Intelligent cerebrospinal fluid drainage monitoring system and monitoring method

Abstract

The invention relates to the technical field of medical instruments, in particular to an intelligent cerebrospinal fluid drainage monitoring system and method. The system comprises a drainage catheter, a drainage bag and a monitoring terminal, a high-precision pressure sensor, an ultrasonic flow sensor, a data acquisition and transmission module and an intelligent monitoring decision module are integrated, and corresponding pressure analog signals are acquired, calibrated and output in real time through the high-precision pressure sensor; a flow analog signal corresponding to compensation output is collected in real time through an ultrasonic flow sensor; synchronously acquiring the pressure analog signal and the flow analog signal through an ADC (Analog to Digital Converter) through a data acquisition and transmission module to generate pressure data and flow data, and transmitting the pressure data and the flow data to a monitoring terminal; and the monitoring terminal calculates the average sliding pressure and the drainage efficiency and decides to generate a multi-stage alarm signal corresponding to cerebrospinal fluid drainage. Accurate data support can be provided for neurosurgery cerebrospinal fluid drainage assistance by means of a high-precision sensor and intelligent monitoring processing.

Description

Technical Field

[0001] The present invention relates to the technical field of medical devices, and in particular to an intelligent cerebrospinal fluid drainage monitoring system and a monitoring method. Background Art

[0002] Cerebrospinal fluid drainage technology is widely used to treat neurological diseases such as hydrocephalus, intraventricular hemorrhage, and increased intracranial pressure. It primarily involves draining excess cerebrospinal fluid through a drainage device to relieve intracranial pressure and restore normal brain function. Currently, traditional monitoring relies on medical staff manually observing the cerebrospinal fluid in the drainage tube at regular intervals, measuring the drainage volume using a simple ruler, and relying on experience to determine whether the cerebrospinal fluid pressure is normal.

[0003] In addition, similar patents such as CN113577418A disclose an intelligent cerebrospinal fluid external drainage monitoring system, including a processing module, a flow rate control device, a flow sensor, a pressure sensor, a temperature sensor, a micro camera, a communication module, a medical information recognition device and a human-computer interaction module; the processing module is electrically connected to the flow rate control device, the flow sensor, the pressure sensor, the temperature sensor, the micro camera, the communication module, the alarm device, the medical information recognition device and the human-computer interaction module respectively. This application can select different drainage modes according to individual differences of patients and their condition, realize intelligent control of drainage status, and improve the safety of cerebrospinal fluid drainage. It has the advantages of high precision, high accuracy, high pertinence and high reliability, which improves the level of medical informationization, reduces the workload of nursing, and improves the efficiency of hospital management. It has great clinical value and social significance. However, it only has basic flow rate and pressure monitoring and alarm functions, and its monitoring accuracy is low. It cannot dynamically compensate for output errors caused by environmental changes. For example, traditional pressure sensors are not accurate enough and are prone to errors due to temperature fluctuations, sensor aging and other factors. The measurement accuracy of flow sensors has not met the ideal requirements. At the same time, it also lacks intelligent diagnosis and early warning mechanisms, and cannot promptly remind medical staff to intervene when abnormal situations occur, thereby reducing the efficiency of abnormal early warning. Summary of the Invention

[0004] In order to solve the above-mentioned technical problems existing in existing cerebrospinal fluid drainage intelligent devices, the present invention provides an intelligent cerebrospinal fluid drainage monitoring system with high monitoring accuracy, high measurement accuracy and high early warning efficiency, including a drainage catheter, a drainage bag and a monitoring terminal. The system also integrates a high-precision pressure sensor, an ultrasonic flow sensor, a data acquisition and transmission module and an intelligent monitoring decision module. The intelligent monitoring decision module is composed of the monitoring terminal and the hospital information system, wherein:

[0005] The high-precision pressure sensor is a piezoresistive sensor and is installed 3-5 cm from the end of the drainage catheter corresponding to the ventricle to output a pressure analog signal through an energized bridge conversion;

[0006] The ultrasonic flow sensor uses a piezoelectric ceramic ultrasonic transducer and is installed in the straight section of the drainage catheter corresponding to 5-10 cm downstream of the high-precision pressure sensor. The flow-temperature compensation relationship formula v = 1402.5 + 4.9T - 0.05T is used. 2 Perform dynamic compensation to output flow analog signal, where v is flow rate and T is temperature;

[0007] The data acquisition and transmission module integrates an analog-to-digital converter (ADC) to synchronously acquire the corresponding pressure analog signal and flow analog signal to generate pressure data and flow data. It supports wireless dual-channel transmission using Wi-Fi as the main channel and Bluetooth as the backup channel, and uses AES-256 encryption and automatic retransmission mechanism for redundant transmission to the monitoring terminal.

