CN108325020B - Vein anesthesia multi-parameter index closed-loop monitoring system - Google Patents
Vein anesthesia multi-parameter index closed-loop monitoring system Download PDFInfo
- Publication number
- CN108325020B CN108325020B CN201810193928.4A CN201810193928A CN108325020B CN 108325020 B CN108325020 B CN 108325020B CN 201810193928 A CN201810193928 A CN 201810193928A CN 108325020 B CN108325020 B CN 108325020B
- Authority
- CN
- China
- Prior art keywords
- parameter
- module
- control host
- index
- monitoring
- 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
-
- 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
- A61M5/00—Devices for bringing media into the body in a subcutaneous, intra-vascular or intramuscular way; Accessories therefor, e.g. filling or cleaning devices, arm-rests
- A61M5/14—Infusion devices, e.g. infusing by gravity; Blood infusion; Accessories therefor
- A61M5/142—Pressure infusion, e.g. using pumps
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/48—Other medical applications
- A61B5/4821—Determining level or depth of anaesthesia
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
-
- 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
- A61M5/00—Devices for bringing media into the body in a subcutaneous, intra-vascular or intramuscular way; Accessories therefor, e.g. filling or cleaning devices, arm-rests
- A61M5/14—Infusion devices, e.g. infusing by gravity; Blood infusion; Accessories therefor
- A61M5/168—Means for controlling media flow to the body or for metering media to the body, e.g. drip meters, counters ; Monitoring media flow to the body
-
- 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
- A61M5/00—Devices for bringing media into the body in a subcutaneous, intra-vascular or intramuscular way; Accessories therefor, e.g. filling or cleaning devices, arm-rests
- A61M5/14—Infusion devices, e.g. infusing by gravity; Blood infusion; Accessories therefor
- A61M5/142—Pressure infusion, e.g. using pumps
- A61M2005/14208—Pressure infusion, e.g. using pumps with a programmable infusion control system, characterised by the infusion program
-
- 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
- A61M5/00—Devices for bringing media into the body in a subcutaneous, intra-vascular or intramuscular way; Accessories therefor, e.g. filling or cleaning devices, arm-rests
- A61M5/14—Infusion devices, e.g. infusing by gravity; Blood infusion; Accessories therefor
- A61M5/142—Pressure infusion, e.g. using pumps
- A61M2005/14288—Infusion or injection simulation
- A61M2005/14292—Computer-based infusion planning or simulation of spatio-temporal infusate distribution
-
- 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
- A61M2230/08—Other bio-electrical signals
Abstract
The invention discloses a vein anesthesia multi-parameter index closed-loop monitoring system, which comprises a parameter signal acquisition module, a control host module and an injection device module, wherein the control host module is connected with the parameter signal acquisition module; the parameter signal acquisition module is connected with the control host module and is used for acquiring human body original EEG signals, NIRS signals and EMG signals, converting the signals into parameter indexes capable of monitoring the sedation depth, the analgesia depth and the muscle relaxation degree and transmitting the parameter indexes to the control host module; the control host module is used for parameter identification, model prediction and controller feedback calculation to obtain the dose and injection rate required by the human body in real time and provide monitoring indexes in real time; the control end of the control host module is connected with the injection device module, and the injection device module obtains the dose and the injection speed of the medicament through the control host module and then injects the medicament into a human body. The invention adopts a closed loop system of multi-parameter index monitoring, solves the problem of uncertainty of the traditional parameters, and has the advantages of strong real-time performance, high precision and the like.
Description
Technical Field
The invention relates to the field of anesthesia control, in particular to a multi-parameter index closed-loop monitoring system for intravenous anesthesia.
Background
According to statistics, about 9 million general anesthesia operations are performed every year in China, and the operation amount is expected to reach 1.55 hundred million operations every year in 2018. Clinically important procedures, such as cardiac surgery, neurosurgery and monitoring of patients with brain trauma, require careful administration of anesthesia to ensure intra-and post-operative safety. These have put higher demands on the individual and fine control of the administration of anesthesia.
