Oxygen flow control system of oxygen inhalation machine for pneumology department

Abstract

The invention relates to the technical field of oxygen inhalation machines, in particular to an oxygen flow control system of an oxygen inhalation machine for the pneumology department, which comprises a data acquisition unit used for acquiring relevant parameters of effective oxygen flow regulation; the data processing unit comprises an oxygen demand calculation module, an oxygen flow calculation module, an oxygen flow regulation module and a parameter abnormity alarm module. Various physiological parameters, namely related parameters influencing oxygen flow regulation, of a patient are collected in real time, the collected related parameters are calculated and processed through the data processing unit, the predicted oxygen demand of the patient can be obtained, and then the oxygen regulation amount needing to be regulated by the oxygen inhalation machine is calculated through the predicted oxygen demand; compared with a traditional oxygen adjusting mode, the scheme has the advantages that the individual demand of each patient can be responded more accurately, so that the oxygen supply accuracy is improved, and the risk caused by insufficient or excessive supply is avoided.

Description

Technical Field

[0001] The invention relates to the technical field of oxygen concentrators, and in particular to an oxygen flow control system for an oxygen concentrator in the respiratory department. Background Art

[0002] Oxygen concentrators are medical devices used to provide oxygen support and are widely used in hospitals, home care and emergency situations. With the aging population and the increase in respiratory diseases, oxygen therapy plays an important role in improving patients' quality of life, alleviating symptoms and increasing survival rates.

[0003] After searching, Chinese patent No. CN202011030690.7 discloses an oxygen flow control system and method for respiratory support equipment. The above scheme detects and collects the oxygen flow of the intake pipeline in real time through a flow sensor, and the processor determines whether the oxygen flow of the intake pipeline has changed; if it has changed, the processor returns to initialization, and the flow sensor collects the oxygen flow of the intake pipeline again, and compares the collected oxygen flow with the standard oxygen flow. Within the preset deviation value range, the aperture of the proportional valve is controlled by the fuzzy control circuit, and then the oxygen flow of the intake pipeline is adjusted to output a stable oxygen flow to the user end.

[0004] Most traditional oxygen flow control methods rely on flow sensors, but in clinical practice, this method has certain limitations. For example, individual differences in patients' breathing, activity status, and oxygen demand make it difficult to accurately control oxygen flow. In addition, different disease states (such as chronic obstructive pulmonary disease, pneumonia, heart failure, etc.) require more precise regulation of oxygen flow based on specific factors.

[0005] Therefore, an oxygen flow control system for a respiratory oxygen inhaler is proposed to solve the above-mentioned problems. Summary of the invention

[0006] Technical issues solved

[0007] In view of the above-mentioned shortcomings of the prior art, the present invention provides an oxygen flow control system for a respiratory oxygen concentrator, which can effectively solve the problem that the oxygen flow of the oxygen concentrator in the prior art cannot be finely adjusted according to the individual differences of patients.

[0008] Technical Solution

[0009] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0010] The present invention provides an oxygen flow control system for a respiratory oxygen inhaler, comprising a data acquisition unit for obtaining relevant parameters for effective oxygen flow regulation;

[0011] It also includes a data processing unit, which includes:

[0012] The oxygen demand prediction calculation module is used to build an oxygen demand prediction model based on relevant parameters and calculate the predicted oxygen demand

[0013] Oxygen flow calculation module, used to calculate oxygen flow according to relevant parameters and oxygen demand Construct an oxygen regulation model and calculate the oxygen regulation amount ΔO;

[0014] Oxygen flow regulation module, based on predicted oxygen demand The oxygen flow rate is calculated by the oxygen adjustment amount ΔO PID control and accelerated response mechanism are introduced to control the oxygen flow rate O f Adjust speed for optimization;

[0015] The parameter abnormality alarm module is used to define the warning range of relevant parameters and detect whether there are abnormal items in the relevant parameters. If there are abnormal items in the relevant parameters, an alarm will be issued.

[0016] Furthermore, the related parameters include: heart rate HR, physical activity intensity A, carbon dioxide emission CO2, blood oxygen saturation SpO2, respiratory rate f R , tidal volume V T , positive end-expiratory pressure (PEEP).

[0017] Furthermore, the method of constructing the oxygen demand prediction model includes:

[0018] Defining and predicting oxygen requirements There are related parameters with nonlinear relationships. Based on the polynomial regression model, we get Y=β0+β1·X1+β2·X2+...+β n ·X k +β n+1 X1 2 +β n+2 X2 2 +...+β n+k ·X k 2 +...+ε; where Y is the dependent variable, i.e., the predicted oxygen demand β0 is the intercept of the model; X is the independent variable, i.e., the relevant parameter; k represents the number of independent variables; β1, β2, β3, …, β n Corresponding to the independent variable X k The regression coefficient of n represents the total number of regression coefficients; X1 2 , X2 2 , …, X k 2is the square term of the independent variable X; ε is the error term; then the oxygen demand prediction model is evaluated to obtain the mean square error MSE, the optimal coefficient vector β, and the determination coefficient R of the oxygen demand prediction model 2 and the residual ε.

[0019] Furthermore, the method of evaluating the oxygen demand prediction model includes:

[0020] Define the loss function as mean square error Where N is the total number of samples; is the model prediction value, that is, the predicted oxygen demand Y i is the true value, that is, the actual oxygen demand O d ;

[0021] Solve the optimal coefficient vector β=(X matrix T X matrix ) -1 X matrix T Y vector Where, X matrix is the independent variable matrix, Y vector is the dependent variable vector, X matrix T is the independent variable matrix X matrix The transpose of X matrix T X matrix represents the transpose of the independent variable matrix X matrix T and the independent variable matrix X matrix The product of itself, (X matrix T X matrix ) -1 Represents the matrix X matrix T X matrix The inverse matrix, X matrix T Y vector Represents the independent variable matrix X matrix The transpose of the dependent variable vector Y vector The product of

[0022] Defining the coefficient of determination In the formula, the error sum of squares is Total sum of squares in is the mean of the dependent variable Y;

[0023] Defining residuals ε i represents the residual of the ith sample.