[0008] The intelligent monitoring and decision-making module uses pressure data and flow data to calculate drainage efficiency in real time; by configuring a random forest machine learning model and dynamically adjusting the pressure threshold according to the hospital information system, a multi-level alarm signal corresponding to cerebrospinal fluid drainage is generated based on the pressure threshold and drainage efficiency decision.

[0009] The intelligent cerebrospinal fluid drainage monitoring system proposed in the present invention generally includes a drainage catheter, a drainage bag and a monitoring terminal, and is also integrated with a high-precision pressure sensor, an ultrasonic flow sensor, a data acquisition and transmission module, and an intelligent monitoring decision module. Compared with the existing technology, the beneficial effect of the present application is that it uses a piezoresistive sensor to measure the cerebrospinal fluid drainage pressure, which can not only measure the pressure of the cerebrospinal fluid drainage system with high precision, but also ensure that accurate data can be provided in various common clinical applications through its measurement range of 0-200mmH2O and resolution of 0.1mmH2O. It also performs dynamic calibration through an energized bridge, which helps to eliminate the zero drift of the sensor itself and the performance degradation caused by long-term use, ensuring the accuracy and reliability of the data. The sensor is installed 3-5cm from the drainage catheter, so that it can accurately reflect the actual pressure state of the ventricular end, greatly reducing the error caused by unstable sensor performance. Secondly, a piezoelectric ceramic ultrasonic transducer is used to measure the flow of cerebrospinal fluid drainage, which has high accuracy (±1mL / h) and a wide measurement range (0-500mL / h). It is installed on the straight section of the drainage catheter 5-10cm downstream of the high-precision pressure sensor to ensure its accurate capture of the drainage flow, so that any flow changes can be fed back in time. At the same time, the compensation relationship formula v=1402.5+4.9T-0.05T is used. 2Dynamic compensation is performed to make flow measurement more accurate under different temperature conditions, eliminating the interference of temperature changes on measurement, thereby improving the reliability of flow data and providing real-time flow data. Then, by integrating a 16-bit ADC for synchronous data acquisition at a 1000Hz sampling rate, it is possible to collect analog signals of pressure and flow at a high frequency, and has a signal-to-noise ratio of >90dB, making the collected data clearer and more accurate. On this basis, the system supports dual-channel wireless transmission of Wi-Fi and Bluetooth, which can provide stable network connections in different environments, ensuring that data can be transmitted quickly and accurately to the monitoring terminal. The use of the AES-256 encryption algorithm also ensures data security during transmission, and the automatic retransmission mechanism ensures that in the event of signal loss or interference, data can be effectively restored and no important monitoring information will be lost. This can provide instant feedback on dynamic changes in drainage speed, thereby ensuring the security, accuracy and efficiency of data transmission. Finally, by calculating the sliding average pressure and drainage efficiency in real time, combined with dynamic adjustment of the pressure threshold, the system can ensure that it can respond flexibly according to different clinical situations. When the system detects abnormal fluctuations in pressure data and flow data, it can immediately analyze and trigger multi-level alarm signals to promptly remind medical staff to take necessary intervention measures. Through the introduction of the random forest machine learning model, the entire system can learn and optimize the pressure threshold through historical data, automatically adapt to the individual differences of different patients, thereby improving the system's adaptability to different clinical scenarios. The biggest advantage of this step lies in the intelligent dynamic adjustment and alarm mechanism, which can improve the system's abnormal warning efficiency and accuracy while ensuring data accuracy.

[0010] Preferably, the high-precision pressure sensor further includes:

[0011] Biocompatible titanium alloy shell with a parylene moisture barrier coating on the inner surface;

[0012] The self-test circuit detects the impedance change corresponding to the energized bridge every 15-30 minutes, and triggers a calibration request when the impedance change deviation is greater than 5%, wherein the calibration request is performed once every 7 days through the corresponding regular calibration mechanism, wherein the calibration operation is calibrated according to the calibration formula, P 实际 =k×P 原始 +b; where P 实际 is the calibrated pressure source, k is the calibration coefficient, P 原始 is the original pressure source, b is the offset;

[0013] The optimized pressure transmission structure includes a 2mm diameter silicone buffer diaphragm, which has been verified by CFD simulation to reduce the measurement error caused by cerebrospinal fluid drainage to ≤0.5mmH2O.

