in FIGS. 2 and 3, an LSTM model (or related optimization model such as GRU, etc.) is selected, which is modified from the RNN model, where h N ＝σ(W h *h N-1 +W I *x N ) W represents the parameters of the required training, sigma represents the sigmoid activation function, y N ＝σ(W o *h N ) The following can be obtained by combining fig. 2, 3 and the above formula: h is a N Can be defined by h N-1 、x N The premise of the obtained is that h is considered as N And h N-1 、x N There is a correlation that continues throughout the transmission process when a parameter is consistently greater than 1 or less than 1, such that as the transmission process continues to accumulate, it is easy for the parameter value to change exponentially or consistently toward 0 (i.e., gradient explosion or disappearance), thus selecting the LSTM model to model h in the RNN model N Is divided into h N And c N Wherein c N Is a variable which changes faster, h N Is a variable with slower variation, and the specific formula is c N ＝z f ×c N-1 +z i ×z，h N ＝z o ×tanh(c N-1 )，y N ＝σ(W*h N ) Where "x" denotes the multiplication of the element matrices, "+" element matrices are added, and z is all defined by h N-1 Multiplied by x, provided only with different activation functions and parameters, where z f 、z i 、z o Respectively representing a forgetting door, an input door and an output door, respectively having different functions, and not being specifically unfolded here; the LSTM model uses the mean square error as an evaluation criterion,wherein y is i Is the actual value +.>For predictive value, the related parameters such as training batch number, learning rate, neuron number, output result number and iteration number can be adjusted by different modes by using a common machine learning training framework (such as TensorFlow) to obtain a time sequence model with relatively small MSEThis serves as a master model for predicting future physiological parameters;
s40, selecting a LightGBM model to train future physiological parameters, analyzing a final noninvasive respiratory treatment result according to the distribution of training samples, adjusting the parameters to determine an optimal classification model, and predicting the probability of treatment failure through the optimal classification model;
specifically, in this embodiment, the LightGBM model is preferably used for training data, the core algorithm of the LightGBM model is developed from a decision tree, and compared with other decision tree (such as GBDT) algorithms, the LightGBM model has the characteristics of low memory, high operation efficiency, higher accuracy, parallelization operation and the like, future physiological parameters are divided into a training set and a testing set according to the ratio of 4:1, the training set is used for training the LightGBM model, and the testing set is used for testing and evaluating; the model algorithm comprises a logistic regression model, a support vector classification model, a neural network model and a decision tree model, and the relevant parameters of the LightGBM model comprise a learning rate, a tree depth, the number of leaf nodes and the number of tool boxes, and in the embodiment, key parameters can be initialized: the learning rate can be set to 0.1, the tree depth can be set to 4, the number of leaf nodes is set to 8, the number of tool boxes is set to 50, other parameters can be modified and perfected according to actual conditions, the success and failure proportion of noninvasive respiratory therapy is analyzed according to the distribution of training samples, and an index can be selected as a measurement standard adjustment parameter, such as recall rate (the proportion of correctly predicted samples in failure) or accuracy rate (the proportion of correctly predicted samples in all samples), so that an optimal model is determined;
s50, distributing training data of a user and training data weights in the time sequence master model, and retraining the time sequence master model;
specifically, based on the time series mother model, a targeted retraining can be performed on the patient in use, a database is required to be built when the model retrains, data and labels thereof are sequentially stored according to a mode of collecting training data in the time series model, in the retraining process, the MSE is calculated differently from the initial training process, the weight ratio of the patient data in use to the data in the mother model can be given to 9:1 or 4:1 in the retraining process, namely, the MSE of the patient data in use is calculated and multiplied by 0.9 or 0.8, and the MSE of the data in the mother model is calculated and multiplied by 0.1 or 0.2, and other processes are consistent with the training process.
The method for predicting the treatment failure probability of the noninvasive ventilator according to the embodiment of the present invention can predict future physiological parameters of a patient through a time series model, and predict future physical conditions of the patient; the probability of non-invasive respiratory therapy failure can be predicted on the basis of future physiological parameters through a classification model; retraining the time sequence model by distributing the data weight of the patient data and the time sequence mother model in use so as to improve the accuracy of predicting future physiological parameters; in addition, while predicting the probability of non-invasive respiratory failure, the intermediate margin predicts future physiological parameters for other purposes, such as predicting whether a related drug is needed when the blood PH is too low or too high.