It is well known that three indicators can be used for intraoperative anesthesia depth monitoring, namely sedation, analgesia and muscle relaxation. For clinical anesthesia, monitoring research on sedation is the most, and there are many ways to estimate the depth of sedation based on the generation of a single index by the brain electrical wave, such as bispectral index, Nacrotrend index, entropy index, wavelet index, etc. In analgesia, it is difficult to extract a single level of analgesia from EEG because the analgesic drugs act on the brain differently from the sedative drugs. Monitoring of analgesia is heart rate variability, mean arterial blood pressure, etc. The monitoring of the muscle relaxation is relatively simple, the muscle relaxation degree in the operation can be effectively monitored through the myoelectric component, and no drug interaction exists between the muscle relaxation agent and the sedative and analgesic. At the same time, there is a synergistic effect between sedative drugs (such as propofol) and analgesic drugs (such as remifentanil), which can lead to errors in the estimation of the fixed PKPD model.
Although much research has been done on anesthesia monitoring and closed-loop control, most systems currently available only consider a single indicator of sedation or analgesia, and do not consider synergy between analgesic and sedation drugs. Therefore, no anesthesia monitoring system can simultaneously provide multiple parameter indexes of sedation, analgesia and the like in anesthesia at present.
Disclosure of Invention
The invention aims to provide a vein anesthesia multi-parameter index closed-loop monitoring system which can simultaneously acquire EEG and NIRS signals, extract related indexes of sedation and analgesia from the EEG and the NIRS, and respectively realize corresponding closed-loop control on parameters of analgesia, sedation and the like by mainly using propofol-remifentanil combined anesthesia and considering the synergistic effect of combined medication.
In order to realize the purpose, the following technical scheme is adopted: the system comprises a parameter signal acquisition module, a control host module and an injection device module; the parameter signal acquisition module is connected with the control host module and is used for acquiring human body original EEG signals, NIRS signals and EMG signals, converting the signals into parameter indexes capable of monitoring the sedation depth, the analgesia depth and the muscle relaxation degree and transmitting the parameter indexes to the control host module; the control host module is used for parameter identification, model prediction and controller feedback calculation to obtain the dose and injection rate required by the human body in real time and provide monitoring indexes in real time; the control end of the control host module is connected with the injection device module, and the injection device module acquires the dose and the injection speed of the medicament through the control host module and then injects the medicament into a human body;
the parameter signal acquisition module consists of an EEG-NIRS signal acquisition system and is provided with an EEG signal module, an NIRS signal module and an electromyographic signal module; the EEG signal module comprises the acquisition of original signals and the extraction of sedation indexes; the NIRS signal module comprises the steps of collecting an original NIRS signal and extracting an analgesic index; the electromyographic signal module comprises the steps of collecting original signals and extracting muscle relaxation indexes;
the control host module comprises a monitoring interface, a drug-substituted model, parameter identification, a controller, index optimization and a database; the monitoring interface has a manual input function and a parameter index monitoring function; the drug-substituted model is a patient model considering the synergistic effect of the mixed use of the drugs; the parameter identification is model parameter identification based on a least square method; the controller is a model prediction control and a closed-loop controller based on an extended prediction adaptive algorithm; the index optimization is parameter optimization through feature extraction and unsupervised deep neural network learning; the database establishes patient data information for storage;
the injection device module consists of sedation, analgesia and muscle relaxation drug injectors which are designed in parallel, and transmits three drug injection speed information of sedation, analgesia and muscle relaxation calculated by the control host module to the injectors for drug injection of patients by establishing an RS232 serial port communication protocol with the control host module;
the sedation depth parameter LIA is obtained by integrating, extracting features and learning a deep neural network of feature components of four indexes of PLZC, SFS, MPF and BetaRatio;
wherein, the calculation formula of PLZC is as follows:
the SFS is calculated by the formula: SFS ═ log (P)0.5-47HZ/P40-47HZ) (ii) a The calculation formula of MPF is:
the formula for BetaRatio is: BetaRatio ═ log (P)30-47Hz/P11-20Hz)。
The LIA is obtained by SVM or random forest training on the basis of the parameters.