[0014] The biocompatible titanium alloy shell of the present invention has excellent mechanical properties and biocompatibility, can effectively resist corrosion in the internal environment, and ensure the stability and safety of long-term implantation in the body. At the same time, the good affinity of titanium alloy with human tissue makes it widely used in the biomedical field, which can provide strong physical protection to prevent the impact of external physical impact or chemical erosion in the body on internal electronic components. The polyparaxylene coating has good insulation properties, which helps to improve the electrical stability of the device and reduce circuit interference. The design of the self-test circuit is one of the key factors to ensure the accurate measurement and stable performance of the device. It performs a self-test every 15-30 minutes, and detects the impedance change corresponding to the energized bridge in real time, which helps to promptly detect any abnormal conditions in the circuit and ensure that the device is always in the best working state. If the deviation of the impedance change exceeds 5%, the self-test circuit will automatically trigger a calibration request. This mechanism can ensure that the device maintains accuracy during use and avoid measurement errors caused by environmental changes or wear during use. The purpose of the pressure conduction optimization structure design is to solve the measurement error problem caused by cerebrospinal fluid drainage. The 2mm diameter silicone buffer diaphragm plays a vital role in this structure. The silicone material is highly elastic and flexible, and can adapt to complex environmental changes in the body, ensuring the accurate transmission of pressure signals during the measurement process. Verified by CFD simulation, the design successfully controls the measurement error caused by cerebrospinal fluid drainage within 0.5mmH2O. This optimized design effectively improves the measurement accuracy of the equipment while reducing the risk of misdiagnosis or delayed treatment due to measurement errors.

[0015] Preferably, the self-test circuit further comprises measuring the impedance response difference ΔR of the energized bridge under 0.1mA / 1mA dual current excitation and calculating the corresponding temperature drift compensation amount. Where α represents the temperature drift coefficient of the material corresponding to the energized bridge, R0 represents the internal resistance of the energized bridge, and a calibration request can also be triggered when ΔT>2°C.

[0016] The present invention can effectively improve the accuracy and stability of the circuit by measuring the impedance response difference of the energized bridge under 0.1mA / 1mA dual current excitation and calculating the corresponding temperature drift compensation. Specifically, temperature changes will have a significant impact on the impedance of the energized bridge. Especially in precision measurement applications, temperature drift will cause errors in the measured value, thereby affecting the final result. Therefore, by measuring the impedance response difference of the energized bridge under 0.1mA / 1mA dual current excitation, the impact of temperature changes on the bridge measurement results can be accurately captured, and the temperature drift compensation can be further calculated. This compensation is calculated based on the temperature drift coefficient and the internal resistance characteristics of the energized bridge, ensuring that the circuit can compensate for the errors caused by temperature changes in real time. In addition, when the system detects that the temperature change exceeds 2°C, it can trigger a calibration request. This trigger mechanism can effectively avoid the accumulation of measurement errors caused by sudden temperature changes. The combination of temperature drift compensation and calibration requests can further improve the adaptability and reliability of the self-test circuit, avoid the expansion of errors, and thus ensure long-term stable operation.

[0017] Preferably, the ultrasonic flow sensor further includes:

[0018] The anti-interference unit uses wavelet transform technology, including Daubechies4 wavelet basis function, to separate the effective signal component and the noise signal component corresponding to the flow simulation signal to complete the noise reduction operation;

[0019] The dynamic calibration unit automatically performs zero-point calibration every 12-24 hours to temporarily block the connection between the drainage catheter and the drainage bag through the solenoid valve during the calibration process, and records the zero flow reference value ε0 corresponding to the ultrasonic flow sensor. At the same time, 32 sampling values are collected in the zero flow state as a moving median filter to eliminate abnormal values with an amplitude > 3ε0.

[0020] The present invention adopts wavelet transform technology, especially Daubechies4 wavelet basis function, to complete the noise reduction operation by separating the effective signal components and noise signal components in the flow simulation signal. Wavelet transform is an efficient signal processing technology that can analyze signals in the time domain and frequency domain simultaneously. It is particularly suitable for processing non-stationary signals. The flow simulation signal usually contains multiple different frequency components, some of which come from equipment noise or external interference. These noise signals will have a negative impact on the accuracy of the system. By using the Daubechies4 wavelet basis function, the effective components and noise components in the signal can be effectively separated. This method can accurately extract the real changes in the flow signal by decomposing and reconstructing the signal, while effectively suppressing the influence of external environment or system noise. This not only improves the quality of the signal, but also reduces the measurement error caused by noise. At the same time, by utilizing the dynamic calibration unit to automatically perform zero-point calibration every 12-24 hours, the accuracy of the equipment during long-term use is ensured. Zero-point calibration is a common calibration method designed to eliminate drift and deviation of the sensor system and ensure that the measurement system can always provide accurate reference values. Temporarily blocking the connection between the drainage catheter and the drainage bag through the solenoid valve can avoid affecting the calibration process when the flow changes, ensuring that the sensor is in a zero-flow state during calibration, thereby improving the stability and accuracy of the signal.