[0024] Furthermore, the method of constructing the oxygen regulation model includes:

[0025] Defining oxygen regulation

[0026] ΔO=K adj ·(HR target -HR)+K PEEP ·PEEP target -PEEP)+K SpO2 (SpO2 target -SpO2):

[0027] In the formula, HR target The target heart rate that needs to be adjusted for the patient, K adj Indicates the patient's current heart rate HR and target heart rate HR target Adjustment factor for the difference between target K is the target positive end-expiratory pressure value that needs to be adjusted for the patient. PEEP Indicates the patient's current PEEP and target PEEP target Adjustment factor for the difference between target The target blood oxygen saturation that needs to be adjusted for the patient, K SpO2 Indicates the patient's current blood oxygen saturation SpO2 and target blood oxygen saturation SpO2 target The adjustment factor for the difference between .

[0028] Furthermore, the PID control and acceleration response mechanism optimizes the oxygen flow rate adjustment speed in the following manner:

[0029] Introduce the acceleration response factor t, that is, time point t, and then define the dynamic regulation model of oxygen flow. Get the oxygen flow rate O at time t f (t), that is, the current flow value to be calculated; where G p is the proportional gain; e(t) represents the error at the current moment, and the calculation formula is G i is the integral gain; G d is the differential gain; G acc To accelerate the response gain.

[0030] Furthermore, the integral gain G i The method of obtaining includes: first setting G i =0 as the initial value, and then gradually increase the integral gain G i , observe the response of the oxygen flow dynamic regulation model at the same time, and record the integral gain G each time i Adjusted oxygen flow O f(t) and the error e(t) at the current moment, until the steady-state error of the oxygen flow dynamic regulation model Close to zero, determine the optimal integral gain G i ;

[0031] The differential gain G d The acquisition method includes: first calculating the error change Δe(t) = e(t)-e(t-1); then using the differential gain G d Update the current output of the oxygen flow dynamic regulation model: oxygen flow O f (t) = O f (t-1)+K d Δe(t); then apply a changing input signal, i.e. predict oxygen demand And detect O f (t), and record the error e(t) and the error change Δe(t); then adjust the differential gain G d The value of the differential gain G is determined by adjusting the output fluctuation of the oxygen flow dynamic regulation model until it decreases and responds smoothly. d Optimal value;

[0032] The acceleration response gain G acc The acquisition methods include: using the acceleration response gain G acc Update the current output of the oxygen flow dynamic regulation model: oxygen flow O f (t) = O f (t-1)+K acc Δe(t); then apply a rapidly changing input signal, i.e., predict oxygen demand Then observe and predict oxygen demand When the oxygen flow rate O changes f (t), and whether the oxygen flow dynamic regulation model can return to a new steady state; finally, using the real-time monitoring data, the acceleration response gain G is gradually adjusted. acc Specifically, a threshold η of the error change is set, and when |Δe(t)>η, the acceleration response gain G is triggered. acc The adjustment formula is G acc (t) = G acc (t-1)+α·Δe(t); where G acc (t-1) represents the acceleration response gain value at time point t-1; α is the step size.

[0033] Furthermore, the warning range of the defined relevant parameters includes:

[0034] Heart rate HR warning range [HR min ,HR max ], HR minis the minimum heart rate warning threshold, HR max is the maximum heart rate warning threshold;

[0035] SpO2 warning range min ,SpO2 max ], SpO2 min The minimum blood oxygen saturation warning threshold, SpO2 max It is the highest blood oxygen saturation warning threshold;

[0036] Respiratory rate R The warning range [f Rmin ,f Rmax ], f Rmin is the minimum respiratory rate warning threshold, f Rmax is the maximum respiratory rate warning threshold;

[0037] Tidal volume V T The warning range [V Tmin ,V Tmax ],V Tmin is the minimum tidal volume warning threshold, V Tmax is the maximum tidal volume warning threshold;

[0038] Carbon dioxide emission CO2 warning range (CO2 ≤ CO2 max ), CO2 max A unilateral cap on maximum CO2 emissions;

[0039] Warning range of positive end-expiratory pressure (PEEP) min ,PEEP max ], PEEP min is the minimum positive end-expiratory pressure warning threshold, PEEP max It is the maximum positive end-expiratory pressure warning threshold;

[0040] Warning range of body movement intensity A (A≤A max ), A max It is the unilateral upper limit of the maximum body movement.

[0041] Furthermore, the method of detecting whether there are abnormal items in the relevant parameters includes:

[0042] Check heart rate HR, blood oxygen saturation SpO2, respiratory rate f in turn R , tidal volume V T , carbon dioxide emissions CO2, positive end-expiratory pressure PEEP, and body movement intensity A, whether they exceed their warning range or unilateral upper limit. If they exceed their warning range or unilateral upper limit, they are abnormal items.

[0043] Beneficial Effects

[0044] Compared with the prior art, the technical solution provided by the present invention has the following beneficial effects:

[0045] The present invention collects various physiological parameters of the patient in real time through a data collection unit, that is, relevant parameters that affect oxygen flow regulation, and calculates and processes the collected relevant parameters through a data processing unit to obtain the patient's predicted oxygen demand, and then calculates the oxygen adjustment amount that needs to be adjusted by the oxygen concentrator through the predicted oxygen demand, and finally obtains the oxygen flow that can meet the individual needs of different patients. Compared with the traditional oxygen regulation method, this solution can more accurately respond to the individual needs of each patient, thereby improving the accuracy of oxygen supply and avoiding the risks caused by insufficient or excessive supply;

[0046] This solution introduces an accelerated response factor when calculating the oxygen flow rate that meets the individual needs of patients, and then defines a dynamic oxygen flow adjustment model. When the patient's physiological parameters change rapidly, the oxygen flow rate can be adjusted quickly to meet the patient's needs. After each oxygen flow adjustment, the error is recalculated, so that the system can continuously monitor the patient's response, so that each adjustment is based on the latest physiological data, thereby achieving continuous optimization. This adaptive ability makes oxygen therapy more reliable;

[0047] Finally, a parameter abnormality alarm module is set in this scheme to f During the adjustment process, by setting the warning thresholds for the above-mentioned related parameters, the oxygen demand is predicted. When any parameter is abnormal during calculation, an early warning will be triggered to remind medical staff, thereby significantly improving patient safety and treatment effects, and avoiding the situation where there are abnormal items in related parameters when estimating predicted oxygen demand. Even if the calculated results can be met, it will lead to misjudgment of the patient's condition, which in turn leads to major clinical crises and wrong decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0049] Figure 1 Schematic diagram of the structure of the oxygen flow control system of the oxygen concentrator in the embodiment of the present invention.