It is apparent that the above-described embodiments are merely preferred embodiments of the present invention, not all of which are shown in the drawings, which do not limit the scope of the invention. This invention may be embodied in many different forms, but rather, embodiments are provided in order to provide a thorough and complete understanding of the present disclosure. Although the invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that modifications may be made to the embodiments described in the foregoing description, or equivalents may be substituted for elements thereof. All equivalent structures made by the content of the specification and the drawings of the invention are directly or indirectly applied to other related technical fields, and are also within the scope of the invention.
Claims (10)
1. A method of predicting the probability of failure of a noninvasive ventilator treatment comprising the steps of:
s10, collecting training data of a user and recording a final treatment result of the user, wherein the training data comprise basic characteristics, illness states and monitoring data of the user;
s20, inputting the training data into a time sequence model and a classification model and calibrating labels;
s30, selecting an LSTM model to train the training data, taking a time sequence model with relatively small MSE as a time sequence master model, and predicting future physiological parameters through the time sequence master model;
s40, selecting a LightGBM model to train the future physiological parameters, analyzing a final noninvasive respiratory treatment result according to the distribution of training samples, adjusting the parameters so as to determine an optimal classification model, and predicting the probability of treatment failure through the optimal classification model;
s50, distributing the training data of the user and the training data weight in the time sequence master model, and retraining the time sequence master model.
2. The method according to claim 1, wherein in step S40, the training data is divided into a training set and a test set according to a predetermined ratio, the training set is used for training the LightGBM model, and the test set is used for test evaluation.
3. A method of predicting the failure probability of a noninvasive ventilator treatment in accordance with claim 2, wherein the training set to test set ratio is 4:1.
4. A method according to claim 3, wherein in step S50, the weight ratio of the training data of the user to the training data in the time series parent model is 9:1 or 4:1.
5. The method according to claim 1, wherein in the step S30, the machine learning training framework used for the LSTM model training is a TensorFlow, and the relevant parameters include the number of training lots, the learning rate, the number of neurons, the number of output results, and the number of iterations.
6. The method according to claim 5, wherein in the step S40, the model algorithm includes a logistic regression model, a support vector machine classification model, a neural network model, and a decision tree model.
7. The method according to claim 6, wherein the parameters related to the LightGBM model in step S40 include learning rate, tree depth, number of leaf nodes, and number of tool boxes.
8. The method according to claim 1, wherein in the step S10, the basic characteristics include sex and age, the disease condition includes diagnosis of disease, whether or not the patient has been suffering from hypertension, diabetes and cerebrovascular disease, and the monitoring data includes body temperature, consciousness index GCS, diastolic blood pressure, systolic blood pressure, respiratory rate, heart rate, CO 2 Partial pressure, O 2 Partial pressure, blood pH, blood oxygen concentration, oxygen concentration fraction FiO in inhaled air 2 。
9. The method according to claim 8, wherein in step S10, the final treatment result includes non-invasive treatment failure, non-invasive treatment success and non-determinable.
