Peritoneal drainage system based on hemoglobin real-time monitoring and intelligent early warning

Abstract

The invention relates to the technical field of peritoneal drainage, in particular to a peritoneal drainage system based on hemoglobin real-time monitoring and intelligent early warning. According to the technical scheme, the device comprises an abdominal cavity drainage tube, an optical sensing module, a microprocessor module and a wireless communication module, wherein the microprocessor module is configured to perform moving average filtering and dynamic trend modeling on hemoglobin concentration data; and triggering early warning according to a preset multi-level alarm condition. The hemoglobin concentration in the peritoneal drainage liquid is accurately monitored in real time, intelligent early warning and a personalized response mechanism are combined, the limitation of traditional manual observation is broken through, the timeliness and accuracy of bleeding risk identification are remarkably improved, the infection risk is reduced through the closed design, the medical care working process is optimized through the system integration capacity, and the medical care efficiency is improved. Data sharing and remote cooperation are achieved, intervention time windows are shortened, patient prognosis is improved, nursing resources can be released, medical quality is improved, and a comprehensive and efficient solution is provided for postoperative monitoring.

Description

Technical Field

[0001] The present invention relates to the technical field of abdominal drainage, and in particular to an abdominal drainage system based on real-time hemoglobin monitoring and intelligent early warning. Background Art

[0002] Pancreaticoduodenectomy (Whipple procedure), a key surgical procedure for treating periampullary cancer and pancreatic cancer, has a high incidence of postoperative bleeding complications of 8-19%, with approximately 3-5% requiring reoperation for severe bleeding. Currently, clinical evaluation relies on medical staff manually observing the color changes of peritoneal drainage fluid every 1-2 hours to assess bleeding, which has significant limitations:

[0003] The subjectivity and lag of manual observation

[0004] Color judgment errors are large: Clinical studies have shown that the Kappa value for consistency in drainage fluid color grading (clear, reddish, bloody) among different medical staff is only 0.42-0.58 (moderate consistency), and the misjudgment rate is particularly high at night or in emergency situations. For example, when the drainage fluid hemoglobin concentration is in the critical range of 1-3g / dL, the probability of misjudgment is high.

[0005] Early signs of bleeding are difficult to detect: In the early stages of postoperative bleeding, drainage fluid may appear only slightly turbid or pale pink, difficult to detect with the naked eye. Studies have shown that when hemoglobin concentration rises by 20% from baseline (indicating early bleeding), manual observation has a high rate of missed diagnosis.

[0006] Delayed intervention: The average time from the onset of bleeding to the discovery of typical bloody drainage by medical staff is 1.8 hours, by which time blood loss may have reached 500-800 ml. A retrospective study of 237 patients after Whipple surgery showed that the mortality rate in patients whose bleeding was delayed for more than 2 hours was 3.2 times that of the group receiving prompt intervention (P < 0.01).

[0007] Currently, there is a lack of medical devices on the market that can continuously monitor changes in hemoglobin concentration in peritoneal drainage fluid in real time. Although optical sensing technology has made progress in the field of blood analysis, existing devices have the following problems: they are not optimized for peritoneal drainage scenarios and cannot adapt to dynamic fluid environments and complex component interference;

[0008] The lack of a personalized alarm threshold algorithm makes it difficult to distinguish normal postoperative exudation from active bleeding;

[0009] It is not integrated with the hospital information system, and automated early warning and medical resource scheduling cannot be achieved.

[0010] In summary, this application proposes a peritoneal drainage system based on real-time hemoglobin monitoring and intelligent early warning. Summary of the Invention

[0011] The purpose of the present invention is to address the problem in the background technology that there is a lack of medical equipment on the market that can monitor the changes in hemoglobin concentration in peritoneal drainage fluid in real time and continuously, and to propose a peritoneal drainage system based on real-time monitoring and intelligent early warning of hemoglobin.

[0012] The technical solution of the present invention is a peritoneal drainage system based on real-time hemoglobin monitoring and intelligent early warning, comprising:

[0013] Abdominal drainage tube: equipped with a transparent measuring window for drainage fluid flow;

[0014] Optical sensing module: non-contact mounted outside the transparent measurement window, comprising a dual-wavelength light source and a photoelectric detection unit; the dual-wavelength light source comprises micro-LEDs with wavelengths of 660nm and 940nm, which are used to emit light signals to penetrate the drainage fluid, and the photoelectric detection unit receives the transmitted light signals and converts them into electrical signals;

[0015] Microprocessor module: connected to the optical sensor module, used to collect electrical signals in real time and calculate hemoglobin concentration, perform dynamic trend analysis and multi-level alarm judgment;

[0016] Wireless communication module: connected to the microprocessor module, used to transmit monitoring data and alarm signals to external terminals;

[0017] The microprocessor module is configured as follows:

[0018] Perform moving average filtering and dynamic trend modeling on hemoglobin concentration data;

[0019] Trigger early warnings based on pre-set multi-level alarm conditions, including:

[0020] Level 1 alarm: Hemoglobin concentration increases by more than 20% within 5 minutes;

[0021] Level 2 alarm: The concentration continues to rise for 10 consecutive minutes and the cumulative increase is ≥30%;

[0022] Level 3 alarm: concentration increases rapidly by ≥40% within 3 minutes;

[0023] The graded alarm signals are sent to the medical staff terminal in real time through the wireless communication module.