[0021] Preferably, the noise reduction operation is performed by setting a corresponding dynamic threshold, wherein the dynamic threshold is specifically Wherein σ represents the standard deviation corresponding to the noise signal component, and N represents the signal length corresponding to the flow simulation signal.

[0022] The present invention optimizes the signal processing process by setting a dynamic threshold, wherein the setting of the dynamic threshold is based on the standard deviation of the noise signal component and the signal length of the flow simulation signal. This strategy not only improves the flexibility and adaptability of signal processing, but also effectively enhances the anti-interference ability of the system in complex environments. By adopting the dynamic threshold setting method, the noise reduction operation can be automatically adjusted according to the changes in the current signal, ensuring that the signal processing always adapts to changes in the environment. In specific implementation, the calculation of the dynamic threshold takes into account the standard deviation of the noise signal component and the signal length of the flow simulation signal. The standard deviation of the noise signal component reflects the fluctuation amplitude of the signal. The larger the standard deviation, the more severe the noise fluctuation. The noise reduction operation needs to be processed more sensitively. The signal length is related to the duration and frequency of the flow simulation signal. A longer signal length means that the signal changes slower, and the influence of noise requires a smoother noise reduction operation, thereby optimizing the effect of signal noise reduction.

[0023] Preferably, the synchronous acquisition is specifically to align the corresponding pressure analog signal and flow analog signal through time stamps with an accuracy of ±0.1ms.

[0024] The synchronous acquisition operation of the present invention aligns the corresponding pressure analog signal and flow analog signal through time stamps with an accuracy of ±0.1 milliseconds. The core goal of this design is to ensure the temporal consistency of the pressure signal and the flow signal so that subsequent data analysis, processing and fault diagnosis can be more accurate. In complex flow and pressure monitoring systems, the temporal relationship between the flow signal and the pressure signal is crucial. Flow and pressure often have a mutually dependent relationship. Changes in flow will directly affect pressure, and pressure fluctuations will affect the stability of flow. Therefore, if there is a time deviation between the signals, it will lead to erroneous data interpretation, which will in turn affect the performance evaluation and fault diagnosis of the system. By setting the timestamp alignment with an accuracy of ±0.1 milliseconds, the synchronous acquisition of the two signals is ensured, so that the instantaneous change relationship between them can be accurately captured during signal analysis.

[0025] Preferably, when the wireless dual-channel transmission mode is subjected to electromagnetic interference and causes the corresponding packet loss rate to be greater than 5%, the system switches to the backup channel within 0.5 seconds or less.

[0026] The wireless dual-channel transmission method of the present invention can switch to the backup channel within 0.5 seconds or less when the packet loss rate exceeds 5% due to electromagnetic interference. This design significantly improves the reliability and stability of the system in harsh environments, especially when wireless communication is frequently subject to electromagnetic interference. Electromagnetic interference (EMI) is a common challenge in modern wireless communications, especially in industrial environments or when there are a large number of electrical devices around the equipment. EMI can cause wireless signal loss or transmission errors. A wireless transmission packet loss rate exceeding 5% generally indicates poor transmission quality, resulting in incomplete or distorted data, which has a significant negative impact on the system's real-time monitoring and data accuracy. This step, through the wireless dual-channel transmission method, allows the system to automatically switch to the backup channel within 0.5 seconds or less when the packet loss rate on the main channel exceeds 5%. This switching speed is critical, enabling communication to be restored in an extremely short time, avoiding system stalls or malfunctions caused by data loss. After switching to the backup channel, the system can continue to transmit data seamlessly, ensuring the continuous and smooth flow of monitoring data and avoiding the impact of communication interruptions.

[0027] Preferably, the intelligent monitoring decision module includes:

[0028] By dynamically setting the corresponding window length and smoothing the instantaneous pressure fluctuations of the pressure data based on the window length, the pressure data after fluctuation smoothing is generated;

[0029] By setting a sliding interval corresponding to 3-5 minutes and performing a sliding average calculation on the pressure data after fluctuation smoothing based on the sliding interval, the sliding average pressure P is obtained.avg ;

[0030] The instantaneous drainage volume Q of the cerebrospinal fluid is extracted from the corresponding flow data through the time end point corresponding to the sliding interval;

[0031] According to the instantaneous drainage volume Q and the sliding average pressure P avg The drainage efficiency is calculated in real time based on the ratio between

[0032] By configuring a random forest machine learning model in the monitoring terminal and dynamically adjusting the pressure threshold according to the patient's age, weight and historical data in the hospital information system, the pressure threshold is combined with the sliding average pressure P avg And the drainage efficiency v decision generates a multi-level alarm signal corresponding to cerebrospinal fluid drainage.