[0050] Figure 2 The figure is a schematic diagram of the working process of the parameter abnormality alarm module in an embodiment of the present invention. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] The present invention will be further described below in conjunction with the embodiments.

[0053] Example:

[0054] This case proposes an oxygen flow control system for a respiratory oxygen inhaler. Figure 1 ,The system includes a data acquisition unit and a data processing unit.

[0055] The data acquisition unit is used to obtain relevant parameters that affect oxygen flow regulation, that is, the patient's physiological parameters, and the relevant parameters include:

[0056] The patient's heart rate HR reflects the number of heart beats per minute (unit: beats / min), which directly affects the body's demand for oxygen. An increase in heart rate usually indicates an increase in physical activity or an increase in metabolic rate, thereby increasing the demand for oxygen. It is obtained through a heart rate monitor;

[0057] The patient's physical activity intensity A refers to the patient's activity. Generally, the higher the intensity, the greater the demand for oxygen. Quantifying the physical activity intensity A can help assess the patient's condition and the amount of oxygen needed. The method for obtaining it is: A = ω1·S A +ω2·S HR +ω3·S TQ , where S A It represents the patient's activity intensity score. By wearing a motion sensor on the patient, the patient's movement trajectory, speed, acceleration and other information are obtained based on the motion sensor, and the patient's activity duration and activity intensity are calculated. Finally, it is converted into a quantitative activity index. By analyzing the acquired acceleration data, the activity intensity can be divided into static, light, moderate and high activity, and different thresholds are used to distinguish the exercise intensity. The common activity threshold is metabolic equivalent METS, and then the patient's activity intensity score S can be obtained. A ; S HR Indicates the patient's heart rate score, which can be obtained through a motion sensor or a heart rate monitor; S TQrepresents the patient's activity time score, which can also be obtained through motion sensors; ω1, ω2, and ω3 are corresponding weight factors, which can be adjusted based on clinical experience;

[0058] The patient's carbon dioxide emissions (CO2) refer to the amount of carbon dioxide produced by the patient's body metabolism, which is usually proportional to the oxygen consumption. The increase in carbon dioxide indicates that the body is consuming more oxygen, so its quantitative information is crucial for regulating oxygen supply. The carbon dioxide emissions (CO2) can be monitored in real time through a breathalyzer (unit: L / min);

[0059] The patient's blood oxygen saturation SpO2 indicates the saturation level of oxygen in the blood, which directly reflects the patient's oxygen acquisition. A decrease in blood oxygen saturation SpO2 indicates insufficient oxygen supply or that the patient's respiratory system cannot work properly. Blood oxygen saturation SpO2 is obtained through real-time monitoring of a pulse blood pressure monitor and is usually output in percentage form.

[0060] The patient's respiratory rate f R , refers to the number of times a patient breathes per minute (unit: times / min), which is usually proportional to the body's demand for oxygen. A higher respiratory rate indicates that the body is trying to get more oxygen, which can be obtained through real-time monitoring of respiratory monitoring equipment;

[0061] The tidal volume V breathed by the patient T , represents the volume of air inhaled per breath (unit: L), tidal volume V T The size of will affect the amount of oxygen that can be provided in each breath, so it is an important parameter for regulating the oxygen flow rate, which can be obtained through real-time monitoring of the respiratory flow meter;

[0062] The patient's positive end-expiratory pressure (PEEP) refers to the positive pressure in the patient's lung airway at the end of exhalation. PEEP can reduce the patient's resistance during inhalation and maintain the development of the patient's alveoli, thereby increasing the functional residual capacity and the effective ventilation area of ​​the lungs, which helps to improve the exchange of oxygen and carbon dioxide and reduce the demand for high-concentration oxygen. It is obtained through airway pressure sensor monitoring, and the acquisition method is PEEP = P exp -P atm , where P exp It represents the pressure value of the last exhalation phase of each respiratory cycle of the patient, P atm is the atmospheric pressure value;

[0063] Oxygen flow rate f , refers to the volume of oxygen actually provided to the patient by the oxygen concentrator. It is an important parameter that directly affects the patient's oxygen supply and is obtained by monitoring the oxygen supply pipeline through a flow meter.

[0064] The data processing unit includes a predicted oxygen demand calculation module, an oxygen flow calculation module and an oxygen flow regulation module. The predicted oxygen demand module is used to calculate and process the relevant parameters obtained by the data acquisition unit to obtain the patient's predicted oxygen demand. Predicting oxygen needs It reflects the amount of oxygen required by the patient under a specific physiological state, directly affects the regulation of oxygen flow, and ensures that the patient gets enough oxygen; the oxygen flow calculation module is used to calculate the predicted oxygen demand. As an input item, the oxygen adjustment amount ΔO is obtained through calculation and processing. The oxygen adjustment amount ΔO and the predicted oxygen demand The oxygen flow rate O that meets the patient's oxygen demand can be calculated f ; Then the real-time oxygen flow output by the oxygen concentrator is controlled through the oxygen flow regulation module.

[0065] Specifically, the oxygen demand prediction calculation module calculates and processes the relevant parameters in the following way: define and predict oxygen demand There are related parameters with nonlinear relationships, including:

[0066] Heart rate HR. The relationship between heart rate HR and oxygen demand may be linear to a certain extent, because an increase in heart rate HR generally means that the body needs more oxygen to meet the heart's pumping needs. However, this relationship is not completely linear, because an increase in heart rate HR may also be a reflection of other factors (such as body temperature, pain, anxiety, etc.), which may not directly increase oxygen demand;

[0067] The relationship between physical activity intensity A and oxygen demand is usually positive, but also nonlinear; at low-intensity exercise, the increase in oxygen demand may be proportional to the increase in physical activity intensity A; while at high-intensity exercise, the increase in oxygen demand may become less significant;

[0068] The relationship between carbon dioxide emissions CO2, oxygen demand and carbon dioxide emissions CO2 is usually linear, because oxygen consumption and carbon dioxide production have a certain proportional relationship in the metabolic process. However, under extreme conditions or in certain pathological states, this ratio may change;

[0069] Blood oxygen saturation SpO2. When blood oxygen saturation SpO2 is within the normal range, the body is able to maintain an adequate oxygen supply. However, once SpO2 drops, especially when it is below 90%, oxygen demand increases dramatically because the body needs more oxygen to maintain the function of vital organs.

[0070] Respiratory rate R , under normal circumstances, the respiratory rate f RIncreases in oxygen intake may result in increased oxygen uptake, but the relationship is not linear because excessively rapid breathing (such as hyperventilation) may lead to excessive excretion of carbon dioxide, causing respiratory alkalosis, which may adversely affect oxygen transport and utilization;

[0071] Tidal volume V T , increase tidal volume V T Oxygen intake can be increased, but this increase is limited, and excessive tidal volumes may cause alveolar overdistension and barotrauma.

[0072] Then a polynomial regression model is constructed, namely the oxygen demand prediction model:

[0073] Y=β0+β1·X1+β2·X2+...+β n ·X k +β n+1 X1 2 +β n+2 X2 2 +...+β n+k ·X k 2 +...+ε;

[0074] Where, Y(predicted oxygen demand ) is the dependent variable; β0 is the intercept of the model; X is the independent variable, k represents the number of independent variables, X k represents the kth independent variable X. Specifically, in this case, the independent variable X has six related parameters, namely X1 (heart rate HR), X2 (physical activity intensity A), X3 (carbon dioxide emission CO2), X4 (blood oxygen saturation SpO2), X5 (respiratory frequency f R )、X6(Tidal Volume V T );β1,β2,β3,…,β n Corresponding to the independent variable X k The regression coefficient indicates the effect of a unit change in the independent variable X on the predicted oxygen demand. The degree of influence of n is the total number of regression coefficients associated with the model; X1 2 , X2 2 , …, X k 2 is the square term of the independent variable X, X k 2 represents the square term of the kth independent variable, indicating a nonlinear relationship; ε is the error term. In this scheme, the relationship between the dependent variable and the independent variable is nonlinear, so the square of the independent variable or a higher power can explain the interaction between the variables, and can better fit the data in different data segments, thereby providing more accurate prediction results.

[0075] After the oxygen demand prediction model is built, it needs to be evaluated to determine the fit and prediction ability of the polynomial regression model. The specific evaluation methods include:

[0076] First, define the loss function as the mean square error Where N is the total number of samples; is the model prediction value, that is, the predicted oxygen demand Y i is the true value, that is, the actual oxygen demand O d Specifically, a historical data set of relevant parameters and actual oxygen demand of several sample patients is collected. d ; Then the data set is cleaned, missing values ​​and outliers in the data set are checked, and appropriate processing (such as interpolation of missing values ​​and removal of outliers) is performed to ensure the integrity and quality of the data. Then the patient's historical data set is imported into the oxygen demand prediction model to obtain the predicted oxygen demand. Then, by calculating the MSE, we can measure the difference between the model prediction value and the true value, that is, the predicted oxygen demand and actual oxygen demand O d The square form of MSE can effectively focus on larger errors because the square of the error will also be magnified. The smaller the value of MSE, the better the predictive ability of the polynomial regression model.

[0077] Then, in order to improve the prediction accuracy of the oxygen demand prediction model as a whole, the coefficient vector β (which contains the regression coefficients of all independent variables X in the polynomial regression model, if there are k independent variables X, then β is a k*1 column vector) is optimized by minimizing the MSE. The optimization method is: solve the coefficient vector β = (X matrix T X matrix ) - 1 X matrix T Y vector ;

[0078] In the formula, X matrix is the independent variable matrix, which contains all the independent variable values. Each row represents the observed value of a sample, that is, the historical data set of the relevant parameters in this case, and each column represents an independent variable, that is, the relevant parameters in this scheme. For example, in this case, the historical data set of the relevant parameters is converted into a table format, which can be expressed as:

[0079] HR A CO2 SpO2 <![CDATA[f R ]]> <![CDATA[V T ]]> 70 1 0.03 95 16 500 75 2 0.04 97 18 550 ... ... ... ... ... ...

[0080] Then extract data from the above table to form the independent variable matrix X matrix, the shape is N*m, where N is the number of observations and m is the number of independent variables, which is 6 in the above example;

[0081] Y vector is the dependent variable vector, which is a vector containing all the real measured dependent variables, i.e., the actual oxygen demand O of all patient samples in this case. d , each actual oxygen demand O d The oxygen demand value corresponding to each observation; for example, in this case, the actual oxygen demand O of all sample patients d Arranged into a one-dimensional array, it can be expressed as:

[0082] <![CDATA[O d ]]> 250 270 ...

[0083] Then create the dependent variable vector Y vector , the shape is N*1;

[0084] For example, there are two sets of historical data sets:

[0085] First historical dataset:

[0086] HR:70, A:1, CO2:0.03, SpO2:95, f R :16, V T :500

[0087] O d :250

[0088] Second historical dataset:

[0089] HR:75, A:2, CO2:0.04, SpO2:97, f R :18, V T :550

[0090] O d :270

[0091] Then the independent variable matrix X matrix and the dependent variable vector Y vector The construction is as follows:

[0092]

[0093] X matrix T is the independent variable matrix X matrix The transpose of the independent variable matrix X is matrix The number of rows becomes the number of independent variables, the number of columns becomes the number of observations, and the shape is m*N;

[0094] X matrix T Xmatrix represents the transpose of the independent variable matrix X matrix T and the independent variable matrix X matrix The product of itself is an m*m matrix (the number of independent variables is the same as the number of dependent variables), and the elements of this matrix represent the correlation between the independent variables X;

[0095] (X matrix T X matrix ) -1 Represents the matrix X matrix T X matrix The inverse matrix of is used to adjust the mutual influence between the independent variables X when solving the coefficient vector β;

[0096] X matrix T Y vector Represents the independent variable matrix X matrix The transpose of the dependent variable vector Y vector The product of is an m*1 column vector. This product can determine the influence of each independent variable X on the dependent variable Y, indicating the weighted contribution of the dependent variable Y to the independent variable X.

[0097] Then, through the above formula, an optimal coefficient vector β can be obtained, so that the polynomial regression model can best fit the observed data and minimize the gap between the predicted value and the true value, that is, predict the oxygen demand and actual oxygen demand O d The gap increases the oxygen flow rate O f The precision of regulation can provide patients with the required amount of oxygen more accurately, avoiding overdose or underdose.

[0098] Then, the coefficient of determination for evaluating the goodness of fit of the oxygen demand prediction model was defined In the formula, the error sum of squares is Used to measure the difference between the predicted value and the true value; the total sum of squares It is used to measure the difference between the true value and the mean. is the mean of the dependent variable Y. Then the coefficient of determination R is calculated by the above formula 2 The value obtained is between 0 and 1. The closer it is to 1, the better the fit of the polynomial regression model.

[0099] Finally, define the residual ε i Represents the residual of the i-th sample. By i and the corresponding residual ε iDraw a residual graph (usually a scatter plot), and then observe the random distribution of the residual ε in the residual graph. If the residual graph shows random scatter, it indicates that the assumptions of the polynomial regression model are well met. If it is observed that the residual ε varies with the predicted value Y i If the data do not change with changes in the data (such as the hole shape or other patterns), then the polynomial regression model is not appropriate.

[0100] In summary, in this scheme, by calculating the mean square error MSE, coefficient vector β, and determination coefficient R 2 and residual ε, can effectively evaluate the predictive ability of the oxygen demand prediction model.

[0101] Furthermore, in this case, the patient's predicted oxygen demand is calculated by the oxygen demand prediction model. After that, the oxygen demand will be predicted The oxygen flow calculation module is input into the oxygen flow calculation module, which combines the above-mentioned related parameters to build a regression model, namely the oxygen regulation model, and calculates and processes the oxygen regulation amount ΔO, and then predicts the oxygen demand The oxygen flow rate O is calculated by the oxygen adjustment amount ΔO f .

[0102] Specifically, the construction method of the oxygen regulation model, that is, the calculation and processing method of the oxygen regulation amount ΔO includes:

[0103] Defining oxygen regulation

[0104] ΔO=K adj ·(HR target -HR)+K PEEP ·PEEP target -PEEP)+K SpO2 (SpO2 target -SpO2); where,

[0105] The oxygen adjustment amount ΔO indicates the degree of adjustment of oxygen quality based on the gap between the patient's current physiological state and the target state; HR target The target heart rate that the patient needs to adjust is usually set by the doctor based on the patient's specific needs; K adj Indicates the patient's current heart rate HR and target heart rate HR target Adjustment factor for the difference between target The target positive end-expiratory pressure value that needs to be adjusted for the patient is usually set to the ideal PEEP value to optimize the patient's oxygenation; K PEEP Indicates the patient's current PEEP and target PEEP target Adjustment factor for the difference between targetThe target blood oxygen saturation that needs to be adjusted for the patient is usually set to the ideal SpO2 value to ensure that the patient gets enough oxygen; SpO2 Indicates the patient's current blood oxygen saturation SpO2 and target blood oxygen saturation SpO2 target The adjustment factor for the difference between the two.

[0106] Oxygen flow rate f The calculation methods include:

[0107] It is worth mentioning that the oxygen flow rate O calculated in the oxygen flow calculation module is f After that, the oxygen flow rate regulating module adjusts the oxygen flow rate O f Rapid feedback to the oxygen breathing machine system, thereby adjusting the oxygen flow O f Adjust the oxygen flow rate O f During the adjustment process, the oxygen flow rate O f The speed of the adjustment is controlled to avoid the oxygen flow O f The adjustment speed is too slow, causing the patient to be in critical condition due to insufficient oxygen supply, or the oxygen flow rate O f If the adjustment speed is too fast, the oxygen supply will fluctuate, affecting the comfort and safety of the patient. f The oxygen flow calculation module introduces PID control and acceleration response mechanism to process the oxygen flow regulation speed. The processing methods of PID control and acceleration response mechanism include:

[0108] Introduce the acceleration response factor t, t represents the time point, and then define the dynamic regulation model of oxygen flow Get the oxygen flow rate O at time t f (t), which is the current flow value to be calculated.

[0109] In the formula, O f (t-1) represents the oxygen flow rate at the previous moment t-1, which is the reference value for the current oxygen flow rate adjustment;

[0110] G p is the proportional gain, which is used to control the influence of the current error on the oxygen flow adjustment; e(t) represents the error at the current moment, and the calculation formula is: Represents the difference between the target oxygen demand and the current oxygen flow; G p e(t) represents the error at the current moment multiplied by the proportional gain, which gives the contribution of the error at the current moment to the adjustment of the oxygen flow rate;

[0111] G i is the integral gain, which is used to accumulate the errors at all previous time points to compensate for long-term errors; It represents the sum of the error values ​​from time 0 to time t, reflecting the total error over the past period of time. It represents the past accumulated error multiplied by the integral gain, which is used to help accurately adjust the oxygen flow to eliminate the steady-state error of the system;

[0112] G d is the differential gain, which indicates the influence of the current error change rate on the oxygen flow adjustment; [e(t)-e(t-1)] indicates the difference between the error at the current moment and the error at the previous moment, which indicates the change of the error; G d [e(t)-e(t-1)] represents the change in error multiplied by the differential gain, which is used to take into account the rate of change of the error, help predict future behavior, and provide a dynamic response to the oxygen flow rate;

[0113] G acc To accelerate the response gain, it is a new adjustment factor, representing the sensitivity of flow adjustment; G acc [e(t)-e(t-1)] represents the change in error multiplied by the acceleration response gain, providing an acceleration response factor t to further adjust the oxygen flow rate so as to respond quickly when the error changes significantly, especially in the case of rapid changes in physiological state.

[0114] Furthermore, through the above-mentioned oxygen flow dynamic adjustment model, it can be ensured that the oxygen flow can be adjusted quickly and accurately when the patient's physiological parameters change rapidly, so that the oxygen concentrator can provide the required oxygen flow more effectively under different circumstances.

[0115] Among them, the integral gain G i The historical error is weighted so that the oxygen flow dynamic regulation model can reach the target value in the long term. i Ways to obtain include:

[0116] First set G i =0 as the initial value; then gradually increase the integral gain G i For example, you can choose to increase by 0.1 or 0.05 each time to observe the response of the oxygen flow rate dynamic regulation model; then record the output oxygen flow rate O of the oxygen flow rate dynamic regulation model after each adjustment. f (t) and the error e(t) at the current moment; finally, until the steady-state error of the oxygen flow dynamic regulation model is observed Close to zero, determine the optimal integral gain G i .

[0117] Example, assuming oxygen demand is predicted Initial oxygen flow O f (0) = 8 L / min, the initial cumulative error is 0;

[0118] Calculate initial error

[0119] Update cumulative error CE = 0 + e(0) = 0 + 2 = 2;

[0120] Select the integral gain G i =0.1, update oxygen flow O f (1) = O f (0)+K i ·CE=8+0.1·2=8.2;

[0121] Calculate the error at the next moment

[0122] Update cumulative error CE = 0 + e(1) = 2 + 1.8 = 3.8;

[0123] Adjust the integral gain G i =0.2, update oxygen flow O f (2) = O f (1)+K i CE = 8.2 + 0.2 3.8 = 8.96;

[0124] Continue this process by repeating the above steps to continuously calculate and gradually adjust the integral gain G i , and monitor the performance of the oxygen flow dynamic regulation model until the optimal integral gain G is found i value.

[0125] Among them, the differential gain G d It is used to respond to the rate of error change, predict the future behavior of the oxygen flow dynamic regulation model, and help slow down the oscillation of the oxygen flow dynamic regulation model and increase stability. d Ways to obtain include:

[0126] First, calculate the error change Δe(t)=e(t)-e(t-1), that is, at each time point t, the difference between the current error and the error at the previous moment;

[0127] Second, use the differential gain G d Update the current output of the oxygen flow dynamic regulation model:

[0128] Oxygen flow rate f (t) = O f (t-1)+K d Δe(t);

[0129] Then a rapidly changing input signal is applied, i.e. the predicted oxygen demand For example From 10L / min suddenly changed to 15L / min;

[0130] Then, after each fast-changing input signal is input, the actual output O of the oxygen flow dynamic regulation model is monitored. f (t), and record the error e(t) and the error change Δe(t);

[0131] Next, adjust the differential gain G d The value of; initially set to a small initial value, such as G d =0.1; then gradually increase G d , for example, increase by 0.1 or 0.2 each time;

[0132] After that, monitor the differential gain G d During adjustment, the oxygen flow rate dynamically adjusts the output oxygen flow rate O of the model. f (t) and error e(t), and pay attention to the stability and response time of the oxygen flow dynamic regulation model;

[0133] Finally, adjust the differential gain G d The value of the differential gain G is determined until the output fluctuation of the oxygen flow dynamic regulation model is reduced and the response is smoother without causing obvious delay. d Optimal value.

[0134] Example, assuming the setting condition is to predict oxygen demand Initial oxygen flow O f (0) = 15 L / min;

[0135] Calculate initial error

[0136] Time step 1: Set the derivative gain G d The initial value is 0.1;

[0137] Calculate the error at the next moment

[0138] Calculate the error change Δe(1)=e(1)-e(0)=5-5=0;

[0139] Update oxygen flow O f (1) = O f (0)+K d Δe(1)=15+0.1·0=15;

[0140] Time step 2: Assume that the set condition changes to predict oxygen demand

[0141] Calculate the current error

[0142] Calculate the error change Δe(2)=e(2)-e(1)=10-5=5;

[0143] Update oxygen flow O f (2) = O f (1)+K d ·Δe(2)=15+0.1·5=15.5;

[0144] Continue this process and repeat the above steps and adjust the differential gain G d , if the response is slow, increase the differential gain G d (For example, adjust to 0.2), observe whether the response time is shortened and no oscillation is introduced, and finally determine the differential gain G d Optimal value.

[0145] Among them, the acceleration response gain G acc It is to speed up the response of the oxygen flow dynamic regulation model to significant error changes. Acceleration response gain G acc Ways to obtain include:

[0146] First, calculate the error change Δe(t)=e(t)-e(t-1);

[0147] Secondly, use the acceleration response gain G acc Update the current output of the oxygen flow dynamic regulation model:

[0148] Oxygen flow rate f (t) = O f (t-1)+K acc Δe(t);

[0149] Then, a rapidly changing input signal, the predicted oxygen demand, is applied. For example The oxygen flow rate suddenly changes from 20L / min to 45L / min, simulating a sudden pathological situation, and monitoring how the oxygen flow rate dynamic regulation model responds;

[0150] Then, observe the oxygen flow rate O output by the dynamic adjustment model under these rapid changes. f (t) changes, and whether the oxygen flow dynamic regulation model can quickly return to a new steady state;

[0151] Finally, using the real-time monitoring data, the acceleration response gain G is gradually adjusted. acc Specifically, a threshold η of the error change is set. When |Δe(t)>η, the acceleration response gain G is triggered. acc The adjustment formula is the acceleration response gain G acc (t) = G acc (t-1)+α·Δe(t); where G acc(t-1) represents the acceleration response gain value at time point t-1, which is the basis for continued feedback adjustment based on the previous oxygen flow dynamic regulation model state; α is the step size, which is a constant that determines the amplitude of each adjustment. If the α value is large, the adjustment speed of the acceleration response gain will be faster, and the oxygen flow dynamic regulation model will be more sensitive to error changes. If the α value is small, the adjustment speed of the acceleration response gain will be slower, and the oxygen flow dynamic regulation model will respond more smoothly to error changes, but will not respond enough to rapidly changing errors. For example, when the acceleration response gain G is set acc The initial value of is 0.1, and then try to increase or decrease it step by step, and record the dynamic adjustment error change according to the reaction of the oxygen flow dynamic adjustment model. And in order to prevent over-adjustment, set the acceleration response gain G acc The maximum and minimum values ​​of :

[0152] Example, assuming the setting condition is to predict oxygen demand Initial oxygen flow O f (0) = 25L / min, initial acceleration response gain G acc =0.1;

[0153] Time step 0:

[0154] Calculate the current error

[0155] Time step 1: Simulate the rapid change of set conditions

[0156] Calculate new error

[0157] Calculate the error change Δe(1)=e(1)-e(0)=15-5=10;

[0158] Update oxygen flow O f (1) = O f (0)+K acc Δe(1)=25+0.1·10=26;

[0159] Record and monitor the output of the oxygen flow dynamic regulation model. At this time, the oxygen flow O f (1) = 26 L / min;

[0160] Continue this process and repeat the above calculations in each subsequent time step and analyze the response of the oxygen flow dynamic regulation model to rapid changes.

[0161] By monitoring error changes, sudden change tests and establishing feedback loops, the acceleration response gain G can be effectively adjusted. acc, ensuring that the oxygen flow dynamic regulation model can respond quickly when faced with significant error changes, thereby ensuring the stability and effectiveness of the oxygen flow dynamic regulation model.

[0162] It is worth noting that the data processing unit also includes a parameter abnormality alarm module, which can f During the adjustment process, the above-mentioned multiple related parameters (i.e., the patient's physiological parameters) have complex interrelationships with the patient's oxygen demand. When making an estimate, if there are abnormal items in the relevant parameters, even if the calculated results can meet the requirements, it will lead to misjudgment of the patient's condition, which will lead to major clinical crises and decision-making errors. The parameter abnormality alarm module sets warning thresholds for the above-mentioned relevant parameters to predict oxygen demand. During calculation, when any parameter is abnormal, an early warning will be triggered to alert medical staff, thereby significantly improving patient safety and treatment outcomes.

[0163] Specifically, the warning method of the parameter abnormality alarm module includes: defining the warning range of relevant parameters, including:

[0164] Heart rate HR warning range [HR min ,HR max ], HR min is the minimum heart rate warning threshold, HR max is the maximum heart rate warning threshold;

[0165] SpO2 warning range min ,SpO2 max ], SpO2 min The minimum blood oxygen saturation warning threshold, SpO2 max It is the highest blood oxygen saturation warning threshold;

[0166] Respiratory rate R The warning range [f Rmin ,f Rmax ], f Rmin is the minimum respiratory rate warning threshold, f Rmax is the maximum respiratory rate warning threshold;

[0167] Tidal volume V T The warning range [V Tmin ,V Tmax ],V Tmin is the minimum tidal volume warning threshold, V Tmax is the maximum tidal volume warning threshold;

[0168] Carbon dioxide emission CO2 warning range (CO2 ≤ CO2 max ), CO2 maxA unilateral cap on maximum CO2 emissions;

[0169] Warning range of positive end-expiratory pressure (PEEP) min ,PEEP max ], PEEP min is the minimum positive end-expiratory pressure warning threshold, PEEP max It is the maximum positive end-expiratory pressure warning threshold;

[0170] Warning range of body movement intensity A (A≤A max ), A max It is the unilateral upper limit of the maximum body movement.

[0171] Define an alarm mechanism to check if each parameter exceeds the threshold range, such as Figure 2 As shown: Check the heart rate HR in turn to see if it exceeds the warning range. If it does, an alarm is issued. Otherwise, continue to check whether the blood oxygen saturation SpO2 exceeds the warning range. If it does, an alarm is issued. Otherwise, continue to check the respiratory rate f R Whether it exceeds the warning range, if yes, an alarm is issued, otherwise continue to detect the tidal volume V T Whether it exceeds the warning range, if yes, an alarm is issued; otherwise, continue to detect whether the carbon dioxide emissions CO2 exceed the unilateral upper limit, if yes, an alarm is issued; otherwise, continue to detect whether the positive end-expiratory pressure PEEP exceeds the warning range, if yes, an alarm is issued; otherwise, continue to detect whether the body movement intensity A exceeds the unilateral upper limit, if yes, an alarm is issued; otherwise, it indicates that all relevant parameters are normal.

[0172] By monitoring and obtaining the patient's physiological parameters in real time and comparing them with the preset thresholds, the abnormal conditions of the patient can be identified in time and the alarm can be triggered, providing automated and timely feedback to medical personnel, helping to quickly take necessary medical interventions and improve the safety and quality of care for patients. And through the comprehensive monitoring of multiple related parameters, the patient's physiological state can be more comprehensively evaluated, reducing the risk caused by the error of a single parameter.

[0173] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the protection scope of the technical solutions of the embodiments of the present invention.

Claims

1. An oxygen flow control system for a respiratory oxygen inhaler, comprising a data acquisition unit for obtaining relevant parameters for effective oxygen flow regulation, characterized in that: It also includes a data processing unit, which includes: The oxygen demand prediction calculation module is used to build an oxygen demand prediction model based on relevant parameters and calculate the predicted oxygen demand Oxygen flow calculation module, used to calculate oxygen flow according to relevant parameters and oxygen demand Construct an oxygen regulation model and calculate the oxygen regulation amount ΔO; Oxygen flow regulation module, based on predicted oxygen demand The oxygen flow rate is calculated by the oxygen adjustment amount ΔO PID control and accelerated response mechanism are introduced to control the oxygen flow rate O f Adjust speed for optimization; The parameter abnormality alarm module is used to define the warning range of relevant parameters and detect whether there are abnormal items in the relevant parameters. If there are abnormal items in the relevant parameters, an alarm will be issued.

2. The oxygen flow control system of a respiratory oxygen inhaler according to claim 1, characterized in that: The relevant parameters include: heart rate HR, physical activity intensity A, carbon dioxide emission CO2, blood oxygen saturation SpO2, respiratory rate f R , tidal volume V T , positive end-expiratory pressure (PEEP).

3. The oxygen flow control system of a respiratory oxygen inhaler according to claim 2, characterized in that: The method of constructing the oxygen demand prediction model includes: Defining and predicting oxygen requirements There are related parameters with nonlinear relationships. Based on the polynomial regression model, we get Y=β0+β1·X1+β2·X2+...+β n ·X k +β n+1 X1 2 +β n+2 X2 2 +...+β n+k ·X k 2 +...+ε; where Y is the dependent variable, i.e., the predicted oxygen demand β0 is the intercept of the model; X is the independent variable, i.e., the relevant parameter; k represents the number of independent variables; β1, β2, β3, …, β n Corresponding to the independent variable X k The regression coefficient of n represents the total number of regression coefficients; X1 2 , X2 2 , …, X k 2 is the square term of the independent variable X; ε is the error term; then the oxygen demand prediction model is evaluated to obtain the mean square error MSE, the optimal coefficient vector β, and the determination coefficient R of the oxygen demand prediction model 2 and the residual ε.