[0024] Optionally, the optical sensing module includes:

[0025] Polycarbonate shell: Dimensions are 25mm long x 10mm wide x 5mm thick, with a central groove that fits over the transparent window of the drainage tube.

[0026] Quick clamp: located on the back of the housing, used to fix the module to the outside of the drainage tube;

[0027] Dual-wavelength LED light source and photodiode: symmetrically distributed in the groove, with a distance of 3-5mm between the light source and the detector to ensure effective reception of the light signal after penetrating the drainage fluid.

[0028] Optionally, the microprocessor module is further configured to:

[0029] The rate of change of hemoglobin concentration was calculated using the formula:

[0030]

[0031] Among them, C t is the current concentration, C t-Δt is the concentration at the previous time point, and Δt is the preset time interval;

[0032] Based on the rate of change and time series data, the bleeding risk is predicted using a dynamic trend model;

[0033] The microprocessor module performs calculation of the hemoglobin concentration change rate and trend prediction, specifically including:

[0034] Dynamic baseline calibration: Based on the hemoglobin concentration data of the first 2 hours after surgery, establish the patient's individualized baseline concentration C base , the calculation formula is:

[0035]

[0036] Where N is the number of data points within the first 2 hours after surgery (N=12, data sampling interval is 10 minutes), C t is the hemoglobin concentration measured at time point t;

[0037] Calculation of rate of change: Use the exponentially weighted moving average (EWMA) method to smooth the data and eliminate the interference of instantaneous fluctuations. The calculation formula is:

[0038] C smoothed =αC t +(1-α)C t-1 (α=0.7)

[0039] Among them, C smoothed is the smoothed hemoglobin concentration value, α is the smoothing coefficient, C t is the original concentration value at the current time point, C t-1 is the smoothed concentration value at the previous time point;

[0040] Calculate the rate of change based on the smoothed data:

[0041]

[0042] where ΔC is the percentage change in hemoglobin concentration relative to baseline;

[0043] Trend prediction model: Time series analysis (ARIMA model) is used to predict the concentration changes in the next 30 minutes. If the predicted value exceeds the baseline by 50% and meets the alarm conditions, an early warning will be triggered.

[0044] Optionally, the multi-level alarm condition further includes:

[0045] When a level 3 alarm is triggered, the hospital emergency system will be activated and the patient's location information will be pushed;

[0046] The alarm level is tied to the medical staff's response priority, and level 2 and above alarms require a pop-up reminder on the mobile app;

[0047] The multi-level alarm mechanism is tied to the clinical response strategy, including:

[0048] Level 1 alarm: A yellow alert is sent to the nurse station terminal, requiring a review of the drainage fluid within 30 minutes;

[0049] Level 2 alarm: A red pop-up window is pushed to the attending physician's mobile terminal, and the electronic medical record system is simultaneously activated to retrieve the patient's coagulation function data;

[0050] Level 3 alarm: Activates the hospital emergency system, automatically allocates an emergency bed and broadcasts the patient's location to the surgical team. At the same time, it controls the drainage tube solenoid valve to limit the drainage speed to avoid increased blood loss.

[0051] Optionally, the wireless communication module supports dual-mode transmission of low-power Bluetooth and Wi-Fi, and integrates medical Internet of Things protocols (such as HL7 and FHIR) to connect with the hospital's central monitoring system; the data transmission method of the wireless communication module includes:

[0052] Dual-mode redundant transmission: Data is uploaded to the hospital's private cloud via Wi-Fi first. If the signal is lost, it switches to the Bluetooth Mesh network and is relayed by the ward relay node.

[0053] Data encryption protocol: AES-256 is used to encrypt drainage fluid data packets, and the packet header is appended with the patient ID hash value (SHA-3 algorithm) to prevent information tampering;

[0054] Offline caching mechanism: The built-in Flash memory caches data for at least 72 hours when the network is disconnected, and automatically resumes the data when the connection is restored.

[0055] Optionally, the microprocessor module integrates a machine learning model to generate personalized warning thresholds through historical patient data training. The machine learning model is a lightweight convolutional neural network (CNN), and its training and execution process includes:

[0056] Input layer: Receives multidimensional time series data, including hemoglobin concentration, patient heart rate (obtained from the monitor via Bluetooth), and drainage fluid temperature (measured by thermocouples embedded in the tube wall);

[0057] Feature extraction layer: Use one-dimensional convolution kernel (window size = 5 minutes) to extract local trend features and capture long-term dependencies through LSTM layer;

[0058] Output layer: Generate personalized alarm threshold C alarm , the calculation formula is:

[0059] C alarm =0.8×C base +0.2×C pop

[0060] Among them, C alarm is the final personalized alarm threshold, C base is the patient's individual baseline concentration, C pop is the average risk threshold for a group of patients undergoing similar surgeries;

[0061] Incremental learning: After completing monitoring of every 100 patients, the model parameters are automatically updated and submitted to the doctor for review.

[0062] Optionally, the optical sensing module is further equipped with a near-infrared LED light source with a wavelength of 805 nm to correct for ambient light interference and improve hemoglobin detection accuracy, specifically by the following steps:

[0063] 1) Collect multi-wavelength transmitted light intensity and calculate absorbance ratio:

[0064]

[0065] Among them, A 660 : absorbance of light with a wavelength of 660 nm in drainage fluid; A 805 : absorbance of light with a wavelength of 805 nm in drainage fluid; a 940 : absorbance of light with a wavelength of 940 nm in the drainage fluid; R, R': absorbance ratio;

[0066] 2) Inverse hemoglobin concentration based on the ratio-concentration calibration curve;

[0067] The precision optimization method of the multi-wavelength detection includes:

[0068] Three-wavelength differential absorption detection: Calculate the normalized absorbance ratio using a 660nm / 940nm / 805nm light source combination:

[0069]

[0070] Among them, R1 and R2 are the normalized absorbance ratios used to eliminate ambient light interference;

[0071] Dynamic calibration database: pre-stores R1-R2 mapping tables at different hemoglobin concentrations and uses bilinear interpolation to match real-time data;

[0072] Elimination of hemolysis interference: When R2>1.2×R1, it is determined to be interference from free hemoglobin, and spectral resampling is initiated and abnormal data points are eliminated.

[0073] Optionally, the optical sensing module and the drainage tube are of a split design, the sensing module can be disassembled, disinfected and reused, and the drainage tube is a disposable sterile component; the split design specifically includes:

[0074] Quick connection interface: The groove of the optical sensor module is embedded with magnetic alignment pins, and ferromagnetic rings are embedded on both sides of the transparent window of the drainage tube;

[0075] The housing material can withstand 134°C high-temperature steam sterilization, and the LED light source and photodiode are encapsulated with medical epoxy resin.

[0076] Optionally, the system is further extended to be applied to chest drainage in thoracic surgery or pelvic drainage in gynecology. The cross-department extension function is achieved by:

[0077] Scenario-based algorithm switching:

[0078] Thoracic Surgery Mode: Added neutrophil elastase detection (calculated by the 940nm / 735nm absorbance ratio) for infection warning;

[0079] Gynecology Mode: Integrates pH sensor data. When pH > 7.5 and hemoglobin levels rise, it indicates the risk of fallopian tube damage.

[0080] Optionally, the optical sensing module integrates an adaptive ambient light compensation algorithm, specifically including:

[0081] The ambient light intensity is periodically collected when no drainage fluid is flowing;

[0082] The ambient light background noise is deducted from the real-time signal. The calculation formula is:

[0083] I corrected =I measured -I background

[0084] Among them, I corrected To indicate the corrected signal strength; I measured Indicates the measured signal strength; I background represents the background signal intensity;

[0085] The ambient light compensation is achieved through hardware-software collaboration:

[0086] Hardware layer: A narrow-band filter film is deposited on the surface of the photodiode, allowing only 660±5nm, 940±5nm, and 805±5nm light to pass through;

[0087] Software layer:

[0088] When there is no drainage fluid flowing, the ambient light background signal I is collected at a frequency of 100 Hz. ambient ;

[0089] The wavelet noise reduction algorithm (Daubechies 4 basis function) is used to separate the ambient light and the effective signal. The compensation formula is:

[0090] I corrected =I raw -0.9×I ambient ×e -t / 10 (t is the time interval since the last calibration)

[0091] Among them, I corrected : Corrected effective light intensity signal, I raw : The original light intensity signal received by the photodiode, I ambient : ambient light background intensity, t: time interval from the last ambient light calibration, e -t / 10 : Exponential decay factor.

[0092] Compared with the prior art, this application has at least one of the following beneficial technical effects:

[0093] This technology breaks through the limitations of traditional manual observation and enables continuous dynamic monitoring of hemoglobin concentration in peritoneal drainage fluid, promptly detecting early signs of bleeding. Through multi-wavelength differential absorption technology and a hemolysis interference elimination algorithm, it significantly improves detection accuracy and reduces interference from ambient light and non-target substances.

[0094] Personalized alarm thresholds are established based on machine learning models to automatically identify bleeding risk trends and avoid missed diagnoses and over-treatment. A multi-level alarm mechanism is integrated with the hospital information system to automate the entire response process, from notifying medical staff to dispatching emergency resources.

[0095] Reduce manual observation frequency, freeing up nursing resources for more critical patient care tasks. Digital recording and analysis capabilities support clinical decision-making and improve the efficiency of medical quality management.

[0096] The closed design avoids frequent contact with the drainage system, reducing the chance of nosocomial infection. The split structure facilitates disinfection and reuse, complying with hospital infection control regulations. Seamlessly connects to the hospital's existing information system to integrate and share patient data. Early warning of bleeding risks shortens the intervention window and reduces the incidence of postoperative complications. Objective data supports the development of personalized treatment plans and promotes rapid patient recovery.

[0097] This invention breaks through the limitations of traditional manual observation by accurately monitoring the hemoglobin concentration in peritoneal drainage fluid in real time, combining intelligent early warning and personalized response mechanisms, and significantly improves the timeliness and accuracy of bleeding risk identification. Its closed design reduces the risk of infection, and the system integration capability optimizes the medical workflow and realizes data sharing and remote collaboration. It not only shortens the intervention time window and improves patient prognosis, but also frees up nursing resources and improves medical quality, providing a comprehensive and efficient solution for postoperative monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0098] Figure 1 This is a structural diagram of a peritoneal drainage system based on real-time hemoglobin monitoring and intelligent early warning. DETAILED DESCRIPTION

[0099] The technical solution of the present invention is further described below in conjunction with specific embodiments.

[0100] Example 1

[0101] like Figure 1 As shown in FIG, the present invention proposes a peritoneal drainage system based on real-time hemoglobin monitoring and intelligent early warning, which includes a peritoneal drainage tube, an optical sensor module, a microprocessor module, and a wireless communication module. Each component is described in detail below.

[0102] 1. In this embodiment, the abdominal drainage tube is provided with a transparent measuring window for the flow of drainage fluid.

[0103] Second, an optical sensing module is non-contactly mounted outside the transparent measurement window and includes a dual-wavelength light source and a photoelectric detection unit. The dual-wavelength light source includes micro-LEDs with wavelengths of 660nm and 940nm, which are used to transmit light signals to penetrate the drainage fluid. The photoelectric detection unit receives the transmitted light signals and converts them into electrical signals. The optical sensing module includes:

[0104] Medical-grade, highly transparent polycarbonate housing: Dimensions are 25mm long x 10mm wide x 5mm thick, with a central groove that fits the transparent window of the drainage tube.

[0105] Quick clamp: located on the back of the housing, used to fix the module to the outside of the drainage tube;

[0106] Dual-wavelength LED light source and photodiode: symmetrically distributed in the groove, with a distance of 3-5mm between the light source and the detector to ensure effective reception of the light signal after penetrating the drainage fluid.

[0107] In this embodiment, the optical sensing module is further equipped with a near-infrared LED light source with a wavelength of 805 nm to correct for ambient light interference and improve the accuracy of hemoglobin detection. Specifically, the following steps are performed:

[0108] 1) Collect multi-wavelength transmitted light intensity and calculate absorbance ratio:

[0109]

[0110] Among them, A 660 : absorbance of light with a wavelength of 660 nm in drainage fluid; A 805 : absorbance of light with a wavelength of 805 nm in the drainage fluid; A 940 : absorbance of light with a wavelength of 940 nm in the drainage fluid; R, R': absorbance ratio;

[0111] 2) Inverse hemoglobin concentration based on the ratio-concentration calibration curve;

[0112] The precision optimization method of the multi-wavelength detection includes:

[0113] Three-wavelength differential absorption detection: Calculate the normalized absorbance ratio using a 660nm / 940nm / 805nm light source combination:

[0114]

[0115] Among them, R1 and R2 are the normalized absorbance ratios used to eliminate ambient light interference;

[0116] Dynamic calibration database: pre-stores R1-R2 mapping tables at different hemoglobin concentrations and uses bilinear interpolation to match real-time data;

[0117] Elimination of hemolysis interference: When R2>1.2×R1, it is determined to be interference from free hemoglobin, and spectral resampling is initiated and abnormal data points are eliminated.

[0118] It is worth noting that the optical sensor module and the drainage tube are split-type designs. The sensor module can be disassembled, disinfected and reused, and the drainage tube is a disposable sterile component. The split-type design specifically includes:

[0119] Quick connection interface: Magnetic alignment pins are embedded in the groove of the optical sensor module, and ferromagnetic rings are embedded on both sides of the transparent window of the drainage tube for millimeter-level precision self-alignment;

[0120] The shell material can withstand 134°C high-temperature steam sterilization, and the LED light source and photodiode are encapsulated with medical epoxy resin to ensure that the transmittance attenuation is less than 5% after 100 sterilizations.

[0121] In addition, the optical sensing module integrates an adaptive ambient light compensation algorithm, specifically including:

[0122] The ambient light intensity is periodically collected when no drainage fluid is flowing;

[0123] The ambient light background noise is deducted from the real-time signal. The calculation formula is:

[0124] I corrected =I measured -I background

[0125] Among them, I corrected To indicate the corrected signal strength; I measured Indicates the measured signal strength; I background represents the background signal intensity;

[0126] Ambient light compensation is achieved through hardware-software collaboration:

[0127] Hardware layer: A narrow-band filter film is deposited on the surface of the photodiode, allowing only 660±5nm, 940±5nm, and 805±5nm light to pass through;

[0128] Software layer:

[0129] When there is no drainage fluid flowing, the ambient light background signal I is collected at a frequency of 100 Hz. ambient ;

[0130] The wavelet noise reduction algorithm (Daubechies 4 basis function) is used to separate the ambient light and the effective signal. The compensation formula is:

[0131] I corrected =I raw -0.9×I ambient ×e -t / 10 (t is the time interval since the last calibration)

[0132] Among them, I corrected : Corrected effective light intensity signal, I raw : The original light intensity signal received by the photodiode, I ambient : ambient light background intensity, t: time interval from the last ambient light calibration, e -t / 10 : Exponential decay factor.

[0133] The optical sensing module utilizes a multi-wavelength light source (660nm / 940nm / 805nm) and a non-contact design to accurately measure drainage fluid hemoglobin concentration in real time. A narrowband filter and ambient light compensation algorithm effectively eliminate external interference. Combined with hemolysis interference detection technology, the module ensures stable and accurate signal acquisition. Its modular structure allows for quick installation on the transparent window of the drainage tube, eliminating the need for invasive procedures and reducing the risk of infection.

[0134] 3. A microprocessor module connected to the optical sensor module is used to collect electrical signals in real time and calculate hemoglobin concentration, perform dynamic trend analysis and multi-level alarm judgment; the microprocessor module is further configured as follows:

[0135] The rate of change of hemoglobin concentration was calculated using the formula:

[0136]

[0137] Among them, C t is the current concentration, C t-Δt is the concentration at the previous time point, and Δt is the preset time interval;

[0138] Based on the rate of change and time series data, the bleeding risk is predicted using a dynamic trend model;

[0139] The microprocessor module performs calculation of the hemoglobin concentration change rate and trend prediction, specifically including:

[0140] Dynamic baseline calibration: Based on the hemoglobin concentration data of the first 2 hours after surgery, establish the patient's individualized baseline concentration C base , the calculation formula is:

[0141]

[0142] Where N is the number of data points within the first 2 hours after surgery (N=12, data sampling interval is 10 minutes), C t is the hemoglobin concentration measured at time point t;

[0143] Calculation of rate of change: Use the exponentially weighted moving average (EWMA) method to smooth the data and eliminate the interference of instantaneous fluctuations. The calculation formula is:

[0144] μ smoothed =αC t +(1-α)C t-1 (α=0.7)

[0145] Among them, C smoothed is the smoothed hemoglobin concentration value, α is the smoothing coefficient, C t is the original concentration value at the current time point, C t-1 is the smoothed concentration value at the previous time point;

[0146] Calculate the rate of change based on the smoothed data:

[0147]

[0148] where ΔC is the percentage change in hemoglobin concentration relative to baseline;

[0149] Trend prediction model: Time series analysis (ARIMA model) is used to predict the concentration changes in the next 30 minutes. If the predicted value exceeds 50% of the baseline and meets the alarm conditions, the warning is triggered in advance. The microprocessor module integrates dynamic baseline calibration, exponentially weighted smoothing filtering and time series prediction algorithms to analyze the trend of hemoglobin concentration changes in real time and support personalized bleeding risk assessment. The machine learning model (CNN) generates exclusive alarm thresholds based on individual patient data to avoid the limitations of fixed thresholds; multi-level alarm logic (one / two / three levels) corresponds to different response strategies, from basic review to emergency activation, to achieve layered warning and precise intervention. In this embodiment, the microprocessor module is configured as follows:

[0150] Perform moving average filtering and dynamic trend modeling on hemoglobin concentration data;

[0151] Trigger early warnings based on pre-set multi-level alarm conditions, including:

[0152] Level 1 alarm: Hemoglobin concentration increases by more than 20% within 5 minutes;

[0153] Level 2 alarm: The concentration continues to rise for 10 consecutive minutes and the cumulative increase is ≥30%;

[0154] Level 3 alarm: concentration increases rapidly by ≥40% within 3 minutes;

[0155] The wireless communication module sends a graded alarm signal to the medical staff terminal in real time. The multi-level alarm conditions further include:

[0156] When a level 3 alarm is triggered, the hospital emergency system will be activated and the patient's location information will be pushed;

[0157] The alarm level is tied to the medical staff's response priority, and level 2 and above alarms require a pop-up reminder on the mobile app;

[0158] The multi-level alarm mechanism is tied to the clinical response strategy, including:

[0159] Level 1 alarm: A yellow alert is sent to the nurse station terminal, requiring a review of the drainage fluid within 30 minutes;

[0160] Level 2 alarm: A red pop-up window is pushed to the attending physician's mobile terminal, and the electronic medical record system is simultaneously activated to retrieve the patient's coagulation function data;

[0161] Level 3 alarm: Activates the hospital emergency system, automatically allocates an emergency bed and broadcasts the patient's location to the surgical team. At the same time, it controls the drainage tube solenoid valve to limit the drainage speed to avoid increased blood loss.

[0162] It is worth noting that the microprocessor module integrates a machine learning model to generate personalized warning thresholds through historical patient data training. The machine learning model is a lightweight convolutional neural network (CNN). Its training and execution process includes:

[0163] Input layer: Receives multidimensional time series data, including hemoglobin concentration, patient heart rate (obtained from the monitor via Bluetooth), and drainage fluid temperature (measured by thermocouples embedded in the tube wall);

[0164] Feature extraction layer: Use one-dimensional convolution kernel (window size = 5 minutes) to extract local trend features and capture long-term dependencies through LSTM layer;

[0165] Output layer: Generate personalized alarm threshold C alarm , the calculation formula is:

[0166] C alarm =0.8×C base +0.2×C pop

[0167] Among them, C alarm is the final personalized alarm threshold, C base is the patient's individual baseline concentration, C pop is the average risk threshold for a group of patients undergoing similar surgeries;

[0168] Incremental learning: After completing monitoring of every 100 patients, the model parameters are automatically updated and submitted to the doctor for review.

[0169] 4. A wireless communication module connected to the microprocessor module is used to transmit monitoring data and alarm signals to an external terminal. The wireless communication module supports dual-mode transmission of low-power Bluetooth and Wi-Fi, integrates medical Internet of Things protocols (such as HL7 and FHIR), and connects to the hospital's central monitoring system. The data transmission method of the wireless communication module includes:

[0170] Dual-mode redundant transmission: Data is uploaded to the hospital's private cloud via Wi-Fi first. If the signal is lost, it switches to the Bluetooth Mesh network and is relayed by the ward relay node.

[0171] Data encryption protocol: AES-256 is used to encrypt drainage fluid data packets, and the packet header is appended with the patient ID hash value (SHA-3 algorithm) to prevent information tampering;

[0172] Offline caching mechanism: The built-in Flash memory caches at least 72 hours of data when the network is disconnected, and automatically resumes transmission after the connection is restored. The wireless communication module adopts dual-mode redundant transmission (Wi-Fi + Bluetooth Mesh) and medical-grade encryption protocol (AES-256 + SHA-3) to ensure the stability and security of data transmission. The offline caching function ensures that monitoring data is not lost in the event of a network disconnection, and automatically resumes transmission after the connection is restored; it seamlessly connects to the hospital information system through the HL7 / FHIR protocol, supports remote monitoring and real-time linkage with multiple terminals (medical care apps, central monitoring, emergency systems), and improves the efficiency of medical collaboration.

[0173] Example 2

[0174] 1. Overall system architecture

[0175] The peritoneal drainage system includes a peritoneal drainage tube, an optical sensor module, a microprocessor module, a wireless communication module, and an external terminal. The tube is made of medical silicone and features a 15mm long, 8mm wide transparent measurement window in the middle. Its inner wall is smooth to minimize fluid retention. A solenoid valve is integrated downstream, and the microprocessor module controls on / off and drainage speed.

[0176] 2. Implementation details of the optical sensing module

[0177] The optical sensor module is clamped to the outside of the drainage tube's transparent window by a spring-loaded slider mechanism on the back of a quick-release clamp. A magnetic alignment pin adheres to the tube's ferromagnetic ring for millimeter-level alignment. Within the medical-grade, highly transparent polycarbonate housing, a dual-wavelength LED light source and photodiode are symmetrically spaced 4mm apart. The light source emits 660nm, 940nm, and 805nm near-infrared light that penetrates the drainage fluid. The hardware layer uses narrowband filters to limit stray light, while the software layer collects ambient light background signals at a frequency of 100Hz in the absence of drainage fluid. This background noise is then subtracted using a wavelet noise reduction algorithm to ensure accurate acquisition of the transmitted light signal.

[0178] 3. Microprocessor module data processing flow

[0179] Dynamic baseline calibration: Within the first 2 hours after surgery, 12 sets of hemoglobin concentration data are collected at 10-minute intervals to calculate the individualized baseline concentration.

[0180] Where N is the number of data points within the first 2 hours after surgery, C t is the hemoglobin concentration measured at time point t, which serves as the benchmark for subsequent change rate calculations.

[0181] Signal smoothing and trend analysis: The exponentially weighted moving average (EWMA, α = 0.7) method was used to smooth the real-time concentration data to eliminate transient interference such as infusion shock and patient position changes. The smoothed data were subjected to time series analysis using the ARIMA model to predict the concentration change trend in the next 30 minutes.

[0182] Multi-level alarm trigger logic:

[0183] Level 1 alarm: When the hemoglobin concentration rises by more than 20% within 5 minutes ( ) and sends a yellow alert to the nurse station, prompting a recheck within 30 minutes.

[0184] Level 2 alarm: If the concentration continues to rise for 10 consecutive minutes and the cumulative increase is ≥30%, a red pop-up window will be triggered on the attending physician's mobile terminal, and coagulation function data (such as platelet count and prothrombin time) in the electronic medical record system will be retrieved simultaneously.

[0185] Level 3 alarm: If the concentration suddenly rises by 40% or more within 3 minutes, the hospital emergency system will be immediately activated, automatically allocating an emergency bed, broadcasting the patient's location to the surgical team, and controlling the solenoid valve to limit the drainage rate to 5 ml / min to delay blood loss.

[0186] 4. Wireless Communication and System Integration

[0187] The wireless communication module utilizes a dual-mode design with Bluetooth Low Energy and Wi-Fi. Monitoring data (including hemoglobin concentration, alarm level, and patient ID) is encrypted (AES-256 + SHA-3 hash) and uploaded to the hospital's private cloud via Wi-Fi. If the signal is interrupted, it switches to the Bluetooth Mesh network for transmission via the ward's relay node. Built-in Flash memory caches 72 hours of data during network outages and automatically resumes transmission upon reconnection. Using HL7, FHIR, and other medical IoT protocols, it connects to the hospital's central monitoring system and electronic medical record system in real time, enabling data sharing and coordinated response.

[0188] 5. Hardware Design and Reuse Mechanism

[0189] The optical sensor module and drainage tube utilize a separate structure. The sensor module housing is autoclavable at 134°C. The LED light source and photodiode are encapsulated with medical epoxy resin, achieving a transmittance loss of less than 5% after 100 sterilization cycles and making it reusable. The drainage tube is a disposable, sterile component, reducing the risk of infection. The quick-action clamp supports drainage tubes with diameters ranging from 6 to 12 mm, adaptively gripping with a spring-loaded slider to meet the specific specifications of drainage tubes used in various departments, including general surgery, thoracic surgery, and gynecology.

[0190] 6. Expanded Application Across Departments

[0191] Thoracic Surgery Mode: When the test object is pleural drainage fluid, the system automatically switches to the thoracic surgery algorithm, estimates the neutrophil elastase concentration through the 940nm / 735nm absorbance ratio, and combines the changes in hemoglobin concentration to simultaneously warn of bleeding and infection risks.

[0192] Gynecological mode: An integrated pH sensor monitors the pH value of pelvic drainage fluid. When the pH is greater than 7.5 and the hemoglobin concentration increases, it indicates possible fallopian tube damage and triggers a targeted examination process.

[0193] 7. Machine Learning Model Training and Update

[0194] The microprocessor module has a built-in lightweight CNN model, and the input data includes hemoglobin concentration time series data, heart rate signal synchronized with Bluetooth, and drainage fluid temperature measured by thermocouples on the tube wall. The model generates personalized alarm thresholds (C alarm =0.8×C base +0.2×C pop Among them, C alarm is the final personalized alarm threshold, C base is the patient's individual baseline concentration, C pop is the average risk threshold for patients undergoing similar surgeries). After every 100 patients are monitored, the model parameters are automatically updated and pushed to the doctor's terminal for review to ensure that the threshold is suitable for different patient groups.

[0195] 8. Workflow

[0196] After surgery, a drainage tube is implanted in the patient's abdominal cavity and connected to an external drainage bag. The optical sensing module is secured to the transparent window using a quick-release clamp. A dual-wavelength light source transmits light signals through the drainage fluid. A photodiode receives the transmitted light and converts it into an electrical signal. After ambient light compensation, this signal is transmitted to a microprocessor. The microprocessor calculates the hemoglobin concentration in real time, performs moving average filtering, EWMA smoothing, and dynamic baseline calibration, and uses the ARIMA model to predict trends. When the concentration changes meet the alarm conditions, the wireless communication module transmits an alarm signal of the corresponding level to an external terminal, triggering a medical response or emergency response process.

[0197] The above specific embodiments are merely several optional embodiments of the present invention. Based on the technical solutions of the present invention and the relevant inspirations of the above embodiments, those skilled in the art may make various alternative improvements and combinations to the above specific embodiments.

Claims

1. A peritoneal drainage system based on real-time hemoglobin monitoring and intelligent early warning, characterized in that: include: Abdominal drainage tube: equipped with a transparent measuring window for drainage fluid flow; Optical sensing module: non-contact mounted outside the transparent measurement window, comprising a dual-wavelength light source and a photoelectric detection unit; the dual-wavelength light source comprises micro-LEDs with wavelengths of 660nm and 940nm, which are used to emit light signals to penetrate the drainage fluid, and the photoelectric detection unit receives the transmitted light signals and converts them into electrical signals; Microprocessor module: connected to the optical sensor module, used to collect electrical signals in real time and calculate hemoglobin concentration, perform dynamic trend analysis and multi-level alarm judgment; Wireless communication module: connected to the microprocessor module, used to transmit monitoring data and alarm signals to external terminals; The microprocessor module is configured as follows: Perform moving average filtering and dynamic trend modeling on hemoglobin concentration data; Trigger early warnings based on pre-set multi-level alarm conditions, including: Level 1 alarm: Hemoglobin concentration increases by more than 20%; Level 2 alarm: The measured hemoglobin concentration continues to increase and the cumulative increase is ≥30%; Level 3 alarm: hemoglobin concentration increases rapidly by ≥40%; The graded alarm signals are sent to the medical staff terminal in real time through the wireless communication module.

2. The peritoneal drainage system based on real-time hemoglobin monitoring and intelligent early warning according to claim 1 is characterized in that: The optical sensing module includes: The polycarbonate shell has a groove in the center that fits the transparent window of the drainage tube; A quick clamp is provided on the back of the housing and is used to fix the module to the outside of the drainage tube; The dual-wavelength LED light source and photodiode are symmetrically distributed in the groove, and the distance between the light source and the detector is 3-5mm.

3. The peritoneal drainage system based on real-time hemoglobin monitoring and intelligent early warning according to claim 1, characterized in that: The microprocessor module is further configured to: The rate of change of hemoglobin concentration was calculated using the formula: Among them, C t is the current concentration, C t-Δt is the concentration at the previous time point, and Δt is the preset time interval; Based on the rate of change and time series data, the bleeding risk is predicted using a dynamic trend model; The microprocessor module performs calculation of the hemoglobin concentration change rate and trend prediction, specifically including: Dynamic baseline calibration: Measure hemoglobin concentration data in advance to establish a patient-specific baseline concentration C base , the calculation formula is: Where N is the number of data points within the first 2 hours after surgery, C t is the hemoglobin concentration measured at time point t; Calculation of rate of change: Use the exponentially weighted moving average method to smooth the data and eliminate the interference of instantaneous fluctuations. The calculation formula is: C smoothed =αC t +(1-α)C t-1 (α=0.7) Among them, C smoothed is the smoothed hemoglobin concentration value, α is the smoothing coefficient, C t is the original concentration value at the current time point, C t-1 is the smoothed concentration value at the previous time point; Calculate the rate of change based on the smoothed data: where ΔC is the percentage change in hemoglobin concentration relative to baseline; Trend prediction model: Time series analysis is used to predict the concentration changes in the next 30 minutes. If the predicted value exceeds the baseline by 50% and meets the alarm conditions, an early warning will be triggered.

4. The peritoneal drainage system based on real-time hemoglobin monitoring and intelligent early warning according to claim 1, characterized in that: The multi-level alarm conditions further include: When a level 3 alarm is triggered, the hospital emergency system will be activated and the patient's location information will be pushed; The alarm level is tied to the medical staff's response priority, and level 2 and above alarms require a pop-up reminder on the mobile app; The multi-level alarm mechanism is tied to the clinical response strategy, including: Level 1 alarm: Send a yellow alert to the nurse station terminal and recheck the drainage fluid; Level 2 alarm: A red pop-up window is pushed to the attending physician's mobile terminal, and the electronic medical record system is simultaneously activated to retrieve the patient's coagulation function data; Level 3 alarm: Activates the hospital emergency system, automatically allocates emergency beds, sends patient location, and controls the drainage tube solenoid valve to limit drainage speed.

5. The peritoneal drainage system based on real-time hemoglobin monitoring and intelligent early warning according to claim 1, characterized in that: The wireless communication module supports dual-mode transmission of low-power Bluetooth and Wi-Fi, integrates the medical Internet of Things protocol, and connects with the hospital's central monitoring system; the data transmission method of the wireless communication module includes: Dual-mode redundant transmission: Data is uploaded to the hospital's private cloud via Wi-Fi first. If the signal is lost, it switches to the Bluetooth Mesh network and is relayed by the ward relay node. Data encryption protocol: AES-256 is used to encrypt drainage fluid data packets, and the packet header is appended with a patient ID hash value to prevent information tampering; Offline caching mechanism: The built-in Flash memory caches data for at least 72 hours when the network is disconnected, and automatically resumes the data when the connection is restored.

6. The peritoneal drainage system based on real-time hemoglobin monitoring and intelligent early warning according to claim 1, characterized in that: The microprocessor module integrates a machine learning model that generates personalized warning thresholds through historical patient data training. The machine learning model is a lightweight convolutional neural network, and its training and execution process includes: Input layer: Receives multidimensional time series data, including hemoglobin concentration, patient heart rate, and drainage fluid temperature; Feature extraction layer: uses one-dimensional convolution kernels to extract local trend features and captures long-term dependencies through the LSTM layer; Output layer: Generate personalized alarm threshold C alarm , the calculation formula is: C alarm =0.8×C base +0.2×C pop Among them, C alarm is the final personalized alarm threshold, C base is the patient's individual baseline concentration, C pop is the average risk threshold for a group of patients undergoing similar surgeries; Incremental learning: After completing monitoring of every 100 patients, the model parameters are automatically updated and submitted to the doctor for review.

7. The peritoneal drainage system based on real-time hemoglobin monitoring and intelligent early warning according to claim 1, characterized in that: The optical sensing module is further equipped with a near-infrared LED light source with a wavelength of 805 nm to correct for ambient light interference and improve hemoglobin detection accuracy, specifically by the following steps: 1) Collect multi-wavelength transmitted light intensity and calculate absorbance ratio: Among them, A 660 : absorbance of light with a wavelength of 660 nm in drainage fluid; A 805 : absorbance of light with a wavelength of 805 nm in the drainage fluid; A 940 : absorbance of light with a wavelength of 940 nm in the drainage fluid; R, R': absorbance ratio; 2) Inverse hemoglobin concentration based on the ratio-concentration calibration curve; The precision optimization method of the multi-wavelength detection includes: Three-wavelength differential absorption detection: Calculate the normalized absorbance ratio using a 660nm / 940nm / 805nm light source combination: Among them, R1 and R2 are the normalized absorbance ratios used to eliminate ambient light interference; Dynamic calibration database: pre-stores R1-R2 mapping tables at different hemoglobin concentrations and uses bilinear interpolation to match real-time data; Elimination of hemolysis interference: When R2>1.2×R1, it is determined to be interference from free hemoglobin, and spectral resampling is initiated and abnormal data points are eliminated.

8. The peritoneal drainage system based on real-time hemoglobin monitoring and intelligent early warning according to claim 1, characterized in that: The optical sensor module and the drainage tube are of a split design. The sensor module can be disassembled, disinfected and reused, and the drainage tube is a disposable sterile component. The split design specifically includes: Quick connection interface: The groove of the optical sensor module is embedded with magnetic alignment pins, and ferromagnetic material rings are embedded on both sides of the transparent window of the drainage tube; The housing material can withstand 134°C high-temperature steam sterilization, and the LED light source and photodiode are encapsulated with medical epoxy resin.

9. The peritoneal drainage system based on real-time hemoglobin monitoring and intelligent early warning according to claim 1, characterized in that: The system is further extended to thoracic surgery chest drainage or gynecological pelvic drainage scenarios. The cross-department expansion function is achieved in the following ways: Scenario-based algorithm switching: Thoracic surgery mode: Added detection of neutrophil elastase for infection warning; Gynecology Mode: Integrates pH sensor data. When pH > 7.5 and hemoglobin levels rise, it indicates the risk of fallopian tube damage.

10. The peritoneal drainage system based on real-time hemoglobin monitoring and intelligent early warning according to claim 1, characterized in that: The optical sensing module integrates an adaptive ambient light compensation algorithm, specifically including: The ambient light intensity is periodically collected when no drainage fluid is flowing; The ambient light background noise is deducted from the real-time signal. The calculation formula is: I corrected =I measured -I background Among them, I corrected To indicate the corrected signal strength; I measured Indicates the measured signal strength; I background represents the background signal intensity; The ambient light compensation is achieved through hardware-software collaboration: Hardware layer: a narrow-band filter film is deposited on the surface of the photodiode to allow light of 660±5nm, 940±5nm, and 805±5nm to pass through; Software layer: When there is no drainage fluid flowing, the ambient light background signal I is collected at a frequency of 100 Hz. ambient ; The wavelet noise reduction algorithm is used to separate the ambient light and the effective signal. The compensation formula is: I corrected =I raw -0.9×I ambient ×e -t / 10 Among them, I corrected : Corrected effective light intensity signal, I raw : The original light intensity signal received by the photodiode, I ambient : ambient light background intensity, t: time interval from the last ambient light calibration, e -t / 10 : Exponential decay factor.

Images