[0033] The present invention effectively eliminates noise and interference in instantaneous data and reduces unnecessary fluctuations by dynamically setting the window length and smoothing the instantaneous pressure fluctuations. Pressure signals are often affected by various factors (such as equipment errors or external interference), which can lead to data instability. Through smoothing, the true change trend of the signal can be better extracted, providing a more stable basis for subsequent analysis and avoiding the impact of error accumulation on system performance. By setting a sliding interval of 3-5 minutes and combining it with a sliding average calculation, it is possible to smooth out sudden fluctuations and periodic interference on a longer time scale, ensuring that reliable pressure data is always obtained in a changing clinical environment. The sliding average pressure, as an expression of the long-term trend, can reflect a more stable pressure state in the system, thereby better comparing drainage efficiency and monitoring pressure changes, and avoiding misjudgment caused by single data fluctuations. Secondly, by extracting the instantaneous drainage volume from the flow data, the drainage status of cerebrospinal fluid can be obtained in real time. The change in the drainage volume of cerebrospinal fluid directly reflects the effectiveness of the drainage process. The extraction of the instantaneous drainage volume helps the system capture the real-time dynamics of drainage more accurately. This step provides the necessary input data for calculating drainage efficiency and lays the foundation for real-time monitoring and subsequent decision-making. Then, drainage efficiency is calculated in real time based on the ratio between instantaneous drainage volume and sliding average pressure, enabling the system to more comprehensively assess the effectiveness of cerebrospinal fluid drainage. Drainage efficiency is calculated by effectively comparing pressure with flow to assess whether the current drainage is meeting the standard. By calculating this ratio, the system can reflect problems in the drainage process in real time, promptly identify insufficient or excessive drainage efficiency, and notify medical staff through the corresponding alarm system to intervene or adjust the treatment plan, thereby avoiding complications caused by excessive or insufficient drainage. Finally, by configuring a random forest machine learning model in the monitoring terminal and dynamically adjusting the pressure threshold based on the patient's age, weight, and historical data, it can more intelligently adapt to the individual differences of each patient. Different physiological characteristics of patients require different pressure thresholds. Through dynamic adjustment, personalized alarm thresholds can be set for each patient to ensure the safety and efficiency of the drainage process. This can accurately determine whether an alarm needs to be triggered, reducing the probability of false alarms and missed alarms.

[0034] As an advantage, the window length corresponding to the dynamic setting is specifically the window length corresponding to the initial 10 sampling points, and the corresponding pressure change rate is calculated based on the window length. Where ΔP represents the pressure change value, Δt represents the window length, and the corresponding pressure change rate is determined at the same time. When , the corresponding window length is shortened to 5 sampling points, otherwise it is kept at 10 sampling points.

[0035] The present invention sets the initial window length to 10 sampling points. This is to ensure that sufficient historical data can be obtained in the initial stage, so that the calculation of pressure changes is smoother and more accurate. By setting the window length, it is possible to balance the relationship between full consideration of past data and real-time performance. The window length of 10 sampling points provides a relatively stable data basis for the subsequent pressure change rate calculation, reducing the fluctuation error caused by too few sampling points. Next, based on the calculated pressure change rate, the system will make a judgment. Specifically, when the pressure change rate is greater than 5mmH2O / min, the window length will be shortened to 5 sampling points. The core purpose of this adjustment is to improve the response speed of the system. When the pressure changes rapidly, shortening the window length allows the system to react more quickly, avoiding the lag effect introduced by too long historical data, which leads to slow adjustment of the system. For example, in some sudden pressure fluctuations, a quick response can effectively reduce potential risks or losses. Therefore, shortening the window length is to improve the system's adaptability to rapid changes. On the other hand, if the calculated pressure change rate is less than or equal to 5 mmH2O / min, the system maintains the original window length, that is, 10 sampling points. This indicates that the current pressure change is relatively stable and does not require an overly sensitive response. Maintaining a longer window length can provide more historical data, making the pressure change trend smoother.

[0036] Preferably, the process of dynamically adjusting the pressure threshold is specifically as follows:

[0037] Obtaining the patient's age, weight, and historical data from the hospital information system, wherein the historical data includes a sample of similar historical cases (n ≥ 1000);

[0038] The corresponding initial pressure threshold is determined to be 70-200 mmH2O according to the patient's age and weight in the hospital information system;

[0039] By configuring a random forest machine learning model in the monitoring terminal with a tree depth of 10, and inputting historical similar case samples n≥1000 into the corresponding random forest machine learning model for model training, and at the same time comparing and calculating the confidence between historical similar cases >85%, the corresponding pressure threshold is dynamically adjusted according to the corresponding adjustment range ≤5mmH2O.

[0040] The present invention has important clinical significance by obtaining the age, weight and historical data of patients in the hospital information system (including historical similar case samples ≥1000 cases). First of all, the age and weight of the patient are important factors affecting the setting of the pressure threshold. Patients of different age groups and weights have different tolerance to pressure during the drainage process, so personalized pressure thresholds can ensure that patients are drained within a safe range. The existence of these historical case data allows the model to learn more actual situations, thereby avoiding the deviation of a single case and making the prediction results more accurate and representative. In addition, a sufficient number of historical samples (≥1000 cases) ensures the breadth and effectiveness of the data, which helps to improve the reliability and accuracy of the entire system. By determining the initial pressure threshold (70-200mmH2O) based on the patient's age and weight, personalized management of each patient is ensured. Age and weight are key factors in determining the patient's physiological condition and tolerance to pressure. Different age and weight ranges will affect the patient's physiological response. Therefore, setting an initial pressure threshold based on this information provides a preliminary, safe pressure range for each patient. This initial threshold ensures that pressure remains at an appropriate level during CSF drainage, avoiding complications caused by excessively high or low pressure and reducing safety risks associated with standardized thresholds. Finally, configuring a random forest machine learning model to dynamically adjust the pressure threshold further enhances the system's intelligence and accuracy. As a powerful ensemble learning algorithm, random forests can handle complex nonlinear relationships and extract underlying patterns from a large number of historical case histories. Setting a tree depth of 10 allows the model to process more feature layers, thereby more meticulously capturing the complex relationships between patient characteristics and pressure thresholds. By training on a sample of similar historical cases, the model is able to make accurate predictions and decisions based on a large amount of data. Importantly, when the confidence level between similar historical cases exceeds 85%, the system dynamically adjusts the pressure threshold by an adjustment of ≤5 mmH2O. This results in a more precise pressure threshold for each patient, reducing errors and risks during drainage and maximizing patient safety.

[0041] As an advantage, the pressure threshold is combined with the sliding average pressure P avg The drainage efficiency η decision generates multi-level alarm signals corresponding to cerebrospinal fluid drainage, including:

[0042] When the single sliding mean pressure P avg When the pressure exceeds the threshold and lasts for 5-10 seconds, the decision is triggered to generate a yellow first-level prompt alarm signal corresponding to cerebrospinal fluid drainage;

[0043] When the sliding mean pressure P avgWhen the pressure threshold is exceeded and the drainage efficiency η drops sharply to zero, it is automatically determined that the drainage catheter is blocked, and an immediate decision is made to trigger the generation of a secondary emergency alarm signal corresponding to cerebrospinal fluid drainage, which is red.

[0044] The present invention triggers a first-level prompt alarm signal (yellow) when a single sliding average pressure exceeds the pressure threshold and lasts for 5-10 seconds, effectively realizing the monitoring and reminder of abnormal pressure fluctuations. The sliding average pressure is a smooth pressure data representation method that can filter out the noise in the instantaneous fluctuation, so that the system can accurately capture the real pressure change trend. When the pressure continues to exceed the threshold and maintains for a period of time (5-10 seconds), the system will issue a yellow prompt alarm signal. This mechanism enables the system to identify potential risks in advance and promptly notify medical staff to take appropriate measures to avoid adverse effects on patients due to excessive pressure during drainage. The yellow alarm signal serves as a first-level prompt alarm signal to remind medical staff that they need to pay attention, but it will not cause excessive tension or interference, thereby effectively ensuring the safety of patients without disrupting clinical work. Then, by monitoring when the sliding average pressure exceeds the pressure threshold and the drainage efficiency drops suddenly to zero, it automatically determines that the drainage catheter is blocked and triggers the secondary emergency alarm signal (red), which can immediately take emergency measures when serious abnormalities occur. A sudden drop in drainage efficiency to zero usually means catheter blockage or other serious problems. In this case, immediate response is crucial. Through intelligent judgment and the triggering of the secondary emergency alarm, the system can quickly notify medical staff to prevent patients from developing further complications due to poor cerebrospinal fluid drainage. The setting of this red alarm signal ensures that when the system detects a critical situation, it can be handled as a priority to ensure that patients receive timely intervention in the shortest time.

[0045] Preferably, the second technical solution of the present invention is an intelligent cerebrospinal fluid drainage monitoring method, which is implemented based on the intelligent cerebrospinal fluid drainage monitoring system as described above. The intelligent cerebrospinal fluid drainage monitoring method includes:

[0046] S01: Real-time acquisition and dynamic calibration of the corresponding pressure analog signal are performed through a high-precision pressure sensor;

[0047] S02: Using an ultrasonic flow sensor to collect data in real time and perform wavelet denoising and temperature compensation to output a flow simulation signal corresponding to cerebrospinal fluid drainage;

[0048] S03: The output pressure analog signal and flow analog signal are synchronously collected by a high-performance and high-precision analog-to-digital converter ADC to generate corresponding pressure data and flow data, and then transmitted to the monitoring terminal;

[0049] S04: The sliding average processing is performed on the monitoring terminal to calculate the sliding average pressure and drainage efficiency, and the pressure threshold is dynamically adjusted by configuring the random forest machine learning model. Based on the pressure threshold combined with the sliding average pressure and drainage efficiency decision, a multi-level alarm signal corresponding to cerebrospinal fluid drainage is generated.

[0050] The present invention has the following beneficial effects:

[0051] (1) Real-time acquisition and dynamic calibration of the output pressure analog signal through high-precision pressure sensors is the core of ensuring the accuracy of pressure data during cerebrospinal fluid drainage. This high-precision sensor can capture pressure changes with extremely high precision, ensuring the accuracy and reliability of sensor readings. In addition, through dynamic calibration, the system error of the sensor can be corrected in real time to avoid measurement errors caused by equipment aging, environmental changes or other external factors, thereby ensuring the stability and consistency of the data.

[0052] (2) Real-time collection of flow data by ultrasonic flow sensors and the implementation of wavelet denoising and temperature compensation processing can help improve the accuracy and stability of flow measurement. During cerebrospinal fluid drainage, changes in flow directly affect the drainage effect, so accurate flow measurement is crucial. Ultrasonic flow sensors can accurately monitor the flow of cerebrospinal fluid, while wavelet denoising technology can effectively filter out high-frequency noise in the signal, reduce interference caused by environmental factors or the sensor itself, preserve important information, and effectively remove noise components, thereby improving the quality and accuracy of the signal.

[0053] (3) The pressure analog signal and the flow analog signal are synchronously collected and transmitted to the monitoring terminal through a high-performance and high-precision analog-to-digital converter (ADC), providing the necessary foundation for further data analysis. The high precision of the ADC ensures the accuracy of the sensor signal when it is converted into a digital signal, ensuring that the original information of the data will not be lost or deformed due to the conversion process. Through synchronous collection, the pressure and flow data can maintain temporal consistency. The monitoring terminal can receive these high-quality data in real time and transmit them to the core analysis part of the system, so as to maintain a high degree of real-time performance and quickly respond to changes in the patient's status. At the same time, real-time data transmission can provide doctors with immediate feedback, facilitating timely adjustments to drainage plans and avoiding potential risks.

[0054] (4) By performing sliding average processing on the monitoring terminal to calculate the sliding average pressure and drainage efficiency, and combining the random forest machine learning model to dynamically adjust the pressure threshold, it is possible to achieve intelligent management of the cerebrospinal fluid drainage process. The sliding average pressure can effectively smooth instantaneous fluctuations and reduce errors caused by short-term data noise, so that the system can more accurately reflect the true pressure trend during the drainage process. This processing method can eliminate false alarm signals caused by instantaneous pressure fluctuations, thereby improving the accuracy and reliability of the alarm system. By configuring the random forest machine learning model, the system can automatically adjust the pressure threshold based on historical data and real-time data, optimize pressure control during the drainage process, ensure that the patient is always within a safe pressure range, and further improve the flexibility and intelligence level of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 Schematic diagram of the structure of the intelligent cerebrospinal fluid drainage monitoring system of the present invention;

[0056] Figure 2 Schematic diagram of module connections between corresponding integrated modules of the intelligent cerebrospinal fluid drainage monitoring system of the present invention;

[0057] Figure 3 This is a flowchart of the intelligent cerebrospinal fluid drainage monitoring method of the present invention. DETAILED DESCRIPTION

[0058] The present invention will be further described below with reference to the accompanying drawings and examples, but they are not intended to limit the present invention.

