Post-thoracic-operation leakage monitoring system and treatment method

Through the combination of micro sensor networks and machine learning algorithms, we can monitor and analyze changes in the chest fluid in real time and dynamically adjust the automation device, solving the subjective deviation, information omission and data isolation problems of the existing system, and achieving efficient and personalized leakage The fluid monitoring and treatment plan ensures data security and information sharing, and improves the ability to respond to complex diseases.

CN118986314AInactive Publication Date: 2024-11-22FUWAI CENTRAL CHINA CARDIOVASCULAR HOSPITAL
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202411096237.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-12
Publication Date
2024-11-22
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing post-thoracic fluid leakage monitoring system relies on physical observation and periodic measurement, has subjective deviations and information omissions, cannot capture dynamic changes in real time, and lacks the ability to comprehensively analyze multi-parameter data, resulting in clinical decision-making errors, system maintenance and calibration. High requirements and high economic costs, intelligent data analysis and prediction cannot be carried out, data isolation hinders information sharing, patient privacy and security are difficult to guarantee, and lack of ability to cope with complex diseases.

Method used

Using a micro sensor network to monitor the changes in the chest cavity in real time, transmit data wirelessly to the central processing system, analyze data using machine learning algorithms, predict leakage trends and dynamically adjust automation devices, integrate intelligent catheter system and micro pumps to realize personalized monitoring solutions and detailed reports to ensure safe and efficient sharing of data.

Benefits of technology

Real-time monitoring of key indicators in the chest cavity is achieved, the accuracy of prediction of leakage changes is improved and treatment targeted, the dependence on manual operations is reduced, the treatment efficiency and patient satisfaction is improved, data security and information sharing are enhanced, Ability to cope with complex diseases.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118986314A_ABST
    Figure CN118986314A_ABST
Patent Text Reader

Abstract

The invention relates to a chest post-operation leakage monitoring system and a treatment method. The set of micro sensor network is used for monitoring the liquid accumulation amount in the thoracic cavity in real time, including the pH value, the liquid property of the ion concentration, the temperature and the pressure change, and sending the data to the central processing system in a wireless mode; a machine learning algorithm is built in the central processing system, and the central processing system is used for analyzing sensor data, predicting the liquid leakage trend and risk and dynamically adjusting the automation device according to an analysis result; wherein the central processing system performs preprocessing and feature extraction on the collected data, and analyzes the frequency characteristics of the data through Fourier transform; and finally, generating a personalized monitoring scheme according to individual differences of the patients, and generating a postoperative monitoring report. A gate control loop unit GRU based on a machine learning algorithm model is adopted, the time sequence of input data is processed through the internal state of the GRU, and leakage processing method tasks are predicted and classified.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a postoperative fluid leakage monitoring system and a processing method, in particular to a postoperative fluid leakage monitoring system and a processing method for thoracic cavity. Background Art

[0002] At present, the postoperative thoracic fluid leakage monitoring system and treatment method have played an important role in clinical practice, but there are still some shortcomings and disadvantages. First, the traditional monitoring method mainly relies on physical observation and periodic measurement. This method relies on the experience of medical staff, is prone to subjective bias, and is difficult to capture the subtle changes of the dynamic changes of fluid leakage in real time. In the traditional treatment method, fixed time intervals are used for inspection and recording, which will lead to the omission of important information, especially when the amount of leakage changes rapidly or the patient's condition changes suddenly, and timely warning and intervention cannot be achieved. In addition, traditional systems mostly rely on simple physical indicators, including drainage volume and volume, which cannot fully reflect the complex physiological state changes in the patient's body, resulting in clinical decision-making errors. In addition, traditional systems lack the ability to comprehensively analyze multi-parameter data, while modern medicine emphasizes personalized and precise treatment, and requires multi-dimensional analysis of patients' physiological data. Furthermore, many postoperative thoracic fluid leakage monitoring systems are technically complex and cumbersome to operate in actual applications, especially for medical staff who are not skilled enough, which increases the workload and the probability of error. Dependence on the system leads to reduced manual intervention, and excessive reliance on system results ignores clinical experience judgment.

[0003] At the same time, many monitoring devices have high maintenance and calibration requirements. In actual operation, they will face problems such as equipment aging and unstable data transmission, which will affect the accuracy and real-time performance of monitoring. Economic cost is also an issue that cannot be ignored. Many advanced monitoring systems and processing methods involve expensive equipment and complex maintenance, which is a significant obstacle for some medical institutions with limited resources. In addition, due to the limitations of technology and data analysis capabilities, many existing monitoring systems are unable to perform intelligent data analysis and prediction, and can only respond after obvious symptoms appear. This passive processing mode is particularly insufficient when dealing with emergencies. Furthermore, it is difficult for existing systems to integrate seamlessly with other medical systems, and the isolation of data leads to the inability to efficiently share and utilize information. During the postoperative recovery process, patients often involve cooperation from multiple disciplines and departments, and information isolation hinders the efficiency and quality of collaborative work. The inability to effectively integrate and analyze monitoring data with medical records and drug records in real time leads to a lack of targetedness and dynamic adjustment capabilities for treatment plans. Another key issue is patient privacy and data security. In the process of large-scale data collection and storage, how to ensure the security and privacy of patient information has become an important challenge. Data leakage or abuse not only harms individual patients, but also has a negative impact on the reputation and trust of medical institutions. Finally, the existing system is still insufficient in dealing with complex conditions. Complex conditions are often accompanied by the interaction of multiple factors, and traditional monitoring methods are often difficult to effectively analyze in multiple dimensions at the same time. This makes the system lack flexibility and accuracy when dealing with postoperative complications, thus affecting the timeliness and effectiveness of treatment. In particular, the shortcomings of traditional systems are particularly obvious in terms of early identification and intervention of the trend of worsening conditions. Summary of the invention

[0004] The purpose of the present invention is to provide a post-thoracic fluid leakage monitoring system and a processing method, so as to solve some of the disadvantages and shortcomings pointed out in the background technology.

[0005] The present invention solves the above-mentioned technical problems by adopting the following technical solutions:

[0006] A network of micro-sensors that monitors the amount of fluid accumulated in the chest cavity, fluid properties including pH, ion concentration, temperature and pressure changes in real time, and sends the data wirelessly to a central processing system;

[0007] The central processing system has a built-in machine learning algorithm for analyzing sensor data, predicting leakage trends and risks, and dynamically adjusting the automation device according to the analysis results; wherein the central processing system preprocesses and extracts features from the collected data, including statistical features of maximum value, minimum value, mean value, standard deviation, time window features including data differences and change rates, and frequency characteristics of the data analyzed by Fourier transform;

[0008] The automated device drains abnormal fluid accumulation and adjusts internal pressure through a micro pump and catheter system;

[0009] The micro pump is used to monitor the pressure and flow volume in the chest cavity in real time, and dynamically adjust the pump speed and pump pressure according to the sensor data;

[0010] The second is the intelligent regulation of the catheter system, which is designed with an adaptive switch valve that automatically opens or closes according to the properties and amount of the liquid, and optimizes the opening and closing frequency and amplitude of the valve through an algorithm;

[0011] Finally, a personalized monitoring plan is generated based on individual patient differences, and a detailed postoperative monitoring report is generated, including data analysis, risk assessment, and recommended intervention measures, to provide high-precision monitoring and efficient personalized treatment.

[0012] Furthermore, the central processing system construction process includes: first, implementing a data preprocessing module to correct errors and drifts in sensor data, using the function:

[0013]

[0014] Where x, y, and z represent different types of data collected from sensors, including temperature, pressure, and liquid accumulation, and a and b are coefficients that are dynamically adjusted based on the characteristics of the data;

[0015] Next, calculate the liquid accumulation and pressure statistics in real time using the formula:

[0016]

[0017] where x k is a single measurement, is the mean of these values, n is the number of samples;

[0018] Finally, combining the time window characteristics and frequency domain characteristics, the calculus formula is used:

[0019]

[0020] Where f(t) is the sensor data function within the time window, and t is the time parameter, which represents the time elapsed from a specific starting point.

[0021] Furthermore, the micro pump implementation method includes:

[0022] First, we implement an intelligent micro pump, integrate multi-dimensional sensors, and dynamically adjust the pump speed and pressure through the formula:

[0023]

[0024] Where P is the pressure in the thoracic cavity, V is the volume of the fluid, α, β, and λ are adjustment parameters, is the second-order time derivative of pressure, reflecting the acceleration of pressure change, It is the weighted exponential integral of liquid volume over time, indicating the impact of past liquid volume changes on the current state;

[0025] Next, a dynamic regulation algorithm based on fuzzy logic was developed to adjust the operating status of the pump using the formula:

[0026]

[0027] Where x is the real-time pressure data measured from the sensor, y is the pressure setting threshold, z is the pressure adjustment sensitivity, and γ is the adjustment amplitude factor.

[0028] Furthermore, the intelligent adjustment implementation method of the catheter system is as follows:

[0029] S1. Build an intelligent valve system, integrate flow meters and pressure sensors, and dynamically adjust the valve opening and closing through the formula:

[0030]

[0031] Where P is the pressure in the monitored chest cavity, Q is the fluid dynamics, δ, ω, μ are the adjustment parameters, τ is the integral variable, and t represents time;

[0032] S2. An adaptive algorithm developed using machine learning technology processes sensor data related to pleural leakage, using the algorithm formula:

[0033]

[0034] Where x, y represent the real-time and set pressure values, λ, φ are adjustment factors, τ is the integral variable, and t represents time;

[0035] S3, implement a closed-loop control system, adjust the treatment measures based on continuous feedback, using the formula:

[0036]

[0037] Where v(t) represents the valve opening degree, t represents time, and ρ, σ, η, and θ are control parameters.

[0038] Furthermore, a gated recurrent unit (GRU) based on a machine learning algorithm model is used to process the time series of input data using its internal state to predict and classify the leakage treatment method task;

[0039] The machine learning algorithm model is trained by inputting feature data, optimizing the model using a back propagation algorithm to reduce prediction errors, and adjusting and verifying the model to ensure performance;

[0040] The real-time analysis module uses the trained model to predict the data collected in real time, output the probability or risk level of leakage, and trigger an alarm based on the risk level predicted by the model to notify the medical team of the predicted leakage event.

[0041] Furthermore, the gated recurrent unit GRU construction method is as follows:

[0042] First, the gated recurrent unit GRU model is implemented to process the postoperative leakage and pressure changes of the thoracic cavity in real time, using the formula:

[0043]

[0044] Where σ is the activation function, β and b are learning parameters, and W i is the weight matrix, h(ti) is the state of the historical input, which is used to process and remember the time series characteristics in the input data;

[0045] Then integrate the GRU model into the real-time monitoring system to predict the trend and potential risk of leakage, and use the dynamic adjustment algorithm formula:

[0046]

[0047] Where γ is the adjustment coefficient, θ, λ and c are the adjustment parameters, and f(x(τ), y(τ)) is the function representing the pressure and volume data input from the sensor.

[0048] Furthermore, the back propagation algorithm optimization model construction process is as follows:

[0049] S1. First, develop a customized machine learning model, use the gated recurrent unit GRU to process time series medical data, and train it through the formula:

[0050]

[0051] where h t is the hidden state at time step t, o t-1 is the output of the previous time step, β and α are learning parameters, σ is a nonlinear activation function, and W ih ,W ho , and b h They are the weight matrix and bias from input to hidden layer and from hidden layer to output layer respectively;

[0052] S2. Then, the back propagation algorithm is used to optimize the model parameters and reduce the prediction error. The formula is:

[0053]

[0054] Update, where is t time step, o t is the output of the time step, is the loss function, θ represents the model parameters;

[0055] S3. Finally, the model is fully tested and validated through cross-validation and external validation datasets, and the model performance is evaluated using the formula:

[0056]

[0057] where y i is the true label, f(x i ; θ) is the model prediction and n is the number of samples.

[0058] Furthermore, the real-time analysis module construction method is as follows:

[0059] S1. First, a composite formula is used to process the real-time data from the intrathoracic sensor. The formula is:

[0060]

[0061] Where x(τ) represents the sensor reading at time τ, t represents the current time point, and λ is the attenuation coefficient;

[0062] S2. Then, the nonlinear characteristics of leakage dynamics are simulated by a prediction model based on nonlinear dynamics:

[0063]

[0064] where y t is the target predictor variable at the current moment, y s represents the value of the variable at time s, t represents the current time point, α i are model parameters, representing the effects of different time scales, and β(ts) is a time-dependent regulation function;

[0065] S3. Finally, when the leakage risk assessment exceeds the predetermined threshold, an alarm is triggered by the trigger formula:

[0066]

[0067] Where z is the risk level assessed by the model, κ(u) is the sensitivity function corresponding to the risk level, and u is the integral variable.

[0068] Beneficial effects of the present invention:

[0069] First, the system integrates highly sensitive sensors and intelligent analysis modules to achieve real-time monitoring of key indicators such as intrathoracic pressure and fluid accumulation, allowing medical staff to obtain patients' physiological data in a timely manner, quickly respond to changes in fluid leakage, and avoid treatment delays caused by delayed or inaccurate measurements.

[0070] Second, by using machine learning algorithms, including gated recurrent units (GRUs) and dynamic adjustment algorithms, the system can process complex time series data and make real-time predictions of leakage trends and risks. This intelligent analysis capability not only improves the accuracy of predictions, but also enables the system to identify potential risks in advance, provide early warning information, and help medical staff take intervention measures before symptoms worsen.

[0071] Third, by combining the patient's real-time monitoring data and historical medical records, the system can provide personalized treatment recommendations. This capability ensures the pertinence and effectiveness of the treatment plan, helps the medical team develop a treatment strategy that is more suitable for the patient's current condition, and improves treatment effectiveness and patient satisfaction.

[0072] Fourth, the system integrates automated valve adjustment and closed-loop control mechanisms, which can automatically adjust the leakage drainage speed and pressure level according to real-time monitoring data, reducing reliance on manual operations and reducing the workload of medical staff. At the same time, the automated processing method also improves the efficiency and reliability of the overall treatment process. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 The present invention is a flow chart of the post-thoracic fluid leakage monitoring system and processing method.

[0074] Figure 2 This is a module diagram of the central processing system of the present invention.

[0075] Figure 3 This is a module diagram for realizing intelligent regulation of the catheter system of the present invention.

[0076] Figure 4 This is a flow chart of the machine learning algorithm of the present invention. DETAILED DESCRIPTION

[0077] The specific implementation modes of the present invention will be described in detail below in conjunction with the accompanying drawings.

[0078] View attached Figure 1, the postoperative thoracic fluid leakage monitoring system and treatment method include: first, real-time monitoring through a set of micro-sensor networks. These sensors are placed inside the chest cavity or related areas to continuously measure a variety of important parameters, including fluid accumulation, pH value, ion concentration, temperature and pressure changes, capture various properties of the fluid in the chest cavity, and transmit these data to the central processing system in real time wirelessly. The central processing system then pre-processes the collected data, including denoising and smoothing the data to ensure the accuracy and reliability of the data. Feature extraction is then performed, including calculating the maximum, minimum, mean, and standard deviation of the data. These statistical features help the system understand the overall situation and trend of fluid accumulation.

[0079] In addition, the system analyzes the differences and rates of change in the data. These time window features help capture the dynamic changes of fluid accumulation. Through Fourier transform, the system can analyze the frequency characteristics of the data, and when the frequency characteristics of fluid accumulation show abnormal fluctuations, the system can issue an alarm in time. According to the analysis results of the machine learning algorithm, the central processing system adjusts the automation device to deal with the detected fluid accumulation problem. The automation device mainly includes a micro pump and a catheter system. The micro pump is used to control the pressure and flow volume in the chest cavity in real time. According to the sensor data, the system can dynamically adjust the pump speed and pump pressure to timely remove abnormal fluid accumulation and maintain normal pressure in the chest cavity. The catheter system is equipped with an adaptive switch valve with intelligent adjustment. The design of this system allows the valve to open or close automatically according to the nature and amount of the liquid. By optimizing the frequency and amplitude of the valve opening and closing by the algorithm, the system can ensure the effectiveness of the liquid flow and prevent excessive or insufficient discharge. If the pH value and ion concentration of the liquid show abnormalities, the valve will adjust the switch state to cope with different liquid characteristics.

[0080] Finally, the system is able to generate personalized monitoring plans based on individual patient differences. These personalized plans take into account the patient's specific condition, postoperative recovery process, and other health factors to provide customized monitoring and treatment. The generated detailed postoperative monitoring report includes data analysis, risk assessment, and recommended interventions.

[0081] Embodiment 1:

[0082] Combined with Figure 2 In the embodiment, in the postoperative thoracic fluid leakage monitoring system, the error and drift in the sensor data are corrected by the data preprocessing module, and the mathematical model is introduced:

[0083]

[0084] x, y, and z represent the temperature, pressure, and liquid accumulation data collected from the sensor, respectively. Coefficients a and b are parameters that are dynamically adjusted according to the data characteristics and vary in the range of (0.1, 2.0) to adapt to different data characteristics and specific settings of the sensor.

[0085] The sensor data for chest monitoring of the patient is as follows:

[0086] Temperature data: 37.5℃ (converted to a unit suitable for calculation: 37.5);

[0087] Pressure data: 100 mmHg (converted to a unit suitable for calculation as 100);

[0088] Liquid volume data: 50ml (converted to a unit suitable for calculation is 50);

[0089] The initial dynamic adjustment coefficients are set to a=1.2 and b=0.8.

[0090] Apply a preprocessing function to this set of data:

[0091]

[0092] Compute the sum of squares:

[0093] 37.5 2 +100 2 =1406.25+10000=11406.25

[0094] Calculate the exponential part:

[0095] e -11406.25 ≈0

[0096] Since the result of the exponential function is extremely small, it means that when the combination of temperature and pressure is extremely large, the adjustment of the signal has little effect and mainly depends on the adjustment of the second part.

[0097] Compute the logarithmic part:

[0098] log(11406.25+1)=log(11407.25)≈9.34

[0099] Final result:

[0100] F(x,y,z)≈1.2·0+0.8·9.34=7.472

[0101] The calculated result F(x,y,z)≈7.472. After adjustment, the adjusted results of this set of data are used for further analysis and monitoring.

[0102] The embodiment further calculates the statistical characteristics of liquid accumulation and pressure using the following formula:

[0103]

[0104] in:

[0105] x k is each individual measurement value (fluid volume or pressure value);

[0106] is the average of these measurements;

[0107] n is the number of samples, which is all the measurement samples within a period of time;

[0108] k! is a factorial that normalizes the contributions of features of different orders.

[0109] Assume that a post-operative patient is using the monitoring system and the sensor has recorded the following fluid accumulation (in ml) over the past 5 minutes:

[0110] Measurements x: 48,52,50,49,51.

[0111] Calculate the average

[0112]

[0113] The feature G(x) is calculated using the formula:

[0114]

[0115] Calculate item by item:

[0116]

[0117] All entries are 0.

[0118]

[0119] Sum calculation:

[0120] G(x)=-0.933+6.267+0+-0.633+1.717=6.418

[0121] The calculated result G(x)=6.418 indicates the high-order statistical characteristics of these measured data. This value can help the system identify trends or abnormal patterns of fluid accumulation.

[0122] The last embodiment uses the time window characteristics and frequency domain characteristics to further improve the accuracy of data analysis and real-time response capability, and uses the following calculus formula to analyze the data changes within the time window:

[0123]

[0124] in:

[0125] f(t) is the function representation of the sensor data in the time window;

[0126] t is a time parameter, which represents the elapsed time from a specific starting point.

[0127] The sensor recorded the change in liquid volume accumulation (in ml) within 5 minutes, and the measurement data were as follows: Time (minutes): 0, 1, 2, 3, 4, 5;

[0128] Fluid accumulation:50,52,51,54,53,55.

[0129] In a given time window, discrete data is converted into a continuous function representation by linear interpolation:

[0130] For intervals from 0 to 1 minute:

[0131] f(t)=50+(52-50)·(t-0)=50+2t,0≤t<1 For an interval of 1 to 2 minutes:

[0132] f(t)=52-(52-51)·(t-1)=52-t,1≤t<2For an interval of 2 to 3 minutes:

[0133] f(t)=51+(54-51)·(t-2)=51+3(t-2), 2≤t<3 For an interval of 3 to 4 minutes:

[0134] f(t)=54-(54-53)·(t-3)=54-(t-3), 3≤t<4 For an interval of 4 to 5 minutes:

[0135] f(t)=53+(55-53)·(t-4)=53+2(t-4), 4≤t≤5 Calculate the derivative of the data For each time interval, compute the derivative:

[0136]

[0137] Apply the integral formula to calculate the characteristics:

[0138] For 0≤t<1:

[0139]

[0140] For 1≤t<2:

[0141]

[0142] For 2≤t<3:

[0143]

[0144] For 3≤t<4:

[0145]

[0146] For 4≤t≤5:

[0147]

[0148] The calculated H(f) value provides the dynamic characteristics of fluid accumulation changes in different time windows, helping the medical team understand the development trend of the disease.

[0149] Embodiment 2:

[0150] In the postoperative thoracic fluid leakage monitoring system, an intelligent micro pump is used to achieve precise control of the fluid and pressure in the thoracic cavity, and a multi-dimensional sensor is integrated to control the following through a control algorithm:

[0151]

[0152] Where P is the intrathoracic pressure; V is the fluid volume; t is time; α and β are adjustment coefficients, which are determined based on clinical data or predictive tests, ranging from 0.1 to 10, and are set according to the sensitivity and response requirements of the system; λ is the attenuation factor, which affects the weight of historical fluid volume data and is set in the range of 0.1 to 1.

[0153] Assuming that a postoperative patient is using the system for chest monitoring, the system records the following data:

[0154] Pressure data: P(t)=100+10sin(t)mmHg;

[0155] Liquid volume data: V(t)=50+2tml.

[0156] Calculate the second time derivative of pressure:

[0157]

[0158] The calculated result at t=2 is -10sin(2)≈-8.84.

[0159] Compute the integral using numerical integration techniques:

[0160]

[0161] The result of integration is 120 ml.

[0162] Set α = 1.5, β = 0.8, λ = 0.5, then at t = 2:

[0163] S(P,V,2)=1.5·(-8.84)+0.8·120=-13.26+96=82.74

[0164] The calculated S(P,V,2)=82.74 indicates that at a given moment, based on the acceleration of pressure change and the historical cumulative impact of liquid volume, the smart pump needs to take corresponding actions to adjust the pump speed and pump pressure to maintain the pressure and liquid volume in the chest cavity at a safe and appropriate level.

[0165] Next, the embodiment adopts a dynamic adjustment algorithm based on advanced fuzzy logic to dynamically adjust the pump speed and pump pressure according to real-time pressure data, a preset pressure threshold and pressure adjustment sensitivity.

[0166]

[0167] in:

[0168] x is the real-time pressure data measured from the sensor;

[0169] y is the pressure setting threshold, which is set according to clinical needs and depends on the patient's specific health status and postoperative recovery plan;

[0170] z is the pressure adjustment sensitivity, which determines the speed and sensitivity of the response, and a smaller value means a higher sensitivity;

[0171] γ is the adjustment amplitude factor, which defines the maximum range of adjusting the pump speed or pump pressure, and its value is in the range of 0,2.

[0172] It is assumed that the patient is using the monitoring system to control chest pressure after surgery. The data are as follows:

[0173] The pressure data x is recorded once every minute, and the latest measurement value is 100 mmHg;

[0174] The set pressure threshold y is 95 mmHg, which is the safe pressure limit set based on the doctor's clinical assessment;

[0175] The pressure adjustment sensitivity x is set to 10 mmHg, which provides moderate adjustment sensitivity, ensuring that the system response is neither overly sensitive nor sluggish;

[0176] The adjustment amplitude factor γ is set to 1.5 to provide sufficient adjustment amplitude to cope with pressure changes.

[0177] Application of calculation formula:

[0178]

[0179] The calculated adjustment value R(x,y,z)=2.193 indicates that the pump speed or pump pressure needs to be increased to reduce the pressure below the set threshold. A value greater than 2 indicates that a larger adjustment is required, and the smart pump will increase the pump speed or increase the displacement accordingly to quickly reduce the pressure in the chest cavity and avoid discomfort or potential health risks to the patient.

[0180] Embodiment 3:

[0181] Combined with Figure 3 In the embodiment, in the postoperative thoracic fluid leakage monitoring system, the intelligent catheter adjustment system is first implemented, and the valve adjustment is performed by real-time monitoring of the pressure and fluid dynamics in the thoracic cavity, using the following formula:

[0182]

[0183] Parameter value range:

[0184] δ: The sensitivity of the control system to pressure changes, with a range of values ​​(0.1, 2.0);

[0185] ω: Weight coefficient for adjusting fluid dynamics, value range (0.01, 1.0);

[0186] μ: Controls the attenuation rate of the influence of historical data, with a value range of (0.1, 0.5).

[0187] The system monitors the pressure changes and fluid dynamics data as follows:

[0188] Current pressure P(t) = 12 mmHg;

[0189] Pressure change rate

[0190] Fluid flow Q(t) = 5 mL / s.

[0191] Set the parameter values ​​δ = 1.5, ω = 0.5, μ = 0.2, and calculate the historical fluid dynamic impact:

[0192]

[0193] Solving the integral yields:

[0194]

[0195] At t = 5s, the calculation is:

[0196] =25(1-e -1 )≈15.9

[0197] Substituting the result into the formula:

[0198] D(P,Q,t)=1.5(0.2+0.5×15.9)=1.5(0.2+7.95)=1.5×8.15=12.225

[0199] The intelligent system calculates D(P,W,t)=12.225, which exceeds the preset safety threshold. The system adjusts the valve state, increases the drainage speed, and reduces pressure accumulation.

[0200] The embodiment further enhances the intelligent processing capability of the monitoring system and introduces an adaptive algorithm developed based on machine learning technology:

[0201]

[0202] in:

[0203] x is the current real-time monitored pressure value;

[0204] y is the pressure threshold set based on clinical guidelines to determine the risk of leakage;

[0205] λ is the coefficient for adjusting the impact of model output, and its value range is set to (0.5, 2.0), which is used to adjust the sensitivity of model response;

[0206] φ is the coefficient that affects the amplitude of the sine function, and its value range is set to (0.1, 1.0) to adjust the influence of periodic fluctuations;

[0207] τ is the variable of integration, which represents the duration from 0 to the current time t.

[0208] It is assumed that at a certain moment, the pressure measured by the system is x=15 mmHg, the pressure threshold is set to y=12 mmHg, and λ=1.0 and φ=0.8 are selected for calculation.

[0209] Calculate the first part, the logical response:

[0210] log(1+e 15-12 )=log(1+e 3 )≈log(1+20.085)≈log(21.085)≈3.05

[0211] Assuming the pressure x(τ) in the past 5 seconds remains at 15 mmHg, integrate the second part:

[0212]

[0213] The total calculation result is:

[0214] A(x,y,t)=1.0×(3.05+1.04)=4.09

[0215] When A(x,y,t) exceeds the preset alarm trigger threshold, the system will trigger an alarm to notify the medical team of the potential risk of leakage and recommend emergency examination or adjustment of treatment.

[0216] The embodiment further implements a closed-loop control system, which operates by the following control formula:

[0217]

[0218] in:

[0219] v(t) represents the degree of valve opening; ρ, σ, η, and θ are control parameters; ρ controls the overall sensitivity of the reaction, with a range of values ​​(0.5, 2.0); σ: affects the baseline value of the denominator, with a range of values ​​(1, 10); η: adjusts the rate at which the denominator changes with v(t), with a range of values ​​(0.01, 0.1); θ is the basic adjustment parameter to ensure system stability, with a range of values ​​(-0.5, 0.5).

[0220] Assume that at a certain detection moment, the valve opening degree measured by the system is v(t) = 40%, and the parameters are set to ρ = 1.0, σ = 5, η = 0.05, θ = 0.1. First, calculate the change rate of the valve opening degree:

[0221] Assume that the valve opening increases slightly from 40% to 41% in a short period of time, and calculate the derivative part of the change in valve opening:

[0222]

[0223] This can be simplified to:

[0224]

[0225] Therefore, substituting the value into the closed-loop control formula yields:

[0226] C(v,t)=1.0(5.71+0.1)=5.81

[0227] The value of 5.81 indicates that the valve opening needs to be increased to reduce the internal pressure and avoid potential risks.

[0228] Embodiment 4:

[0229] In the embodiment, a gated recurrent unit (GRU) model is used to enhance the intelligent analysis capability of the postoperative thoracic fluid leakage monitoring system. The core formula of the model is as follows:

[0230]

[0231] in:

[0232] Y(t) is the model output at the current time t, which is used to predict leakage risk or pressure trend;

[0233] σ is the activation function sigmoid, which is used to introduce nonlinearity to make the model more expressive;

[0234] β and b are learning parameters that control the model's response strength to input, and their value ranges are (0.5, 2.0) and (-1, 1) respectively;

[0235] W i is a weight matrix responsible for learning the relative importance of different input features in the model, with a value range of (-0.5, 0.5);

[0236] h(ti) is the state of historical input, which indicates the model’s ability to remember past data.

[0237] Assume that at a certain moment, the data collected by the system is as follows:

[0238] Leakage volume time series data: [3, 2.8, 2.9, 3.1, 3.0] mL (data from the last 5 time points)

[0239] Pressure change time series data: [10, 10.5, 10.2, 10.1, 10.3] mmHg

[0240] Model parameter selection:

[0241] β=1.5

[0242] b=0.2

[0243] Using the sigmoid activation function:

[0244] The weighted sum part of the calculation formula is:

[0245] Set the weight matrix W i For [0.3, 0.25, 0.2, 0.15, 0.1], calculate the current prediction output:

[0246]

[0247] Calculated as:

[0248] =0.9+0.7+0.58+0.465+0.3=2.945

[0249] Substituting into the formula we get:

[0250] Y(t)=σ(1.5·2.945+0.2)=σ(4.6175)≈0.99

[0251] The calculated Y(t)≈0.99 indicates that the leakage risk is close to the critical value.

[0252] The embodiment further integrates the GRU model into the real-time monitoring system to predict the trend and potential risk of leakage. The formula is as follows:

[0253]

[0254] in:

[0255] Z(t) is the predicted leakage risk value, which is used to guide medical decision-making;

[0256] γ is the adjustment coefficient, with a value range of (0.5, 2.0), which is used to adjust the sensitivity of the model;

[0257] θ, λ, and c are adjustment parameters, which are used to control the weight, decay rate, and baseline adjustment of the integration results, respectively. The value range of θ is (1, 10), the value range of λ is (0.1, 1.0), and the value range of c is (-1, 1);

[0258] f(x(τ),y(τ)) is a function that combines the pressure and volume data and represents the combined effect of the sensor input.

[0259] The system collects the following data:

[0260] Real-time pressure data x(τ): [10, 10.5, 10.2, 10.1, 10.3] mmHg;

[0261] Real-time volume data y(τ): [3, 2.8, 2.9, 3.1, 3.0] mL;

[0262] Set parameters γ=1.5; θ=5; λ=0.5; c=0.2.

[0263] Set the function f(x(τ), y(τ)) = x(τ) · y(τ) and calculate its integral part:

[0264]

[0265] Set t = 5:

[0266] f(x,y)=[10·3,10.5·2.8,10.2·2.9,10.1·3.1,10.3·3.0]

[0267] Calculate e -λ(t-τ) :

[0268] e -λ(t-τ) =[e -0.5×5 ,e -0.5×4 ,e -0.5×3 ,e-0.5×2 ,e -0.5×1 ]

[0269] Perform an integral operation:

[0270] =3×0.0821+2.8×0.1353+2.9×0.2231+3.1×0.3679+3.0×0.6065≈3.549

[0271] Substituting the result into the formula:

[0272] Z(t)=1.5(5×3.549+0.2)=1.5×(17.745+0.2)=1.5×17.945=26.9175

[0273] The calculated Z(t)=26.9175 indicates that the potential risk of leakage increases and exceeds the safety threshold of 25 set by the system.

[0274] Embodiment 5:

[0275] The example uses an advanced postoperative thoracic fluid leakage monitoring system to develop a customized GRU model to process time series medical data collected from intrathoracic sensors and analyze fluid leakage in real time. The training formula of the model is as follows:

[0276]

[0277] in:

[0278] L(t) is the model loss at time step t;

[0279] h t is the hidden state of the current time step, capturing the characteristics of historical data;

[0280] o t-1 is the output of the previous time step, which is used to recursively update the model state;

[0281] β and α are learning parameters, with value ranges of (0.1, 1.0) and (0.01, 0.1), respectively, which are used to adjust the response strength of the model to the input features;

[0282] σ is the nonlinear activation function sigmoid, which is used to introduce nonlinearity so that the model can handle complex relationships;

[0283] W ih ,W ho ,b h is the weight matrix and bias, with a value range of (-0.5, 0.5).

[0284] The following data series were collected:

[0285] Pressure data sequence: [12.0, 12.2, 11.8, 12.5, 12.3] mmHg;

[0286] Leakage volume data sequence: [2.0, 2.2, 1.9, 2.5, 2.3] mL;

[0287] Parameters β=0.8, α=0.05.

[0288] Use sigmoid as the activation function:

[0289] Calculate the hidden state update part:

[0290] Set the input to the hidden layer weight W ih =[0.1,0.2,0.1,0.2,0.1];

[0291] Set the weight W from the hidden layer to the output layer ho =0.3;

[0292] Deviation b h =0.1.

[0293] Compute the hidden state:

[0294]

[0295] Substituting into the formula:

[0296] L(t)=0.8·(σ(1.56+0.3×12.3+0.1))+0.05·1.56

[0297] =0.8·σ(1.56+3.69+0.1)+0.078

[0298] =0.8·σ(5.35)+0.078≈0.8·0.9953+0.078=0.8762

[0299] The output of the model, L(t)≈0.8762, represents the current loss value. The loss value can be further reduced by optimizing the back-propagation algorithm.

[0300] The embodiment further uses a back propagation algorithm to optimize the model parameters, and updates the parameters using the following formula:

[0301]

[0302] in:

[0303] is the loss function, using mean square error or cross entropy loss;

[0304] θ represents the model parameters, including weights and biases;

[0305] o t is the output at time step t;

[0306] h t is the hidden state at time step t;

[0307] t is the time step and T is the total number of time steps.

[0308] In actual monitoring, the data collected by the system are as follows:

[0309] Pressure data sequence: [12.0, 12.2, 11.8, 12.5, 12.3] mmHg;

[0310] Leakage volume data sequence: [2.0, 2.2, 1.9, 2.5, 2.3] mL.

[0311] Set the initial parameters of the model to:

[0312] Weight matrix W ih =[0.1,0.2,0.1,0.2,0.1];

[0313] Weight matrix W ho =0.3;

[0314] Deviation b h =0.1.

[0315] Set the loss function to mean square error:

[0316]

[0317] in is the predicted value, o t is the true value.

[0318] Compute the output error at a certain time step:

[0319]

[0320] Assuming that at time step t=5, the actual output of the system is o t =2.4mL, predicted value

[0321]

[0322] Calculate the hidden state h t For output o t The derivative of :

[0323]

[0324] Compute the derivatives of the hidden state with respect to the model parameters:

[0325]

[0326] Substitute these values ​​into the back-propagation formula to update the parameters:

[0327]

[0328] Compute the gradient:

[0329] M(5)=0.03·[12.0,12.2,11.8,12.5,12.3]=[0.36,0.366,0.354,0.375,0.369]

[0330] Through the above calculation, the model parameter W ih Will be updated. Set the learning rate to 0.01, and the updated weight matrix is:

[0331] W ih =W ih -0.01·M(5)

[0332] Update results:

[0333] W ih =[0.1,0.2,0.1,0.2,0.1]-0.01·[0.36,0.366,0.354,0.375,0.369]

[0334] =[0.0964,0.19634,0.09646,0.19625,0.09631]

[0335] Through continuous optimization of the back-propagation algorithm, the model parameters are gradually adjusted, the loss function value is reduced, and the prediction error is reduced.

[0336] The final example uses an exponential decay error function to thoroughly test and validate the model using cross-validation and external validation datasets:

[0337]

[0338] in:

[0339] N(θ) is the performance evaluation value of the model, reflecting the overall accuracy of the model prediction;

[0340] y i is the true label of the i-th sample, indicating the actual leakage volume or pressure;

[0341] f(x i ; θ) is the predicted value of the model for the i-th sample under given parameters θ;

[0342] n is the number of samples, representing the total number of data in the validation set.

[0343] The data of labels and model prediction values ​​are as follows:

[0344] True label y: [2.0, 1.8, 2.2, 2.1, 2.3];

[0345] Model predictions f(x;θ): [2.1, 1.7, 2.0, 2.2, 2.4].

[0346] Compute the exponentially decaying error for each sample:

[0347]

[0348] The specific calculation is:

[0349] =[e -4.2 ,e -3.06 ,e -4.4 ,e -4.62 ,e -5.52 ]

[0350] Numerical results:

[0351] =[0.0149,0.046,0.0122,0.0099,0.004]

[0352] Substituting these results into the performance evaluation formula:

[0353] Assuming that there are n=100 samples, the evaluation of the first five samples is:

[0354]

[0355] It turns out that:

[0356] N(θ)=0.00087

[0357] By calculating this exponential decay error, the hospital can accurately evaluate the performance of the model on new data. If the N(θ) value is low, it means that the model has a small deviation in prediction and has good generalization ability.

[0358] Embodiment 6:

[0359] Combined with Figure 4 The embodiment optimizes the real-time analysis module of the postoperative thoracic fluid leakage monitoring system to improve its prediction ability and response speed to fluid leakage, using the formula:

[0360]

[0361] in:

[0362] P(x,t) is the result of processing the sensor data at the current time t, reflecting the leakage trend;

[0363] x(τ) is the sensor reading at time τ, representing the pressure or flow in the chest cavity;

[0364] t is the current time point;

[0365] λ is the attenuation coefficient, which is used to control the impact of historical data on current calculations, and its value range is (0.1, 0.5).

[0366] In the setting of postoperative patient monitoring, the system collects the following data from the sensor:

[0367] Pressure data sequence x(τ): [12.0, 12.2, 11.8, 12.5, 12.3] mmHg;

[0368] Set parameter λ=0.3.

[0369] Calculate the integral part of the formula using the discrete integral approximation:

[0370]

[0371] Specific calculation:

[0372] x i ·e -λ(5-i) =[12.0·e -0 ,12.2·e -0.3 ,11.8·e -0.6 ,12.5·e -0.9 ,12.3·e -1.2 ]

[0373] =[12.0,12.2·0.7408,11.8·0.5488,12.5·0.4066,12.3·0.3012]

[0374] =[12.0,9.03,6.47,5.08,3.71]

[0375]

[0376] Compute squares and integrals:

[0377]

[0378] Substitute the formula to calculate p(x,t):

[0379]

[0380] The calculated P(x, t)≈1.33 indicates that the current leakage trend is relatively stable. Based on this calculation result, the system can determine that the patient's leakage risk is under control. If the P(x, t) value increases above the set threshold, the system will trigger an alarm to remind the medical team to conduct further examinations or adjust treatment strategies.

[0381] Next, the embodiment uses a prediction model based on nonlinear dynamics to simulate the nonlinear characteristics of leakage, and the formula is as follows:

[0382]

[0383] in:

[0384] F(y t ) is the prediction output at the current time t, which is used to indicate the leakage risk or trend;

[0385] y t is the current leakage volume;

[0386] y s represents the value of the variable at time s, used to assess past impacts;

[0387] t is the current time point;

[0388] α i is a model parameter with a value range of (0.1, 1.0), representing the impact of different time scales on the prediction;

[0389] β(ts) is the time-dependent regulation function, with a value range of (0.01, 0.1);

[0390] γ is the adjustment coefficient, and its value range is (0.5, 2.0).

[0391] Assume that at a certain moment, the system collects the following data:

[0392] Leakage volume data t :The current value is 2.5mL, and the previous values ​​are [2.0,2.2,1.9,2.5,2.3]mL;

[0393] Model parameter settings: γ = 1.5, α = [0.2, 0.3, 0.1, 0.2, 0.1], β (ts) = 0.05.

[0394] Calculate the first part, the time scale effect:

[0395] Set to use the second-order derivative and calculate Approximate value of :

[0396]

[0397] Weighted sum:

[0398]

[0399] Compute the second part, the integral:

[0400] Set y s = sin(y s )≈sin([2.0,2.2,1.9,2.5,2.3])≈[0.91,0.81,0.95,0.60,0.75]

[0401]

[0402] Substitute into the formula to calculate F(y t ):

[0403] F(y t )=1.5×(2.5-0.06)+0.201=1.5×2.44+0.201=3.66+0.201=3.861

[0404] The calculated F(y t )≈3.861 represents the current leakage risk prediction value.

[0405] The last embodiment uses a post-thoracic fluid leakage monitoring system. When the leakage risk assessment value z exceeds the set threshold, the trigger formula will be used to start the alarm:

[0406]

[0407] in:

[0408] A(z) is the alarm activation value, which triggers the alarm when it is close to 1;

[0409] z is the risk level assessed by the model;

[0410] κ(u) is the sensitivity function of the risk level, which defines the impact of each risk level on the alarm, and its value range is (0.01, 0.1);

[0411] u is an integral variable representing the cumulative effect within the range of risk levels.

[0412] The patient's current leakage risk assessment value z is set to 4.0, and the risk sensitivity function set by the hospital is κ(u) = 0.05u, which means that as the risk level increases, the sensitivity increases linearly.

[0413] Calculate the integral part:

[0414]

[0415] Solve the integral:

[0416]

[0417] Substitute the integral result into the trigger formula to calculate A(z):

[0418] A(z)=1-e -0.4 =1-e -0.4 ≈1-0.6703=0.3297

[0419] The calculated A(z)=0.3297 has not yet reached the alarm triggering threshold value of 0.8 set by the hospital, so the system will not trigger an alarm.

[0420] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. Postoperative thoracic fluid leakage monitoring system, characterized by The components include: a network of micro-sensors that monitor the amount of fluid accumulated in the chest cavity, fluid properties including pH, ion concentration, temperature and pressure changes in real time, and send the data wirelessly to a central processing system; The central processing system has a built-in machine learning algorithm for analyzing sensor data, predicting leakage trends and risks, and dynamically adjusting the automation device according to the analysis results; wherein the central processing system preprocesses and extracts features from the collected data, including statistical features of maximum value, minimum value, mean value, standard deviation, time window features including data differences and change rates, and frequency characteristics of the data analyzed by Fourier transform; The automated device drains abnormal fluid accumulation and adjusts internal pressure through a micro pump and catheter system; The micro pump is used to monitor the pressure and flow volume in the chest cavity in real time, and dynamically adjust the pump speed and pump pressure according to the sensor data; The second is the intelligent regulation of the catheter system, which is designed with an adaptive switch valve that automatically opens or closes according to the properties and amount of the liquid, and optimizes the opening and closing frequency and amplitude of the valve through an algorithm; Finally, a personalized monitoring plan is generated based on individual patient differences, and a detailed postoperative monitoring report is generated, including data analysis, risk assessment, and recommended intervention measures, to provide high-precision monitoring and efficient personalized treatment.

2. The post-thoracic fluid leakage monitoring system according to claim 1 is characterized in that The central processing system construction process includes: first, implementing a data preprocessing module to correct errors and drifts in sensor data, using the function: Where x, y, and z represent different types of data collected from sensors, including temperature, pressure, and liquid accumulation, and a and b are coefficients that are dynamically adjusted based on the characteristics of the data; Next, calculate the liquid accumulation and pressure statistics in real time using the formula: where x k is a single measurement, is the mean of these values, and n is the number of samples; Finally, combining the time window characteristics and frequency domain characteristics, the calculus formula is used: Where f(t) is the sensor data function within the time window, and t is the time parameter, which represents the time elapsed from a specific starting point.

3. The post-thoracic fluid leakage monitoring system according to claim 1 is characterized in that The micro pump implementation method comprises: First, we implement an intelligent micro pump, integrate multi-dimensional sensors, and dynamically adjust the pump speed and pressure through the formula: Where P is the pressure in the thoracic cavity, V is the volume of the fluid, α, β, and λ are adjustment parameters, is the second-order time derivative of pressure, reflecting the acceleration of pressure change, It is the weighted exponential integral of liquid volume over time, indicating the impact of past liquid volume changes on the current state; Next, a dynamic regulation algorithm based on fuzzy logic was developed to adjust the operating status of the pump using the formula: Where x is the real-time pressure data measured from the sensor, y is the pressure setting threshold, z is the pressure adjustment sensitivity, and γ is the adjustment amplitude factor.

4. The post-thoracic fluid leakage monitoring system according to claim 1 is characterized in that The intelligent adjustment implementation method of the catheter system is as follows: S1. Build an intelligent valve system, integrate flow meters and pressure sensors, and dynamically adjust the valve opening and closing through the formula: Where P is the pressure in the monitored chest cavity, Q is the fluid dynamics, δ, ω, μ are the adjustment parameters, τ is the integral variable, and t represents time; S2. An adaptive algorithm developed using machine learning technology processes sensor data related to pleural leakage, using the algorithm formula: Where x, y represent the real-time and set pressure values, λ, φ are adjustment factors, τ is the integral variable, and t represents time; S3, implement a closed-loop control system, adjust the treatment measures based on continuous feedback, using the formula: Where v(t) represents the valve opening degree, t represents time, and ρ, σ, η, and θ are control parameters.

5. A method for treating postoperative thoracic fluid leakage, characterized in that: Adopt the gated recurrent unit GRU based on the machine learning algorithm model, use its internal state to process the time series of input data, predict and classify the leakage treatment method task; The machine learning algorithm model is trained by inputting feature data, optimizing the model using a back propagation algorithm to reduce prediction errors, and adjusting and verifying the model to ensure performance; The real-time analysis module uses the trained model to predict the data collected in real time, output the probability or risk level of leakage, and trigger an alarm based on the risk level predicted by the model to notify the medical team of the predicted leakage event.

6. The method for treating postoperative thoracic fluid leakage according to claim 5, characterized in that The gated recurrent unit GRU construction method is as follows: First, the gated recurrent unit GRU model is implemented to process the postoperative leakage and pressure changes of the thoracic cavity in real time, using the formula: Where σ is the activation function, β and b are learning parameters, and W i is the weight matrix, h(ti) is the state of the historical input, which is used to process and remember the time series characteristics in the input data; Then integrate the GRU model into the real-time monitoring system to predict the trend and potential risk of leakage, and use the dynamic adjustment algorithm formula: Where γ is the adjustment coefficient, θ, λ and c are the adjustment parameters, and f(x(τ), y(τ)) is the function representing the pressure and volume data input from the sensor.

7. The method for treating postoperative thoracic fluid leakage according to claim 5, characterized in that The back propagation algorithm optimization model construction process is as follows: S1. First, develop a customized machine learning model, use the gated recurrent unit GRU to process time series medical data, and train it through the formula: where h t is the hidden state at time step t, o t-1 is the output of the previous time step, β and α are learning parameters, σ is a nonlinear activation function, and W ih ,W ho , and b h They are the weight matrix and bias from input to hidden layer and from hidden layer to output layer respectively; S2. Then, the back propagation algorithm is used to optimize the model parameters and reduce the prediction error. The formula is: Update, where is t time step, o t is the output of the time step, is the loss function, θ represents the model parameters; S3. Finally, the model is fully tested and validated through cross-validation and external validation datasets, and the model performance is evaluated using the formula: where y i is the true label, f(x i ; θ) is the model prediction and n is the number of samples.

8. The method for treating postoperative thoracic fluid leakage according to claim 5, characterized in that The real-time analysis module construction method is as follows: S1. First, a composite formula is used to process the real-time data from the intrathoracic sensor. The formula is: Where x(τ) represents the sensor reading at time τ, t represents the current time point, and λ is the attenuation coefficient; S2. Then, the nonlinear characteristics of leakage dynamics are simulated by a prediction model based on nonlinear dynamics: where y t is the target predictor variable at the current moment, y s represents the value of the variable at time s, t represents the current time point, α i are model parameters, representing the effects of different time scales, and β(ts) is a time-dependent regulation function; S3. Finally, when the leakage risk assessment exceeds the predetermined threshold, an alarm is triggered by the trigger formula: Where z is the risk level assessed by the model, κ(u) is the sensitivity function corresponding to the risk level, and u is the integral variable.

Citation Information

Cited By

  • Blood multimode infection monitoring method and system and dialysis catheter

    CN121034669A