4. The oxygen flow control system of a respiratory oxygen inhaler according to claim 3, characterized in that: The method of evaluating the oxygen demand prediction model includes: Define the loss function as mean square error Where N is the total number of samples; is the model prediction value, that is, the predicted oxygen demand Y i is the true value, that is, the actual oxygen demand O d ; Solve the optimal coefficient vector β=(X matrix T X matrix ) -1 X matrix T Y vector Where X matrix is the independent variable matrix, Y vector is the dependent variable vector, X matrix T is the independent variable matrix X matrix The transpose of X matrix T X matrix represents the transpose of the independent variable matrix X matrix T and the independent variable matrix X matrix The product of itself, (X matrix T X matrix ) -1 Represents the matrix X matrix T X matrix The inverse matrix, X matrix T Y vector Represents the independent variable matrix X matrix The transpose of the dependent variable vector Y vector The product of Defining the coefficient of determination In the formula, the error sum of squares is Total sum of squares in is the mean of the dependent variable Y; Defining residuals ε i represents the residual of the ith sample.

5. The oxygen flow control system of a respiratory oxygen inhaler according to claim 4, characterized in that: The method of constructing the oxygen regulation model includes: Defining oxygen regulation ΔO=K adj ·(HR target -HR)+K PEEP ·(PEEP target -PEEP)+K SpO2 ·(SpO2 target -SpO2): In the formula, HR target The target heart rate that needs to be adjusted for the patient, K adj Indicates the patient's current heart rate HR and target heart rate HR target Adjustment factor for the difference between target K is the target positive end-expiratory pressure value that needs to be adjusted for the patient. PEEP Indicates the patient's current PEEP and target PEEP target Adjustment factor for the difference between target The target blood oxygen saturation that needs to be adjusted for the patient, K SpO2 Indicates the patient's current blood oxygen saturation SpO2 and target blood oxygen saturation SpO2 target The adjustment factor for the difference between .

6. The oxygen flow control system of a respiratory oxygen inhaler according to claim 1, characterized in that: The PID control and acceleration response mechanism optimizes the oxygen flow rate adjustment speed in the following ways: Introduce the acceleration response factor t, that is, time point t, and then define the dynamic regulation model of oxygen flow. Get the oxygen flow rate O at time t f (t), that is, the current flow value to be calculated; where G p is the proportional gain; e(t) represents the error at the current moment, and the calculation formula is G i is the integral gain; G d is the differential gain; G acc To accelerate the response gain.

7. The oxygen flow control system of a respiratory oxygen inhaler according to claim 6, characterized in that: The integral gain G i The method of obtaining includes: first setting G i =0 as the initial value, and then gradually increase the integral gain G i , observe the response of the oxygen flow dynamic regulation model at the same time, and record the integral gain G each time i Adjusted oxygen flow O f (t) and the error e(t) at the current moment, until the steady-state error of the oxygen flow dynamic regulation model Close to zero, determine the optimal integral gain G i ; The differential gain G d The acquisition method includes: first calculating the error change Δe(t) = e(t)-e(t-1); then using the differential gain G d Update the current output of the oxygen flow dynamic regulation model: oxygen flow O f (t) = O f (t-1)+K d Δe(t); then apply a changing input signal, i.e. predict oxygen demand And detect O f (t), and record the error e(t) and the error change Δe(t); then adjust the differential gain G d The value of the differential gain G is determined by adjusting the output fluctuation of the oxygen flow dynamic regulation model until it decreases and responds smoothly. d Optimal value; The acceleration response gain G acc The acquisition methods include: using the acceleration response gain G acc Update the current output of the oxygen flow dynamic regulation model: oxygen flow O f (t) = O f (t-1)+K acc Δe(t); then apply a rapidly changing input signal, i.e., predict oxygen demand Then observe and predict oxygen demand When the oxygen flow rate O changes f (t), and whether the oxygen flow dynamic regulation model can return to a new steady state; finally, using the real-time monitoring data, the acceleration response gain G is gradually adjusted. acc Specifically, a threshold η of the error change is set, and when |Δe(t)>η, the acceleration response gain G is triggered. acc The adjustment formula is G acc (t) = G acc (t-1)+α·Δe(t); where G acc (t-1) represents the acceleration response gain value at time point t-1; α is the step size.

8. The oxygen flow control system for a respiratory oxygen inhaler according to claim 1, characterized in that: The warning range of the defined relevant parameters includes: Heart rate HR warning range [HR min ,HR max ], HR min is the minimum heart rate warning threshold, HR max is the maximum heart rate warning threshold; SpO2 warning range min ,SpO2 max ], SpO2 min The minimum blood oxygen saturation warning threshold, SpO2 max It is the highest blood oxygen saturation warning threshold; Respiratory rate R The warning range [f Rmin ,f Rmax ], f Rmin is the minimum respiratory rate warning threshold, f Rmax is the maximum respiratory rate warning threshold; Tidal volume V T The warning range [V Tmin ,V Tmax ],V Tmin is the minimum tidal volume warning threshold, V Tmax is the maximum tidal volume warning threshold; Carbon dioxide emission CO2 warning range (CO2 ≤ CO2 max ), CO2 max A unilateral cap on maximum CO2 emissions; Warning range of positive end-expiratory pressure (PEEP) min ,PEEP max ], PEEP min is the minimum positive end-expiratory pressure warning threshold, PEEP max It is the maximum positive end-expiratory pressure warning threshold; Warning range of body movement intensity A (A≤A max ), A max It is the unilateral upper limit of the maximum body movement.

9. The oxygen flow control system of a respiratory oxygen inhaler according to claim 8, characterized in that: The method of detecting whether there are abnormal items in the relevant parameters includes: Check heart rate HR, blood oxygen saturation SpO2, respiratory rate f in turn R , tidal volume V T , carbon dioxide emissions CO2, positive end-expiratory pressure PEEP, and body movement intensity A, whether they exceed their warning range or unilateral upper limit. If they exceed their warning range or unilateral upper limit, they are abnormal items.

Images