10. The method according to claim 9, wherein in the step S10, the basic characteristics of the user described in the middle need to be numerically converted.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111072623.6A CN113782210B (en) | 2021-09-14 | 2021-09-14 | Method for predicting treatment failure probability of noninvasive ventilator |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111072623.6A CN113782210B (en) | 2021-09-14 | 2021-09-14 | Method for predicting treatment failure probability of noninvasive ventilator |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113782210A CN113782210A (en) | 2021-12-10 |
| CN113782210B true CN113782210B (en) | 2024-04-12 |
Family
ID=78843259
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111072623.6A Active CN113782210B (en) | 2021-09-14 | 2021-09-14 | Method for predicting treatment failure probability of noninvasive ventilator |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113782210B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112274741A (en) * | 2020-09-14 | 2021-01-29 | 湖南明康中锦医疗科技发展有限公司 | Method for judging expected effect by respiratory support equipment and respiratory support equipment |
| CN112675393A (en) * | 2020-12-18 | 2021-04-20 | 杭州脉兴医疗科技有限公司 | Ventilator removing management system and method |
| CN113257425A (en) * | 2021-06-30 | 2021-08-13 | 杭州华网信息技术有限公司 | Influenza prediction system, storage medium and device for optimizing LSTM and LightGBM parameters |
| EP3874520A1 (en) * | 2018-11-01 | 2021-09-08 | Imperial College of Science, Technology and Medicine | Predicting physiological parameters |
-
2021
- 2021-09-14 CN CN202111072623.6A patent/CN113782210B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3874520A1 (en) * | 2018-11-01 | 2021-09-08 | Imperial College of Science, Technology and Medicine | Predicting physiological parameters |
| CN112274741A (en) * | 2020-09-14 | 2021-01-29 | 湖南明康中锦医疗科技发展有限公司 | Method for judging expected effect by respiratory support equipment and respiratory support equipment |
| CN112675393A (en) * | 2020-12-18 | 2021-04-20 | 杭州脉兴医疗科技有限公司 | Ventilator removing management system and method |
| CN113257425A (en) * | 2021-06-30 | 2021-08-13 | 杭州华网信息技术有限公司 | Influenza prediction system, storage medium and device for optimizing LSTM and LightGBM parameters |
Non-Patent Citations (3)
| Title |
|---|
| Clinical Intervention Prediction and Understanding with Deep Neural Networks;Harini Suresh 等;Proceedings of Machine Learning for Healthcare;全文 * |
| Dynamic prediction of late noninvasive ventilation failure in intensive care unit using a time adaptive machine model;Xue Feng;Computer Methods and Programs in Biomedicine;第208卷;https://doi.org/10.1016/j.cmpb.2021.106290 * |
| Predicting respiratory failure in patients with COVID-19 pneumonia: a case study from Northern Italy;Davide Ferrari 等;Advances in Artificial Intelligence for Healthcare;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113782210A (en) | 2021-12-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111261282A (en) | Sepsis early prediction method based on machine learning | |
| Silva et al. | Rating organ failure via adverse events using data mining in the intensive care unit | |
| CN109659033A (en) | A kind of chronic disease change of illness state event prediction device based on Recognition with Recurrent Neural Network | |
| Mao et al. | A review of recurrent neural network-based methods in computational physiology | |
| CN107249434B (en) | Robust classifier | |
| CN109065171A (en) | The construction method and system of Kawasaki disease risk evaluation model based on integrated study | |
| CN109119156A (en) | A kind of medical diagnosis system based on BP neural network | |
| CN113096818B (en) | Method for evaluating occurrence probability of acute diseases based on ODE and GRUD | |
| CN111951975A (en) | Sepsis early warning method based on deep learning model GPT-2 | |
| CN114512239B (en) | Cerebral apoplexy risk prediction method and system based on transfer learning | |
| Liang et al. | A convolutional neural network combined with prototype learning framework for brain functional network classification of autism spectrum disorder | |
| Banerjee et al. | Deep neural network based missing data prediction of electrocardiogram signal using multiagent reinforcement learning | |
| Aggarwal et al. | A structured learning approach with neural conditional random fields for sleep staging | |
| CN112542242A (en) | Data transformation/symptom scoring | |
| CN110400610B (en) | Small sample clinical data classification method and system based on multichannel random forest | |
| Ji et al. | Functional brain network classification based on deep graph hashing learning | |
| CN111276242A (en) | Disease diagnosis and disease state evaluation modeling method for patients in intensive care unit of hospital | |
| Zhou et al. | Risk factor refinement and ensemble deep learning methods on prediction of heart failure using real healthcare records | |
| KR102421172B1 (en) | Smart Healthcare Monitoring System and Method for Heart Disease Prediction Based On Ensemble Deep Learning and Feature Fusion | |
| Al Rasyid et al. | Anomalous data detection in WBAN measurements | |
| CN113782210B (en) | Method for predicting treatment failure probability of noninvasive ventilator | |
| CN110598789A (en) | Human fatigue state prediction method and system based on fuzzy perceptron | |
| TWI755731B (en) | Biomedical data risk prediction system and method | |
| US20220262516A1 (en) | Atrial Fibrillation Prediction Model And Prediction System Thereof | |
| CN112686091A (en) | Two-step arrhythmia classification method based on deep neural network |
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 | ||
| CB03 | Change of inventor or designer information |
Inventor after: Dai Zheng Inventor after: Luo Huiyu Inventor after: Duan Jun Inventor before: Dai Zheng Inventor before: Luo Huiyu |
|
| CB03 | Change of inventor or designer information | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |