A wireless urine oxygen partial pressure monitoring device and monitoring method

Abstract

This invention discloses a wireless urine oxygen partial pressure monitoring device and method. The monitoring device includes a multi-lumen catheter, a temperature sensor, a fiber optic oxygen sensor, a flow sensor, a main control circuit, a wireless data transceiver module, a smart terminal, and a urine bag. The monitoring method includes installing the monitoring device; data acquisition; establishing a prediction model and training and validating it; model prediction and result output. The monitoring device of this invention is external, eliminating the risk of damage to the patient's bladder. It can easily incorporate a temperature sensor to help determine whether there is kidney infection or other abnormalities through urine temperature. It also incorporates a flow sensor to monitor the kidney's urination capacity. Combined with the flow sensor, and based on the urine oxygen partial pressure index, multi-parameter comprehensive analysis can more accurately assess the degree and nature of kidney damage.

Description

Technical Field

[0001] This invention belongs to the field of medical device technology, and specifically relates to a wireless urine oxygen partial pressure monitoring device and monitoring method. Background Technology

[0002] Acute kidney injury (AKI) is a clinical syndrome characterized by a rapid decline in kidney function over a short period due to damage to the kidney's structure or function. Its pathogenesis is complex, and it has a high morbidity and mortality rate. Treatment for AKI mainly includes symptomatic and supportive care, as well as renal replacement therapy. Currently, there is no effective treatment to promote or accelerate kidney recovery. Existing methods aim to identify high-risk individuals for AKI and to develop treatment plans in the early stages to prevent progression to more severe stages. The clinical diagnosis of AKI typically uses the diagnostic criteria established in the Kidney Disease Improvement Global Outcomes (KDIGO) guidelines for acute kidney injury: a serum creatinine level ≥0.3 mg / dL (≥26.5 μmol / L) or more than 1.5 times the baseline value within 48 hours, with these conditions confirmed or inferred to have occurred within 7 days; or a persistent 6-hour urine output <0.5 mL / (kg·h), which can predict AKI risk. However, these two tests are time-consuming, often leading to a delay in diagnostic results. In addition, extracorporeal circulation of biomarkers can be established to assist in diagnosis, but this method requires expensive diagnostic equipment and is time-consuming.

[0003] Ischemia-reperfusion injury (IRI) is one of the main causes of acute kidney injury (AKI). IRI leads to hypoxia and oxidative stress in the donor kidney, which further triggers a series of cellular and tissue damages, such as lipid peroxidation of cell membranes, activation of inflammatory responses, and apoptosis. Studies have reported that when there is significant heterogeneity in kidney injury during reperfusion, the effects are most pronounced at the corticomena fold, and the limited tissue oxygen supply and high oxygen demand of the medulla are considered major reasons for the kidney's susceptibility to acute ischemic injury. In the absence of other confounding factors, the partial pressure of oxygen in urine is a good, almost immediate replacement for the partial pressure of oxygen in the renal medulla. This is based on the fact that the straight vessels are the periglomerular capillary network located within the medulla near the collecting ducts, and when urine is first excreted, the partial pressure of oxygen in urine is similar to that in the renal medulla.

[0004] Based on the above analysis, continuous urine oxygen partial pressure measurement can be used as a real-time monitoring method for renal hypoxia and injury risk. Real-time monitoring of oxygen in urine can help assess tissue oxygenation during donor kidney ischemia and reperfusion. There are reports of using polarographic electrodes placed in the bladder to measure urine oxygen partial pressure, thereby monitoring the risk of postoperative AKI in patients. However, implanting a detection probe in the bladder not only increases the patient's financial burden but also easily increases the risk of bladder injury. Furthermore, a single urine oxygen partial pressure indicator cannot comprehensively assess the health status of the kidneys. Based on the problems existing in the current technology, a wireless urine oxygen partial pressure monitoring device and method are needed. Summary of the Invention

[0005] This invention provides a wireless urine oxygen partial pressure monitoring device and method to solve the problems of high cost and risk of bladder damage when measuring urine oxygen partial pressure with polarographic electrodes placed in the bladder.

[0006] To solve the above problems, the technical solution provided by the present invention is as follows:

[0007] This invention provides a wireless urine oxygen partial pressure monitoring device, including a multi-lumen catheter (3), a temperature sensor (4), an optical fiber oxygen sensor (5), a flow sensor (6), a main control circuit (7), a wireless data transceiver module (8), a smart terminal (9), and a urine bag (10); the multi-lumen catheter (3) includes a urine storage chamber (31), and the urine storage chamber (31) is provided with a left connector (32), three top connectors (33), and a right connector (34); the left connector (32) is connected to the bladder of the patient (1) through a first catheter (21), and the right connector (34) is connected to the urine bag (10) through a second catheter (22);

[0008] One end of the temperature sensor (4) is connected to the urine storage chamber (31) via the first top connector (33), and the other end is connected to the main control circuit (7) via a data cable; the fiber optic oxygen sensor (5) includes a fiber optic oxygen sensor probe (51) and a Y-shaped fiber optic cable (53) connected via an SMA interface (52); the fiber optic oxygen sensor probe (51) is connected to the urine storage chamber (31) via the second top connector (33); the Y-shaped fiber optic cable (53) is connected to the main control circuit (7); one end of the flow sensor (6) is connected to the urine storage chamber (31) via the third top connector (33), and the other end is connected to the main control circuit (7) via a data cable; the wireless data transceiver module (8) is connected to the main control circuit (7) and also to the smart terminal (9), the smart terminal (9) including a terminal processor (91) and a user interface (92);

[0009] One branch of the Y-shaped optical fiber (53) is connected to a blue light signal transmitting port (54), and the other branch of the Y-shaped optical fiber (53) is connected to a fluorescent signal receiving port (55). The main control circuit (7) includes a filter (74), a photoelectric conversion circuit (75), a lock-in amplifier circuit (76), an analog-to-digital conversion circuit (77), an MCU chip (71), a signal generating circuit (72), and a blue LED (73) connected in sequence. The blue LED (73) is connected to the blue light signal transmitting port (54), and the filter (74) is connected to the fluorescent signal receiving port (55). The photoelectric conversion circuit (75) is connected to the MCU chip (71), and the MCU chip (71) is also connected to a wireless data transceiver circuit (78). The signal generating circuit (72) is also connected to the lock-in amplifier circuit (76).

[0010] In an optional embodiment of the present invention, the lock-in amplifier circuit (76) includes an AD630 modulation circuit (763), a low-pass filter circuit (764), and a DC bias circuit (765) connected in sequence; the DC bias circuit (765) is connected to the analog-to-digital converter circuit (77); the AD630 modulation circuit (763) is connected to the signal processing circuit (724) of the signal generation circuit (72), and the signal processing circuit (724) is also connected to the blue LED (73) and the MCU chip (71) respectively.

[0011] This invention also provides a wireless urine oxygen partial pressure monitoring method, implemented using a wireless urine oxygen partial pressure monitoring device as described in the above embodiments, comprising the following steps:

[0012] Step S1: Install the wireless urine oxygen partial pressure monitoring device, make the circuit connection, and then debug to ensure that the line is correct and the signal transmission is normal.

[0013] Step S2, data acquisition: After processing the sensor signals using the smart terminal (9), the urine oxygen partial pressure monitored by the fiber optic oxygen sensor (5), the urine temperature monitored by the temperature sensor (4), and the urine flow rate monitored by the flow sensor (6) can be directly read.

[0014] Step S3: Build a prediction model and train and validate it: Use the Logistic regression analysis method to build a big data analysis model for predicting the probability of AKI in patients, train and validate it so that it can analyze and process the patient data collected by the wireless urine oxygen partial pressure monitoring device and give the probability of patients developing AKI.

[0015] Step S4, Model Prediction and Result Output: The patient data collected in step S2 is sent in real time to the big data analysis and prediction model trained in step S3 to assess the current patient's risk of AKI, that is, to use the trained big data analysis and prediction model to predict the monitoring data.

[0016] In an optional embodiment of the present invention, in step S1, the bladder of the patient (1) is connected to the left connector (32) of the multi-lumen catheter (3) through the first catheter (21) to ensure that the bladder of the patient (1) is connected to the urine storage chamber (31), and then the other end of the multi-lumen catheter (3) is connected to the urine bag (10) through the second catheter (22);

[0017] Install a temperature sensor (4), an optical fiber oxygen sensor (5), and a flow sensor (6); connect the three to the main control circuit (7), integrate the wireless data transceiver module (8) on the main control circuit (7), and connect the wireless data transceiver module (8) to the debugging smart terminal (9) via wireless communication.

[0018] In an optional embodiment of the present invention, the fiber optic oxygen sensor (5) in step S2 utilizes the fluorescence quenching effect to detect changes in oxygen concentration. This process is described using the Stern-Volmer equation, as shown in formula (1):

[0019] In the formula: I0 is the initial intensity of the fluorescence signal emitted by the fiber optic oxygen-sensitive sensor probe (51) in an oxygen-free environment; I is the intensity of the fluorescence signal after the fluorescence quenching of the fiber optic oxygen-sensitive sensor probe (51); τ0 is the initial lifetime of the fluorescence signal emitted by the fiber optic oxygen-sensitive sensor probe (51) in an oxygen-free environment; τ is the lifetime of the fluorescence signal after the fluorescence quenching of the fiber optic oxygen-sensitive sensor probe (51); K is the Stern-Volmer constant; x9O2) is the oxygen concentration.

[0020] In an optional embodiment of the present invention, step S3 specifically includes the following steps:

[0021] Step S31, Pre-processing of data: Read and analyze the collected data such as the patient's urine oxygen partial pressure level, urine temperature, and urine flow rate, and remove abnormal data; manually enter the patient's age, gender, weight, medical history, AKI occurrence, and other data through a smart terminal; use the above data as a dataset, and use 80% of the patient data in the dataset as the training set and 20% of the patient data as the test set;

[0022] Step S32, Constructing the connection function: Use the sigmoid function to construct the connection function, mapping the binary variable of whether the patient has AKI to a continuous variable, and construct the fitting function f(z) for the multivariate ordered logistic regression model, the expression of which is as follows:

[0023] The calculated value of f9z) is the probability of the patient developing AKI calculated by the model. Z represents the situation of the patient developing AKI. When Z = 1, it means that the patient has developed AKI. When Z = 0, it means that the patient has not developed AKI. In equation (2), z is the multiple linear regression fitting function, and the expression is as follows: z = w1x1 + w2x2 + w3x3 + b (3); where: b is the intercept; x1, x2, and x3 are the patient's urine oxygen partial pressure data, urine temperature data, and urine flow rate data, respectively; w1, w2, and w3 are the corresponding independent variables x i The weights; let W = [w1, w2, w3], X = [x1, x2, x3], then we get Z = WX T +b;

[0024] Step S33, Derive the probability of AKI occurring as predicted by the model: The probability of AKI occurring as predicted by the model is derived by fitting the function f(z) of the multivariate ordered logistic regression model. The expression is as follows:

[0025]

[0026] Formula (4) represents the probability of AKI occurring as predicted by the model; Formula (5) represents the probability of AKI not occurring as predicted by the model;

[0027] According to formulas (3), (4), and (5), the likelihood term for the i-th sample point is obtained:

[0028]

[0029] The maximum likelihood function lnL(W,b) can then be expressed as:

[0030] Step S34, construct the cost function of the multivariate ordered logistic regression model: The cost function of the multivariate ordered logistic regression model is constructed using the cross-entropy loss function Lloss(W,b). The expression of the cross-entropy loss function is as follows: Let y represent the actual sample value, The cost function can be derived from the model's predicted output:

[0031] Step S35, construct the gradient descent algorithm to solve for the model parameters: The gradient descent (SGD) method is used to solve for the cross-entropy loss function. The gradient G of the cross-entropy loss function is expressed as follows:

[0032]

[0033]

[0034] The iterative formula for the stochastic gradient descent algorithm is:

[0035] W:=W-αG w (12); b:=b-αG b (13); where α is the learning rate, which controls the speed of gradient descent, and the convergence condition is set to the difference between two consecutive gradient calculations being less than 0.001;

[0036] The model parameters W and b are solved using gradient descent on the training set to obtain a big data analysis model for predicting the probability of AKI in patients.

[0037] Step S36, Model Validation: Input the data used as the test set into the model constructed in step S35, calculate the probability of the patient developing AKI, and analyze the model calculation results with the actual results to calculate the accuracy of the big data analysis model used to predict the probability of the patient developing AKI.

[0038] In an optional embodiment of the present invention, the abnormal data removed in step S31 includes: abnormal data with a urine oxygen partial pressure level greater than 250 mmHg or less than 1 mmHg, urine temperature higher than 45°C or lower than 25°C, and urine flow rate greater than 10 mL / min or less than 0.1 mL / min; after removal, the missing data is replaced with the average value of a urine oxygen partial pressure level of 80 mmHg, a urine temperature of 35°C, and a urine flow rate of 1 mL / min.

[0039] In an optional embodiment of the present invention, the accuracy rate in step S36 represents the accuracy of the prediction results of the established model, and the calculation formula is as follows:

[0040] In the formula: TP is the result where both the predicted result and the actual value are "positive", and "positive" means that the patient has AKI; TN is the result where both the predicted result and the actual value are "negative", and "negative" means that the patient has not AKI; FP is the result where the predicted result is "positive" and the actual value is "negative"; FN is the result where the predicted result is "negative" and the actual value is "positive".

[0041] In an optional embodiment of the present invention, in step S4, the threshold for the multivariate ordered logistic regression model is set to 0.5; when the prediction result is greater than 0.5, the patient is determined to have AKI; when the prediction result is less than 0.5, the patient is determined not to have AKI.

[0042] Furthermore, the model prediction results are classified: when the prediction result is less than 0.3, the model outputs a low-risk AKI warning; when the prediction result is between 0.3 and 0.6, the model outputs a medium-risk AKI warning; when the prediction result is greater than 0.6, the model outputs a high-risk AKI warning.

[0043] The model prediction results are sent to the user interface (92) of the smart terminal (9) in real time for the user to view. The user can view the patient's blood pressure, urine oxygen partial pressure level, urine temperature, urine flow rate, AKI risk prediction results, and prediction probability from the display interface.

[0044] Compared with the prior art, the embodiments of the present invention provide a wireless urine oxygen partial pressure monitoring device and its monitoring method, which have the following beneficial effects:

[0045] (1) The monitoring device of the present invention utilizes time-division multiplexing technology and uses AD630 and low-pass filter circuit to form a single-channel phase-locked amplifier structure. The MCU chip generates a high-precision quadrature reference signal in time division and sends it to the single-channel phase-locked amplifier circuit for modulation and operation with the input signal. This can effectively avoid the error caused by the inability of the dual-channel phase-locked amplifier structure to achieve complete parameter symmetry. After experimental testing, the phase accuracy of the input signal detection by the single-channel phase-locked amplifier structure using time-division multiplexing technology is improved by 30%, which can realize real-time continuous and accurate monitoring of the patient's urine oxygen partial pressure level. The device is small in size and easy to install. The monitoring data is sent to the smart terminal through the wireless data transceiver module, avoiding the disadvantages of complex wiring connections causing foreign body sensation and limited movement for the patient. It is suitable for monitoring the patient's urine oxygen partial pressure level in various situations.

[0046] Because the monitoring device is external, there is no need to consider the risk of damage to the patient's bladder. Temperature sensors can be easily incorporated to help determine whether there is infection or other abnormalities in the kidneys through urine temperature. Flow sensors can be incorporated to monitor the kidneys' urination capacity. Combined with urine oxygen partial pressure indicators, multi-parameter comprehensive analysis based on flow sensors can more accurately assess the degree and nature of kidney damage, thus providing a more reliable basis for clinical decision-making.

[0047] (2) The monitoring method of the present invention provides a big data analysis model for predicting the risk of AKI in patients based on logistic regression. This model adopts a multivariate ordered logistic regression model, with independent variables including the patient's urinary oxygen partial pressure level, urine temperature, and urine flow rate; the dependent variable is the risk of the patient developing AKI, which is divided into three categories: low risk, medium risk, and high risk. The model's independent variable data comes from the acquisition results of the wireless urinary oxygen partial pressure monitoring device and the collection of clinical patient information. This data is transmitted to the smart terminal through the wireless data transceiver module and stored in categories. The actual AKI occurrence of the subjects monitored by the wireless urinary oxygen partial pressure monitoring device is input by the smart terminal's data input function and stored in categories. This model can predict the probability of a patient developing AKI based on indicators such as the patient's urinary oxygen partial pressure level, urine flow rate, and urine temperature. The application of this model can help in the prevention and treatment of AKI in clinical patients. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in the embodiments or prior art, the drawings used in the description of the embodiments or prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 This is a schematic diagram of a wireless urine oxygen partial pressure monitoring device provided in an embodiment of this application.

[0050] Figure 2 This is a schematic diagram of a multi-lumen catheter structure in a wireless urine oxygen partial pressure monitoring device provided in an embodiment of this application.

[0051] Figure 3 The circuit connection block diagram of the phase modulator in a wireless urine oxygen partial pressure monitoring device provided in this application embodiment is shown.

[0052] Figure 4 This is a circuit connection block diagram of a lock-in amplifier in a wireless urine oxygen partial pressure monitoring device provided in an embodiment of this application.

[0053] Figure 5 The waveform diagram of the blue LED driving signal and the fluorescence voltage signal in a wireless urine oxygen partial pressure monitoring device provided in the embodiments of this application is shown.

[0054] Figure 6 This is a flowchart of a wireless urine oxygen partial pressure monitoring method provided in an embodiment of this application.

[0055] Figure 7 A linear relationship diagram between measured urine oxygen partial pressure data and the phase difference of fluorescence voltage signal of a wireless urine oxygen partial pressure monitoring device provided in this application embodiment.

[0056] Figure 8 This application provides a wireless urine oxygen partial pressure monitoring device for clinically measuring patients' urine oxygen partial pressure.

[0057] Figure 9 A flowchart illustrating the construction of a risk prediction model for AKI in a wireless urine oxygen partial pressure monitoring device provided in this application embodiment.

[0058] Figure 10 The diagram shows the result of parameter weight calculation for a patient AKI risk prediction model in a wireless urine oxygen partial pressure monitoring device provided in this application embodiment.

[0059] Figure 11 The graph shows the accuracy calculation results of a patient AKI risk prediction model in a wireless urine oxygen partial pressure monitoring device provided in this application embodiment.

[0060] Figure 12 The graph shows the output results of a logistic regression model of urine oxygen partial pressure and patient AKI risk in a wireless urine oxygen partial pressure monitoring device provided in this application embodiment for a patient AKI risk prediction model.

[0061] Figure 13 This is a schematic diagram of a user terminal interface for a wireless urine oxygen partial pressure monitoring device provided in an embodiment of this application.

[0062] Figure 14 This is a circuit diagram of a wireless urine oxygen partial pressure monitoring device provided in an embodiment of this application.

[0063] Figure 15 This is a circuit diagram of a wireless urine oxygen partial pressure monitoring device provided in an embodiment of this application.

[0064] Figure 16 This is a physical diagram of a wireless urine oxygen partial pressure monitoring device provided in an embodiment of this application.

[0065] The attached diagram is labeled as follows: 1-Patient; 2-Urinary catheter; 3-Multi-lumen urinary catheter; 31-Urine storage chamber; 32-Connector; 33-Threaded interface; 4-Temperature sensor; 5-Fiber optic oxygen sensor; 51-Fiber optic oxygen sensor probe; 52-SMA interface; 53-Y-type fiber optic cable; 54-Blue light signal transmitting port; 55-Fluorescence signal receiving port; 6-Flow sensor; 7-Main control circuit; 71-MCU chip; 72-Signal generation circuit; 721-PWM drive signal; 722-Reference signal X; 723-Reference... Signal Y; 724 - Signal processing circuit; 725 - Blue LED driving signal; 73 - Blue LED; 74 - Filter; 75 - Photoelectric conversion circuit; 76 - Lock-in amplifier circuit; 761 - Time-division reference signal; 762 - Fluorescent voltage signal; 763 - AD630 modulation circuit; 764 - Low-pass filter circuit; 765 - DC bias circuit; 77 - Analog-to-digital conversion circuit; 78 - Wireless data transceiver circuit; 8 - Wireless data transceiver module; 9 - Smart terminal; 91 - Terminal processor; 92 - User interface; 10 - Urine bag. Detailed Implementation

[0066] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, not all of them. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to represent selected embodiments of this application.

[0067] like Figure 1 , Figure 2 and Figure 3 As shown, this embodiment of the invention provides a wireless urine oxygen partial pressure monitoring device, including a multi-lumen catheter 3, a temperature sensor 4, a fiber optic oxygen sensor 5, a flow sensor 6, a main control circuit 7, a wireless data transceiver module 8, a smart terminal 9, and a urine bag 10. The multi-lumen catheter 3 includes a urine storage chamber 31, which is equipped with a left connector 32, three top connectors 33, and a right connector 34. The left connector 32 is connected to the bladder of the patient 1 through a first catheter 21, and the right connector 34 is connected to the urine bag 10 through a second catheter 22. All three top connectors 33 are threaded interfaces.

[0068] Temperature sensor 4 is connected at one end to urine storage chamber 31 via the first top connector 33, and at the other end to main control circuit 7 via a data cable. Its function is to monitor urine temperature and transmit the fluorescence signal generated by the probe of temperature sensor 4 to the receiver of the phase modulator / demodulator via the data cable. Fiber optic oxygen sensor 5 includes a fiber optic oxygen sensor probe 51 and a Y-shaped fiber optic cable 53 connected via an SMA interface 52. The fiber optic oxygen sensor probe 51 is connected to urine storage chamber 31 via the second top connector 33. The Y-shaped fiber optic cable 53 is connected to main control circuit 7 and its function is to monitor the partial pressure of urine oxygen near the bladder end of patient 1.

[0069] One end of the flow sensor 6 is connected to the urine storage chamber 31 via the third top connector 33, and the other end is connected to the main control circuit 7 via a data cable. Its function is to monitor urine flow and transmit the fluorescence signal generated by the flow sensor probe to the receiving end of the main control circuit via the data cable. The wireless data transceiver module 8 is connected to the main control circuit 7 and also to the smart terminal 9, used to send the collected urine oxygen partial pressure, urine temperature, and urine flow information to the smart terminal 9. The smart terminal 9 includes a terminal processor 91 and a user interface 92.

[0070] One branch of the Y-shaped optical fiber 53 is connected to a blue light signal transmission port 54, which is used to transmit the blue light excitation signal generated by the main control circuit 7 to the fiber optic oxygen-sensitive sensor probe 51. The other branch of the Y-shaped optical fiber 53 is connected to a fluorescence signal receiving port 55, which is used to transmit the fluorescence emission signal generated by the fiber optic oxygen-sensitive sensor probe 53 to the receiving end of the main control circuit.

[0071] like Figure 3 and Figure 4 As shown, the main control circuit 7 includes a filter 74, a photoelectric conversion circuit 75, a lock-in amplifier circuit 76, an analog-to-digital converter circuit 77, an MCU chip 71, a signal generation circuit 72, and a blue LED 73 connected in sequence. The blue LED 73 is connected to the blue light signal transmitting port 54, and the filter 74 is connected to the fluorescent signal receiving port 55. The photoelectric conversion circuit 75 is connected to the MCU chip 71, and the MCU chip 71 is also connected to a wireless data transceiver circuit 78. The signal generation circuit 72 is also connected to the lock-in amplifier circuit 76. The lock-in amplifier circuit 76 includes an AD630 modulation circuit 763, a low-pass filter circuit 764, and a DC bias circuit 765 connected in sequence. The DC bias circuit 765 is connected to the analog-to-digital converter circuit 77. The AD630 modulation circuit 763 is connected to the signal processing circuit 724 of the signal generation circuit 72, and the signal processing circuit 724 is also connected to the blue LED 73 and the MCU chip 71 respectively.

[0072] The specific signal flow of a wireless urine oxygen partial pressure monitoring device is as follows:

[0073] The MCU chip 71 generates a PWM drive signal 721, which, after biasing processing by the signal generation circuit 72, generates a blue LED drive signal 725 and is transmitted to the blue LED 73. The blue LED 73 generates a blue excitation signal, which is transmitted sequentially through the blue signal emission port 54 and one branch of the Y-type optical fiber 53 to the fiber optic oxygen-sensitive sensor 51, causing the fiber optic oxygen-sensitive sensor to generate a fluorescence emission signal. This fluorescence emission signal is then transmitted sequentially through the other branch of the Y-type optical fiber 53 and the fluorescence signal receiving port 55 to the filter 74. After filtering out the blue excitation signal, the fluorescence emission signal is transmitted to the photoelectric conversion circuit 75, which converts the fluorescence emission signal into a fluorescence voltage signal 762. This fluorescence voltage signal 762 is then sent to the lock-in amplifier circuit 76. At this time, two orthogonal reference signals generated by the control signal generation circuit of the MCU chip 71 are also input in parallel with the fluorescence voltage signal, and the two signals are modulated. The voltage waveforms of the blue LED drive signal 725 and the fluorescence voltage signal 762 are as follows: Figure 5 As shown, the frequencies of the blue LED driving signal 725 and the fluorescent voltage signal 762 are both 40kHz. Finally, the processed signal is converted into a digital signal by the analog-to-digital converter circuit 77 and sent to the MCU chip 71 for data processing. The processing result is sent to the wireless data transceiver module 8 through the wireless data transceiver circuit 78, and finally sent to the terminal processor 91 for data analysis and displayed through the user interface 92.

[0074] To clarify the specific signal processing process in the lock-in amplifier circuit 76, its structure and signal flow are described in detail: The structure of the lock-in amplifier circuit 76 is as follows: Figure 4 As shown, it includes: AD630 modulation circuit 763, low-pass filter circuit 764, DC bias circuit 765, and the signal flow is as follows:

[0075] MCU chip 71 generates a PWM drive signal 721, which is biased by signal processing circuit 724 to obtain a blue LED drive signal 725. This blue LED drive signal is sent to blue LED 73 to generate a blue light excitation signal. At the same time, MCU chip 71 generates orthogonal reference signals X 722 and Y 723 in a time-division manner, which are biased by signal processing circuit 724 respectively. The processed time-division reference signal 761 is sent to AD630 modulation circuit 763 in a time-division manner. The signal flow of blue LED 73 has been described previously and will not be repeated. In AD630 modulation circuit 763, the time-division reference signal 761 is modulated with the fluorescence voltage signal 762 finally input by photoelectric conversion circuit 75. The result is sent to low-pass filter circuit 764 for filtering. The filtered DC signal is forward biased by DC bias circuit 765 and then sent to analog-to-digital conversion circuit 77.

[0076] like Figure 6 As shown, this embodiment of the invention also provides a wireless urine oxygen partial pressure monitoring method, implemented using a wireless urine oxygen partial pressure monitoring device as described in the above embodiment, comprising the following steps:

[0077] Step S1: Install the wireless urine oxygen partial pressure monitoring device, make the circuit connection, and then debug to ensure that the line is correct and the signal transmission is normal.

[0078] Step S2, Data Acquisition: After processing the sensor signals using the smart terminal 9, the urine oxygen partial pressure monitored by the fiber optic oxygen sensor 5, the urine temperature monitored by the temperature sensor 4, and the urine flow rate monitored by the flow sensor 6 can be directly read.

[0079] Step S3: Build a prediction model and train and validate it: Use the Logistic regression analysis method to build a big data analysis model for predicting the probability of AKI in patients, train and validate it so that it can analyze and process the patient data collected by the wireless urine oxygen partial pressure monitoring device and give the probability of patients developing AKI.

[0080] Step S4, Model Prediction and Result Output: The patient data collected in step S2 is sent in real time to the big data analysis and prediction model trained in step S3 to assess the current patient's risk of AKI, that is, to use the trained big data analysis and prediction model to predict the monitoring data.

[0081] In step S1, the patient's bladder is connected to the left connector 32 of the multi-lumen catheter 3 via the first catheter 21, ensuring communication between the patient's bladder and the urine storage chamber 31. The other end of the multi-lumen catheter 3 is then connected to the urine bag 10 via the second catheter 22. A temperature sensor 4, a fiber optic oxygen sensor 5, and a flow sensor 6 are installed and connected to the main control circuit 7. A wireless data transceiver module 8 is integrated into the main control circuit 7 and connects to the debugging smart terminal 9 via wireless communication. After all equipment is installed and debugged, and the wiring is confirmed to be correct and the signal transmission is normal, the monitoring phase begins.

[0082] In step S2, the fiber optic oxygen sensor 5 uses the fluorescence quenching effect to detect changes in oxygen concentration. This process is described by the Stern-Volmer equation, as shown in formula (1):

[0083] In the formula: I0 is the initial intensity of the fluorescence signal emitted by the fiber optic oxygen-sensitive sensor 51 in an oxygen-free environment; I is the intensity of the fluorescence signal after the fluorescence quenching of the fiber optic oxygen-sensitive sensor 51; τ0 is the initial lifetime of the fluorescence signal emitted by the fiber optic oxygen-sensitive sensor 51 in an oxygen-free environment; τ is the lifetime of the fluorescence signal after the fluorescence quenching of the fiber optic oxygen-sensitive sensor 51; K is the Stern-Volmer constant; x(O2) is the oxygen concentration.

[0084] The phase value of the fluorescence signal emitted by the fiber optic oxygen-sensitive sensor 51 is different under different oxygen concentrations. The phase change of the fluorescence signal emitted by the fiber optic oxygen-sensitive sensor 51 is measured based on phase-sensitive detection technology. After the sensor signal is processed by the terminal data processing system, the oxygen concentration of the urine to be tested can be directly output.

[0085] Figure 7 The graph shows a linear fit between the measured urine oxygen partial pressure data and the fluorescence signal phase. The fitting results indicate that the linear correlation between the urine oxygen partial pressure and the fluorescence signal phase measured by the urine oxygen partial pressure detection device described in this invention is greater than 98.9%, enabling accurate measurement of urine oxygen partial pressure. Some clinical test sample data from patients are shown below. Figure 8 As shown.

[0086] The flowchart for building a big data analytics model is as follows: Figure 9 As shown, step S3 specifically includes the following steps:

[0087] Step S31, Pre-processing of data: Read and analyze the collected data such as the patient's urine oxygen partial pressure level, urine temperature, and urine flow rate, and remove abnormal data; manually enter the patient's age, gender, weight, medical history, AKI occurrence, and other data through a smart terminal; use the above data as a dataset, and use 80% of the patient data in the dataset as the training set and 20% of the patient data as the test set;

[0088] Step S32, Constructing the connection function: Use the sigmoid function to construct the connection function, mapping the binary variable of whether the patient has AKI to a continuous variable, and construct the fitting function f(z) for the multivariate ordered logistic regression model, the expression of which is as follows:

[0089] The calculated value of f(z) is the probability of the patient developing AKI calculated by the model. Z represents the situation of the patient developing AKI. When Z = 1, it means that the patient has developed AKI. When Z = 0, it means that the patient has not developed AKI. In formula (2), z is the multiple linear regression fitting function, and the expression is as follows: z = w1x1 + w2x2 + w3x3 + b (3); where: b is the intercept; x1, x2, and x3 are the patient's urine oxygen partial pressure data, urine temperature data, and urine flow rate data, respectively; w1, w2, and w3 are the corresponding independent variables x iThe weights; let W = [w1, w2, w3], X = [x1, x2, x3], then we get Z = WX T +b;

[0090] Step S33, Deriving the probability of AKI occurring as predicted by the model: The probability of AKI occurring as predicted by the model is derived through the fitting function f(z) of the multivariate ordered logistic regression model, and its expression is as follows:

[0091]

[0092] Formula (4) represents the probability of AKI occurring as predicted by the model; Formula (5) represents the probability of AKI not occurring as predicted by the model;

[0093] According to formulas (3), (4), and (5), the likelihood term for the i-th sample point is obtained:

[0094]

[0095] The maximum likelihood function lnL(W,b) can then be expressed as:

[0096] Step S34, construct the cost function of the multivariate ordered logistic regression model: The cost function of the multivariate ordered logistic regression model is constructed using the cross-entropy loss function Lloss(W,b). The expression of the cross-entropy loss function is as follows: Let y represent the actual sample value, The cost function can be derived from the model's predicted output:

[0097] Step S35, construct the gradient descent algorithm to solve for the model parameters: The gradient descent (SGD) method is used to solve for the cross-entropy loss function. The gradient G of the cross-entropy loss function is expressed as follows:

[0098]

[0099] The iterative formula for the stochastic gradient descent algorithm is:

[0100] W:=W-αG w (12); b:=b-αG b (13); where α is the learning rate, which controls the speed of gradient descent, and the convergence condition is set to the difference between two consecutive gradient calculations being less than 0.001;

[0101] The model parameters W and b were solved using gradient descent on the training set to obtain a big data analysis model for predicting the probability of AKI in patients; the calculation results of the parameter weights of this big data analysis model are as follows. Figure 10 As shown, urinary oxygen partial pressure has the greatest impact on whether a patient develops AKI. The logistic regression model output of urinary oxygen partial pressure and the risk of AKI in patients is as follows: Figure 11 As shown.

[0102] Step S36, Model Validation: Input the data used as the test set into the model constructed in step S35, calculate the probability of the patient developing AKI, and analyze the model calculation results with the actual results to calculate the accuracy of the big data analysis model used to predict the probability of the patient developing AKI.

[0103] The abnormal data removed in step S31 includes: urine oxygen partial pressure level greater than 250 mmHg or less than 1 mmHg, urine temperature higher than 45℃ or lower than 25℃, and urine flow rate greater than 10 mL / min or less than 0.1 mL / min. After removal, the missing data is replaced with the average value of urine oxygen partial pressure level of 80 mmHg, urine temperature of 35℃, and urine flow rate of 1 mL / min.

[0104] In step S36, the accuracy rate represents the accuracy of the prediction results of the established model, and the calculation formula is as follows:

[0105] In the formula: TP is the result where both the predicted result and the actual value are "positive", and "positive" means that the patient has AKI; TN is the result where both the predicted result and the actual value are "negative", and "negative" means that the patient has not AKI; FP is the result where the predicted result is "positive" and the actual value is "negative"; FN is the result where the predicted result is "negative" and the actual value is "positive".

[0106] The accuracy of the trained big data analysis and prediction model was calculated using a test set. In this embodiment, the initial accuracy of the model test was 87.71%. On the one hand, this verifies that using the model can provide an auxiliary means for the early prevention of AKI; on the other hand, the 87.71% figure in this embodiment is only the accuracy of the model in its initial test. As the amount of patient data monitored in later applications continues to increase, the various parameters of the model will be corrected and optimized, and the prediction accuracy will be further improved.

[0107] In step S4, the threshold for the multivariate ordered logistic regression model is set to 0.5; when the prediction result is greater than 0.5, the patient is determined to have AKI; when the prediction result is less than 0.5, the patient is determined not to have AKI.

[0108] Furthermore, the model prediction results are classified: when the prediction result is less than 0.3, the model outputs a low-risk AKI warning; when the prediction result is between 0.3 and 0.6, the model outputs a medium-risk AKI warning; when the prediction result is greater than 0.6, the model outputs a high-risk AKI warning.

[0109] The model prediction results are sent to the user interface 92 of the smart terminal 9 in real time for the user to view. The user can view information such as the patient's blood pressure, urine oxygen partial pressure level, urine temperature, urine flow rate, AKI risk prediction results, and prediction probability from the display interface.

[0110] Figure 12 The graph shows the output results of a logistic regression model of urine oxygen partial pressure and patient AKI risk in a wireless urine oxygen partial pressure monitoring device provided in this application embodiment for a patient AKI risk prediction model. Figure 13 This is a schematic diagram of a user terminal interface for a wireless urine oxygen partial pressure monitoring device provided in an embodiment of this application. Figure 14 This is a circuit diagram of a wireless urine oxygen partial pressure monitoring device provided in an embodiment of this application. Figure 15 This is a circuit diagram of a wireless urine oxygen partial pressure monitoring device provided in an embodiment of this application. Figure 16 This is a physical diagram of a wireless urine oxygen partial pressure monitoring device provided in an embodiment of this application.

[0111] In summary, although the present invention has been disclosed above with reference to preferred embodiments, the above preferred embodiments are not intended to limit the present invention. Those skilled in the art can make various modifications and refinements without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention shall be determined by the scope defined in the claims.

Claims

1. A wireless urine oxygen partial pressure monitoring device, characterized in that, The device includes a multi-lumen catheter (3), a temperature sensor (4), a fiber optic oxygen sensor (5), a flow sensor (6), a main control circuit (7), a wireless data transceiver module (8), a smart terminal (9), and a urine bag (10). The multi-lumen catheter (3) includes a urine storage chamber (31), which is provided with a left connector (32), three top connectors (33), and a right connector (34). The left connector (32) is connected to the bladder of the patient (1) through a first catheter (21), and the right connector (34) is connected to the urine bag (10) through a second catheter (22). One end of the temperature sensor (4) is connected to the urine storage chamber (31) via the first top connector (33), and the other end is connected to the main control circuit (7) via a data cable; the fiber optic oxygen sensor (5) includes a fiber optic oxygen sensor probe (51) and a Y-shaped fiber optic cable (53) connected via an SMA interface (52); the fiber optic oxygen sensor probe (51) is connected to the urine storage chamber (31) via the second top connector (33); the Y-shaped fiber optic cable (53) is connected to the main control circuit (7); one end of the flow sensor (6) is connected to the urine storage chamber (31) via the third top connector (33), and the other end is connected to the main control circuit (7) via a data cable; the wireless data transceiver module (8) is connected to the main control circuit (7) and also to the smart terminal (9), the smart terminal (9) including a terminal processor (91) and a user interface (92); One branch of the Y-shaped optical fiber (53) is connected to a blue light signal transmitting port (54), and the other branch of the Y-shaped optical fiber (53) is connected to a fluorescent signal receiving port (55). The main control circuit (7) includes a filter (74), a photoelectric conversion circuit (75), a lock-in amplifier circuit (76), an analog-to-digital conversion circuit (77), an MCU chip (71), a signal generating circuit (72), and a blue LED (73) connected in sequence. The blue LED (73) is connected to the blue light signal transmitting port (54), and the filter (74) is connected to the fluorescent signal receiving port (55). The photoelectric conversion circuit (75) is connected to the MCU chip (71), and the MCU chip (71) is also connected to a wireless data transceiver circuit (78). The signal generating circuit (72) is also connected to the lock-in amplifier circuit (76).

2. The wireless urine oxygen partial pressure monitoring device according to claim 1, characterized in that, The lock-in amplifier circuit (76) includes an AD630 modulation circuit (763), a low-pass filter circuit (764), and a DC bias circuit (765) connected in sequence; the DC bias circuit (765) is connected to the analog-to-digital converter circuit (77); the AD630 modulation circuit (763) is connected to the signal processing circuit (724) of the signal generation circuit (72), and the signal processing circuit (724) is also connected to the blue LED (73) and the MCU chip (71) respectively.

3. A wireless method for monitoring urine oxygen partial pressure, implemented using a wireless urine oxygen partial pressure monitoring device as described in any one of claims 1-2, characterized in that, Includes the following steps: Step S1: Install the wireless urine oxygen partial pressure monitoring device, make the circuit connection, and then debug to ensure that the line is correct and the signal transmission is normal. Step S2, data acquisition: After processing the sensor signals using the smart terminal (9), the urine oxygen partial pressure monitored by the fiber optic oxygen sensor (5), the urine temperature monitored by the temperature sensor (4), and the urine flow rate monitored by the flow sensor (6) can be directly read. Step S3: Build a prediction model and train and validate it: Use the Logistic regression analysis method to build a big data analysis model for predicting the probability of AKI in patients, train and validate it so that it can analyze and process the patient data collected by the wireless urine oxygen partial pressure monitoring device and give the probability of patients developing AKI. Step S4, Model Prediction and Result Output: The patient data collected in step S2 is sent in real time to the big data analysis and prediction model trained in step S3 to assess the current patient's risk of AKI, that is, to use the trained big data analysis and prediction model to predict the monitoring data.

4. The wireless urine oxygen partial pressure monitoring method according to claim 3, characterized in that, In step S1, the patient's (1) bladder is connected to the left connector (32) of the multi-lumen catheter (3) through the first catheter (21) to ensure that the patient's (1) bladder is connected to the urine storage chamber (31). Then, the other end of the multi-lumen catheter (3) is connected to the urine bag (10) through the second catheter (22). Install a temperature sensor (4), an optical fiber oxygen sensor (5), and a flow sensor (6); connect the three to the main control circuit (7), integrate the wireless data transceiver module (8) on the main control circuit (7), and connect the wireless data transceiver module (8) to the debugging smart terminal (9) via wireless communication.

5. The wireless urine oxygen partial pressure monitoring method according to claim 3, characterized in that, In step S2, the fiber optic oxygen sensor (5) uses the fluorescence quenching effect to detect changes in oxygen concentration. This process is described by the Stern-Volmer equation, as shown in formula (1): (1); where: The initial intensity of the fluorescence signal emitted by the fiber optic oxygen-sensitive sensor probe (51) in an oxygen-free environment; I represents the fluorescence signal intensity after fluorescence quenching occurs in the fiber optic oxygen-sensitive sensor probe (51); τ is the initial lifetime of the fluorescence signal emitted by the fiber optic oxygen-sensitive sensor probe (51) in an oxygen-free environment; τ is the lifetime of the fluorescence signal after the fluorescence quenching of the fiber optic oxygen-sensitive sensor probe (51); K is the Stern-Volmer constant. This represents the oxygen concentration.

6. The wireless urine oxygen partial pressure monitoring method according to claim 3, characterized in that, Step S3 specifically includes the following steps: Step S31, Data preprocessing: Read and analyze the collected patient urine oxygen partial pressure level, urine temperature, urine flow rate and other data, and remove abnormal data; Patients' age, gender, weight, medical history, and AKI occurrence were manually entered through smart terminals; this data was used as a dataset, with 80% of the patient data in the dataset used as the training set and 20% of the patient data used as the test set. Step S32, Constructing the connection function: Use the sigmoid function to construct a connection function, mapping the binary variable of whether the patient has AKI to a continuous variable, and constructing a multivariate ordered logistic regression model fitting function. Its expression is as follows: (2); The calculated value is the probability of the patient developing AKI calculated by the model, and Z represents the situation of the patient developing AKI. When Z=1, it means that the patient has developed AKI, and when Z=0, it means that the patient has not developed AKI. In equation (2) The function for fitting multiple linear regression is expressed as follows: (3); where: The intercept; , , These are the patient's urine oxygen partial pressure data, urine temperature data, and urine flow rate data, respectively. , , For the corresponding independent variable The weights; let W = [ , , ],X=[ , , ], thus obtaining Z=WX T + b; Step S33, derive the probability of AKI occurring using the model: fit the function using a multivariate ordered logistic regression model. The probability of AKI occurring, as predicted by this model, is derived as follows: (4)； (5)； Formula (4) represents the probability of AKI occurring as predicted by the model; Formula (5) represents the probability of AKI not occurring as predicted by the model; According to formulas (3), (4), and (5), the likelihood term for the i-th sample point is obtained: (6)； Then the maximum likelihood function It can be represented as: (7); Step S34, construct the cost function for the multivariate ordered logistic regression model: using the cross-entropy loss function. The cost function for constructing the multivariate ordered logistic regression model, the cross-entropy loss function, is expressed as follows: (8); Let y represent the actual sample value, The cost function can be derived from the model's predicted output: (9); Step S35, construct the gradient descent algorithm to solve for the model parameters: The gradient descent (SGD) method is used to solve for the cross-entropy loss function. The gradient G of the cross-entropy loss function is expressed as follows: (10)； (11)； The iterative formula for the stochastic gradient descent algorithm is: (12); (13); where: To control the learning rate and the speed of gradient descent, the convergence condition is set to the difference between two consecutive gradient calculations being less than 0.

001. The model parameters W and b are solved using gradient descent on the training set to obtain a big data analysis model for predicting the probability of AKI in patients. Step S36, Model Validation: Input the data used as the test set into the model constructed in step S35, calculate the probability of the patient developing AKI, and analyze the model calculation results with the actual results to calculate the accuracy of the big data analysis model used to predict the probability of the patient developing AKI.

7. The wireless urine oxygen partial pressure monitoring method according to claim 3, characterized in that, The abnormal data removed in step S31 includes: urine oxygen partial pressure level greater than 250 mmHg or less than 1 mmHg, urine temperature higher than 45℃ or lower than 25℃, and urine flow rate greater than 10 mL / min or less than 0.1 mL / min. After removal, the missing data is replaced with the average value of urine oxygen partial pressure level of 80 mmHg, urine temperature of 35℃, and urine flow rate of 1 mL / min.

8. A wireless urine oxygen partial pressure monitoring method according to claim 6, characterized in that, In step S36, the accuracy rate represents the accuracy of the prediction results of the established model, and the calculation formula is as follows: （14）； In the formula: TP is the result where both the predicted result and the actual value are "positive", and "positive" means that the patient has AKI; TN is the result where both the predicted result and the actual value are "negative", and "negative" means that the patient has not AKI; FP is the result where the predicted result is "positive" and the actual value is "negative"; FN is the result where the predicted result is "negative" and the actual value is "positive".

9. A wireless urine oxygen partial pressure monitoring method according to claim 6, characterized in that, In step S4, the threshold for the multivariate ordered logistic regression model is set to 0.5; when the prediction result is greater than 0.5, the patient is determined to have AKI; when the prediction result is less than 0.5, the patient is determined not to have AKI. Furthermore, the model prediction results are classified: when the prediction result is less than 0.3, the model outputs a low-risk AKI warning; when the prediction result is between 0.3 and 0.6, the model outputs a medium-risk AKI warning. When the prediction result is greater than 0.6, the model outputs a high-risk AKI warning; The model prediction results are sent to the user interface (92) of the smart terminal (9) in real time for the user to view. The user can view the patient's blood pressure, urine oxygen partial pressure level, urine temperature, urine flow rate, AKI risk prediction results, and prediction probability from the display interface.

Images