Further, the analgesia depth parameter NAF is obtained by carrying out multi-scale wavelet decomposition and Bell-Lambert rule transformation on the original NIRS signal and calculating the area under a characteristic curve a-LL; the calculated index NAF can accurately evaluate pain/analgesia reaction in the operation;
NAF＝[(α*LLmin)2+α*β]1/2/12.8 (1)
wherein alpha and beta are constant variables of the analgesia depth parameter NAF, and the value is a constant between 0 and 100; LLmin is the minimum of the area of the characteristic curve A-LL under the adjacent four sub-windows.
Furthermore, the muscle relaxation parameter RPSD is used for judging the muscle relaxation degree of the patient according to the proportion of the myoelectric component to the total energy, namely after the frequency band (more than or equal to 40Hz) where the myoelectric signal is located is extracted, the muscle relaxation degree of the patient is evaluated by calculating the relative power spectral density of the myoelectric component.
Further, the patient model considering the synergistic effect of the drug mixture, aiming at the synergistic effect generated by the combination of propofol and remifentanil, on the basis of the traditional PKPD model, the concentration-response relationship of the two drugs can be represented by the following normalized relationship:
wherein DOA (t) is the anesthetic sedative effect; t is a time(s) variable; theta is a drug concentration effect parameter; e0Is an initial constant (0-100))；Emax(θ) is the maximum possible drug effect at θ; u shapeprop(t) is the effect of propofol concentration; u shapeRem(t) concentration effect for remifentanil; u shapeprop(t)+URem(t) is the mixed drug effect; u shape50(θ) is the maximum effect constant of 50% at θ; γ (θ) is the maximum possible drug effect of the concentration-response relationship at θ.
Further, the parameter identification, namely the patient model identification module, adopts a least square method for identification, and the identified patient models are as follows:
where u represents the input to the model, i.e., the infusion rates for propofol and remifentanil; k is a variable in the infusion speed u matrix and is an integer; y represents an index of anesthesia monitoring, i.e., sedation or analgesia; a is4，a3，a2，a1，a0，b1，b2，b3Identified model parameters.
Further, the index optimization is to adopt a multi-view learning method to carry out secondary optimization and confirmation on index characteristics; the method comprises the following steps:
(1) extracting different characteristics of anesthesia and establishing a characteristic pool of all the characteristics; regarding the characteristics of different age groups as a plurality of visual angles;
(2) adopting the initial distribution probability of the index feature label obtained by the first training of the multilayer neural network as the state feature of anesthesia, and combining the age feature of the patient to retrain the multilayer neural network;
(3) and finding out the most suitable characteristic representation generation state in each state by establishing the association relationship between different characteristics and states.
Compared with the prior art, the invention has the following advantages:
1. realize the anesthesia monitoring of multi-parameter index, provide the state monitoring basis for meticulous control anesthesia.
2. Features of four indexes of PLZC, MPF, SFS and BetaRatio in frequency domain and time domain are extracted and optimized through EEG signals collected by a self-made EEG-NIRS system, a system independent sedation parameter LIA is established, and uncertainty of the existing indexes is solved.
3. According to a self-made EEG-NIRS system, original near infrared signals are subjected to multi-scale decomposition and characteristic extraction to obtain an analgesia parameter NAF (NIRS-Area-Feature) independent of the system, and the obtained analgesia parameter can be used for more accurately evaluating pain/analgesia stimulation degree through clinical experiments.
4. Considering the synergistic effect between the sedative and analgesic drugs, a mixed drug patient model is established, which can make quick response to unexpected surgical stimulation; meanwhile, the anesthesia control system combining manual input and automatic control greatly reduces the workload of an anesthesiologist, improves the operation safety, and enables the anesthesiologist to put more energy into more important operation decisions.
Drawings
FIG. 1 is a block diagram of the system of the present invention.
FIG. 2 is a diagram of the closed loop control of the system of the present invention.
FIG. 3 is a schematic diagram of the multi-parameter anesthesia state optimization of the system of the present invention.
Detailed Description
The invention is further described below with reference to the accompanying drawings:
as shown in figure 1, the system comprises a patient (11), a parameter signal acquisition module (12), a control host module (13) and an injection device module (14);
the patient is connected with a parameter signal acquisition module, the parameter signal acquisition module is composed of an EEG-NIRS signal acquisition system, can synchronously acquire NIRS (121) signals, EEG signals (122) and myoelectric signals (123), extracts related indexes of sedation (141), analgesia (142) and muscle relaxation (143), and transmits parameter index information to a control host module. The control host module comprises a monitoring interface (131) with manual input function and parameter index monitoring, a patient model (134) considering synergistic effect of mixed drug use, model parameter identification (132) based on a least square method, a closed-loop controller (133) based on model prediction control and an extended prediction adaptive algorithm, index optimization (135) through feature extraction and unsupervised deep neural network learning, and a database (136) for establishing patient data information. The control host module is connected with the injection device module, and transmits the types of the medicines required by the patient, the injection medicine amount and the injection speed information calculated by the control host module to the injection device module by establishing an RS232 serial port communication protocol. The injection device module is composed of sedation, analgesia and muscle relaxation drug injectors which are designed in parallel, and drug infusion is carried out on a patient, so that automatic anesthesia closed-loop control is realized.
The multi-index parameters comprise sedation depth, analgesia depth, muscle relaxation degree and brain oxygen metabolism information.
The sedation depth parameter is obtained by obtaining four indexes of permutation Lempel-Ziv complexity (PLZC), synchronous fast-slow ratio (SFS), intermediate frequency spectrum (MPF) and beta ratio (BetaRatio) in time domain and frequency domain through multi-scale decomposition and feature extraction from collected EEG signals, and the sedation depth parameter index LIA can be accurately judged after integration, feature extraction and deep neural network learning.
The specific algorithm of PLZC is described as follows:
(1) for the marker sequence of N points { x (i) < 1 ≦ i ≦ N } totaling m! And, reconstructing it. The Lempel-Ziv complexity is then calculated from the reconstructed marker sequence. The total number of subsequences of a marker sequence (x (n)) has an upper limit, and is denoted as L (n):
L(n)＝c(n)[logαc(n)+1] (1)
where is the number of markers in the current marker sequence, equal to m! .
(2) The PLZC can be defined as a standardized c (n):
(3) where n represents the total length of the marker sequence. When N is large, the PLZC can be simplified to:
the specific algorithm for the beta ratio is as follows:
BetaRatio＝log(P30-47Hz/P11-20Hz) (4)
the MPF specific algorithm is as follows:
the specific algorithm for SFS is as follows:
SFS＝log(P0.5-47HZ/P40-47HZ) (7)
by inputting the characteristic components of the four signals into the characteristic matrix and by means of classification methods such as SVM and random forest, a characteristic curve which accurately accords with the sedation state of a patient in operation can be obtained, the index LIA of the system for monitoring the sedation depth independently is obtained through optimization, and the problems of inaccuracy of a single index and uncertainty and time delay of traditional parameters are solved.
The analgesia depth parameter NAF (NIRS-Area-Feature) is used for carrying out multi-scale wavelet decomposition on an original signal from an acquired NIRS signal. Extracting approximate components (0-0.196Hz) under the scale of 8, and performing wavelet inverse transformation to obtain related trend components of the brain function of the original signals; the change in the relative concentrations of HbO2 and Hb was obtained by performing a Bell-Lambert law transform on the trend components of the two signals. Based on the characteristic L-wave shape generated from this plotted characteristic signal curve, the time between two L-waves is measured. Calculation of the criterion value (θ) using the sequence of intervals of the criterion vector LL over 64 seconds:
then, each LL sample is divided by a standard value (θ):
LLi'＝LLi/θ (9)
and performing energy distribution calculation on the average value and the LL interval sequence after normalization to obtain an A-LL characteristic curve. The 64s semaphore is divided into 4 segments of 16s each. The areas under each 16S sub-window, S1, S2, S3, S4, were recorded by calculating the area values under the characteristic curve A-LL. To derive a fraction of the total window area, a metric of 0-100 is calculated, and we define LLmin-min (S1, S2, S3, S4):
NAF＝[(α*LLmin)2+α*β]1/2/12.8 (10)
experimental analysis and theoretical verification prove that the calculated index NAF can accurately evaluate pain/analgesia response in the operation when alpha is 4.8 and beta is 1.6. By moving the window 64s after each calculation, continuous measurements can be made. The sampling rate of the final parameter depends on the period of the window movement.
The muscle relaxation parameter is used for monitoring the muscle relaxation degree in the operation through Relative Power Spectral Density (RPSD), namely, the muscle relaxation degree of the patient is judged according to the proportion of the myoelectric component to the total energy. The oscillation frequency according to EEG signals can be clinically divided into mainly five sub-bandwidths: delta waves (0-4Hz), theta waves (4-8Hz), alpha waves (8-13Hz), beta waves (13-30Hz), and gamma waves (30-47 Hz). The Power Spectral Densities (PSD) of these sub-bandwidths are calculated using the method of pwelch. The RSPD is calculated as follows:
p (.) represents energy and RPSD (.) represents relative power spectral density. Where f is1And f2Representing low and high frequencies, respectively. p (1,47) represents energy comprising five seed bandwidths (delta wave, theta wave, alpha wave, gamma wave) from 1Hz-47 Hz. According to the method, after the frequency band (more than or equal to 40Hz) of the electromyographic signal is extracted, the Relative Power Spectral Density (RPSD) of the electromyographic component is calculated, and the accuracy can be achievedThe patient was assessed for muscle relaxation and there was no drug interaction between the muscle relaxant and the sedative, analgesic. Through clinical experiments, the method is accurate and feasible in judgment of muscle relaxation.
According to the mixed drug patient model of the system, aiming at the synergistic effect generated by the combination of propofol and remifentanil, on the basis of the traditional PKPD model, the concentration-response relation of the two drugs can be represented by the following normalized relation:
wherein DOA (t) is the anesthetic sedative effect; t is a time(s) variable; theta is a drug concentration effect parameter; e0An initial constant (0-100) of the index; emax(θ) is the maximum possible drug effect at θ; u shapeprop(t) is the effect of propofol concentration; u shapeRem(t) concentration effect for remifentanil; u shapeprop(t)+URem(t) is the mixed drug effect; u shape50(θ) is the maximum effect constant of 50% at θ; γ (θ) is the maximum possible drug effect of the concentration-response relationship at θ.
As shown in fig. 2, the controller of the control system adopts model predictive control and adaptive control algorithm based on extended prediction in the invention from the viewpoint of practicability and reliability. For sedation/analgesia closed-loop control, after a reliable identification model is obtained, the control strategy is as follows:
(1) and carrying out model prediction. Inputs and outputs are predicted for a future time period at the current sampling instant based on the recognition model, the desired sedation/analgesia value, the applied inputs, and the obtained outputs.
(2) Optimal control is performed to bring the future output close to the desired output.
(3) And (4) optimizing rolling. And at the current moment, only adopting the first input with the most control sequences, and repeating the steps at the next sampling moment to obtain the optimal control input.
(4) By repeatedly performing predictive control optimization, an optimal control effect is expected.
The control input signal u may be expressed as: u(s) ═ K1(s) r(s) + K2(s) y(s).
For a closed-loop controller of muscle relaxation, a commonly used muscle retardant is atracurium, and for this purpose, a simple wiener model is adopted to realize single-input single-output control on the muscle relaxation.
The invention adopts the clinical data of each patient, integrates the personal physiological information of the patient into a model through the identification module, and obtains respective model parameters. The least square method is adopted for identification, and the identified patient model is a fourth-order differential equation:
where u represents the input to the model, i.e. the infusion rates of the two drugs; y represents an index of anesthesia monitoring, i.e., sedation or analgesia; a is4，a3，a2，a1，a0，b1，b2，b3Identified model parameters.
The parameter optimization of the system maps the digitized index to different states of anesthesia for control by the controller. As shown in fig. 3, a multi-view learning method is adopted to perform secondary optimization and confirmation on the index features. The specific description is as follows:
(1) based on the acquired multi-source data, different features of anesthesia are extracted and a feature pool of all the features is established. The features of different age groups are considered as multiple views.
(2) Based on multi-modal iterative computation, the initial distribution probability of the index feature label obtained by the first training of the multi-layer neural network is used as the state feature of anesthesia, and the multi-layer bible network is retrained in combination with the age feature of the patient.
(3) Based on the multi-view state mark, the most suitable feature representation generation state in each state is found by establishing the association relation between different features and the states.
The above-mentioned embodiments are merely illustrative of the preferred embodiments of the present invention, and do not limit the scope of the present invention, and various modifications and improvements of the technical solution of the present invention by those skilled in the art should fall within the protection scope defined by the claims of the present invention without departing from the spirit of the present invention.
Claims (6)
1. A vein anesthesia multi-parameter index closed-loop monitoring system is characterized in that: the system comprises a parameter signal acquisition module, a control host module and an injection device module; the parameter signal acquisition module is connected with the control host module and is used for acquiring original EEG signals, NIRS signals and EMG signals of a human body, converting the signals into indexes for monitoring a sedation depth parameter LIA, an analgesia depth parameter NAF and a muscle relaxation parameter RPSD and transmitting the indexes to the control host module; the control host module is used for parameter identification, model prediction and controller feedback calculation to obtain the dose and injection rate required by the human body in real time and provide monitoring indexes in real time; the control end of the control host module is connected with the injection device module, and the injection device module acquires the dose and the injection speed of the medicament through the control host module and then injects the medicament into a human body;
the parameter signal acquisition module consists of an EEG-NIRS signal acquisition system and is provided with an EEG signal module, an NIRS signal module and an electromyographic signal module; the EEG signal module comprises the acquisition of original signals and the extraction of sedation indexes; the NIRS signal module comprises the steps of collecting an original NIRS signal and extracting an analgesic index; the electromyographic signal module comprises the steps of collecting original signals and extracting muscle relaxation indexes;
the control host module comprises a monitoring interface, a drug-substituted model, parameter identification, a controller, index optimization and a database; the monitoring interface has a manual input function and a parameter index monitoring function; the drug-substituted model is a patient model considering the synergistic effect of the mixed use of the drugs; the parameter identification is model parameter identification based on a least square method; the controller is a model prediction control and a closed-loop controller based on an extended prediction adaptive algorithm; the index optimization is parameter optimization through feature extraction and unsupervised deep neural network learning; the database establishes patient data information for storage;
the injection device module consists of sedation, analgesia and muscle relaxation drug injectors which are designed in parallel, and transmits three drug injection speed information of sedation, analgesia and muscle relaxation calculated by the control host module to the injectors for drug injection of patients by establishing an RS232 serial port communication protocol with the control host module;
the sedation depth parameter LIA is obtained by integrating, extracting features and learning a deep neural network of feature components of four indexes of PLZC, SFS, MPF and BetaRatio;
wherein, the calculation formula of PLZC is as follows:
the SFS is calculated by the formula: SFS ═ log (P)0.5-47HZ/P40-47HZ) (ii) a The calculation formula of MPF is:
the formula for BetaRatio is: BetaRatio ═ log (P)30-47Hz/P11-20Hz)；
The LIA is obtained by SVM or random forest training on the basis of the parameters.
2. The system for closed-loop monitoring of intravenous anesthesia with multiple parameter indicators according to claim 1, wherein: the analgesia depth parameter NAF is obtained by carrying out multi-scale wavelet decomposition and Bell-Lambert rule transformation on an original NIRS signal and calculating the area under a characteristic curve A-LL; the calculated index NAF can accurately evaluate pain/analgesia reaction in the operation;
NAF＝[(α*LLmin)2+α*β]1/2/12.8 (1)
wherein alpha and beta are constant variables of the analgesia depth parameter NAF, and the value is a constant between 0 and 100; LLmin is the minimum of the area of the characteristic curve A-LL under the adjacent four sub-windows.
3. The system for closed-loop monitoring of intravenous anesthesia with multiple parameter indicators according to claim 1, wherein: the muscle relaxation parameter RPSD is used for judging the muscle relaxation of the patient according to the proportion of the myoelectric component to the total energy, extracting the frequency band where the myoelectric signal is located, wherein the frequency band is not less than 40Hz, and evaluating the muscle relaxation of the patient by calculating the relative power spectral density of the myoelectric component.
4. The system for closed-loop monitoring of intravenous anesthesia with multiple parameter indicators according to claim 1, wherein: the patient model considering the synergistic effect of the mixed use of the drugs aims at the synergistic effect generated by the combination of the propofol and the remifentanil, and the concentration-response relation of the two drugs is represented by the following normalized relation on the basis of the traditional PKPD model:
wherein DOA (t) is the anesthetic sedative effect; t is a time variable; theta is a drug concentration effect parameter; e0Is an initial constant of the index, and E0Between 0 and 100; emax(θ) is the maximum possible drug effect at θ; u shapeprop(t) is the effect of propofol concentration; u shapeRem(t) concentration effect for remifentanil; u shapeprop(t)+URem(t) is the mixed drug effect; u shape50(θ) is the maximum effect constant of 50% at θ; γ (θ) is the maximum possible drug effect of the concentration-response relationship at θ.
5. The system for closed-loop monitoring of intravenous anesthesia with multiple parameter indicators according to claim 1, wherein: the parameter identification, namely the patient model identification module, adopts a least square method to carry out identification, and the identified patient models are as follows:
wherein u represents the model inputs for the infusion rates of propofol and remifentanil; k is a variable in the infusion speed u matrix and is an integer; y represents an anesthesia monitoring index, and the anesthesia monitoring index is a sedation depth parameter LIA or an analgesia depth parameter NAF; a is4，a3，a2，a1，a0，b1，b2，b3Identified model parameters.
6. The system for closed-loop monitoring of intravenous anesthesia with multiple parameter indicators according to claim 1, wherein: the index optimization is to adopt a multi-view learning method to carry out secondary optimization and confirmation on index characteristics; the method comprises the following steps:
(1) extracting different characteristics of anesthesia and establishing a characteristic pool of all the characteristics; regarding the characteristics of different age groups as a plurality of visual angles;
(2) adopting the initial distribution probability of the index feature label obtained by the first training of the multilayer neural network as the state feature of anesthesia, and combining the age feature of the patient to retrain the multilayer neural network;
(3) and finding out the most suitable characteristic representation generation state in each state by establishing the association relationship between different characteristics and states.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810193928.4A CN108325020B (en) | 2018-03-09 | 2018-03-09 | Vein anesthesia multi-parameter index closed-loop monitoring system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810193928.4A CN108325020B (en) | 2018-03-09 | 2018-03-09 | Vein anesthesia multi-parameter index closed-loop monitoring system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108325020A CN108325020A (en) | 2018-07-27 |
| CN108325020B true CN108325020B (en) | 2021-01-08 |
Family
ID=62929051
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810193928.4A Active CN108325020B (en) | 2018-03-09 | 2018-03-09 | Vein anesthesia multi-parameter index closed-loop monitoring system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108325020B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109157231B (en) * | 2018-10-24 | 2021-04-16 | 阿呆科技（北京）有限公司 | Portable multichannel depression tendency evaluation system based on emotional stimulation task |
| CN109887599B (en) * | 2019-02-25 | 2021-07-06 | 中国计量大学 | Neural network-based traditional Chinese medicine prescription curative effect deduction method |
| CN112190233A (en) * | 2020-10-20 | 2021-01-08 | 黄淮学院 | Remote analysis management system for observing postoperative pain at home of patient |
| CN112870480B (en) * | 2021-01-22 | 2021-12-14 | 燕山大学 | Fuzzy control remote control anesthetic drug infusion pump system |
| CN113440690A (en) * | 2021-08-03 | 2021-09-28 | 复旦大学 | Intelligent quantitative drug administration electric needle injection device based on electromyographic signal feedback |
| CN113648484B (en) * | 2021-09-06 | 2022-11-01 | 北京航空航天大学 | Portable target-controlled infusion intravenous anesthesia device and target-controlled infusion control method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102793534A (en) * | 2012-07-11 | 2012-11-28 | 广西威利方舟科技有限公司 | Anesthesia depth monitoring device and method |
| CN103908249A (en) * | 2014-01-28 | 2014-07-09 | 广西威利方舟科技有限公司 | Anaesthetic balance control device and control method |
| CN105852804A (en) * | 2016-03-24 | 2016-08-17 | 美合实业（苏州）有限公司 | Portable anesthesia depth monitor |
| CN106902421A (en) * | 2017-01-17 | 2017-06-30 | 燕山大学 | The individualized anesthesia closed-loop control system of one kind |
-
2018
- 2018-03-09 CN CN201810193928.4A patent/CN108325020B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102793534A (en) * | 2012-07-11 | 2012-11-28 | 广西威利方舟科技有限公司 | Anesthesia depth monitoring device and method |
| CN103908249A (en) * | 2014-01-28 | 2014-07-09 | 广西威利方舟科技有限公司 | Anaesthetic balance control device and control method |
| CN105852804A (en) * | 2016-03-24 | 2016-08-17 | 美合实业（苏州）有限公司 | Portable anesthesia depth monitor |
| CN106902421A (en) * | 2017-01-17 | 2017-06-30 | 燕山大学 | The individualized anesthesia closed-loop control system of one kind |
Non-Patent Citations (2)
| Title |
|---|
| Design of multichannel functional near-infrared spectroscopy system with application to propofol and sevoflurane anesthesia monitoring;Zhenhu Liang,Yue Gu,Xuejing Duan;《Neurophotonics》;20161005;第3卷(第4期);第1-11页 * |
| EEG熵算法及麻醉状态监测应用研究;梁振虎;《中国博士学位论文全文数据库医药卫生科技辑》;20121015(第10期);E066-2页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108325020A (en) | 2018-07-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108325020B (en) | Vein anesthesia multi-parameter index closed-loop monitoring system | |
| CN105147248A (en) | Physiological information-based depressive disorder evaluation system and evaluation method thereof | |
| CN105249964B (en) | Appraisal procedure is rebuild based on magneticencephalogram and the multi-modal brain function of diffusion tensor | |
| EP2906112B1 (en) | System and method for monitoring and controlling a state of a patient during and after administration of anesthetic compound | |
| CN110598676B (en) | Deep learning gesture electromyographic signal identification method based on confidence score model | |
| Obayya et al. | Automatic classification of sleep stages using EEG records based on Fuzzy c-means (FCM) algorithm | |
| Liu et al. | EEG classification with a sequential decision-making method in motor imagery BCI | |
| CN110193113A (en) | Multi-channel intelligent drug delivery system | |
| Ragnarsdóttir et al. | Automatic detection of target regions of respiratory effort-related arousals using recurrent neural networks | |
| CN114177415A (en) | Intelligent method and system for monitoring anesthesia target control of old people based on electroencephalogram signals | |
| Tang et al. | The novel approach of temporal dependency complexity analysis of heart rate variability in obstructive sleep apnea | |
| CN114652328A (en) | Bidirectional closed-loop brain-computer interaction electro-acupuncture parameter intelligent matching system for insomnia | |
| CN114648040A (en) | Method for extracting and fusing multiple physiological signals of vital signs | |
| CN113974557A (en) | Deep neural network anesthesia depth analysis method based on electroencephalogram singular spectrum analysis | |
| CN105740772A (en) | Real time monitoring method of cortical somatosensory evoked potential of transform joint sparse model | |
| Ghanbari et al. | Brain computer interface with genetic algorithm | |
| Banerjee et al. | Detecting eye movement direction from stimulated Electro-oculogram by intelligent algorithms | |
| CN108241431B (en) | Task adjusting method and device | |
| Goldberger et al. | Fractals and the heart | |
| Sikder et al. | Heterogeneous hand guise classification based on surface electromyographic signals using multichannel convolutional neural network | |
| Mahajan et al. | Automated cardiac state diagnosis from hybrid features of ECG using neural network classifier | |
| Van Den Broek | Monitoring Anesthetic Depth Modification, Evaluation and Application of the Correlation Dimension | |
| Zhou et al. | Application of back propagation neural network and information entropy in deep detection of anesthesia | |
| Zhao et al. | Research on steady state visual evoked potential based on FBCCA | |
| Tseng et al. | ECG sensor verification system with mean-interval algorithm for handling sport issue |
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 |