[0059] To achieve this, please refer to Figures 1 to 3 , Embodiment 1 of the present invention provides an intelligent cerebrospinal fluid drainage monitoring system, including a drainage catheter, a drainage bag, and a monitoring terminal. The system further integrates a high-precision pressure sensor, an ultrasonic flow sensor, a data acquisition and transmission module, and an intelligent monitoring and decision-making module. The intelligent monitoring and decision-making module is composed of the monitoring terminal and a hospital information system. The module connection diagram between the high-precision pressure sensor, the ultrasonic flow sensor, the data acquisition and transmission module, and the intelligent monitoring and decision-making module is shown in FIG. Figure 2 As shown,

[0060] The high-precision pressure sensor is a piezoresistive sensor and is installed 3-5 cm from the end of the drainage catheter corresponding to the ventricle to output a pressure analog signal through an energized bridge conversion;

[0061] In an embodiment of the present invention, the intelligent cerebrospinal fluid drainage monitoring system uses a high-precision pressure sensor that uses a piezoresistive sensor. Its core performance parameters are clearly defined, its measurement range is strictly limited to 0-200 mmH2O, and its resolution reaches 0.1 mmH2O. During actual installation, the sensor is precisely installed 3-5 cm from the corresponding ventricular end of the drainage catheter. This position has been clinically proven to accurately capture cerebrospinal fluid pressure changes. The sensor converts the pressure signal into an analog signal output via an energized bridge, ensuring effective pressure signal acquisition. The structural design of the high-precision pressure sensor balances performance and safety. Its housing is made of biocompatible titanium alloy, and its inner surface is coated with a polyparaxylene moisture barrier. This design effectively prevents external moisture intrusion, ensures stable operation of the sensor's internal circuitry, and meets biocompatibility requirements in medical environments, reducing the risk of patient infection. The self-test circuit is also an important component of the sensor. It detects impedance changes corresponding to the energized bridge every 15-30 minutes. Taking a specific detection as an example, if the impedance change deviation exceeds 5%, a calibration request is immediately triggered. The calibration operation is based on the calibration formula P 实际 =k×P 原始 +b proceed; among them, P 实际 is the calibrated pressure source, k is the calibration coefficient, P 原始 is the original pressure source, b is the offset, and the system is preset to perform a calibration operation every 7 days to ensure the accuracy of the sensor measurement. In addition, the self-test circuit also has a temperature drift compensation function. By measuring the impedance response difference of the energized bridge under 0.1mA / 1mA dual current excitation, the corresponding temperature drift compensation amount is calculated. The calculation formula is Among them, α represents the temperature drift coefficient of the material corresponding to the energized bridge, and R0 represents the internal resistance of the energized bridge. At the same time, when ΔT>2°C, a calibration request can also be triggered to ensure the measurement accuracy of the sensor in different temperature environments. It also includes a pressure conduction optimization structure, which further improves the sensor performance. It contains a 2mm diameter silicone buffer diaphragm. The diaphragm has been verified by CFD simulation and can reduce the measurement error caused by cerebrospinal fluid drainage to ≤0.5mmH2O. During the actual cerebrospinal fluid drainage process, the silicone buffer diaphragm effectively buffers the pressure fluctuations caused by liquid flow, ensuring the accuracy of pressure measurement.

[0062] The ultrasonic flow sensor uses a piezoelectric ceramic ultrasonic transducer and is installed in the straight section of the drainage catheter corresponding to 5-10 cm downstream of the high-precision pressure sensor. The flow-temperature compensation relationship formula v = 1402.5 + 4.9T - 0.05T is used. 2 Perform dynamic compensation to output flow analog signal, where v is flow rate and T is temperature;

[0063] In an embodiment of the present invention, the ultrasonic flow sensor uses a piezoelectric ceramic ultrasonic transducer with a measurement range of 0-500mL / h and an accuracy of ±1mL / h. It is installed downstream of the high-precision pressure sensor, corresponding to a 5-10cm straight section of the drainage catheter. This position can accurately measure the flow rate of cerebrospinal fluid. The sensor is based on the flow-temperature compensation relationship formula v = 1402.5 + 4.9T - 0.05T 2 Dynamic compensation is performed, where v is the flow rate and T is the temperature, to ensure the accuracy of flow measurement under different temperature conditions. The anti-interference unit is the key part of the ultrasonic flow sensor. Wavelet transform technology is used, and Daubechies4 wavelet basis function is selected to process the flow analog signal. In specific operation, the sensor inputs the collected flow analog signal into the anti-interference unit. Daubechies4 wavelet basis function separates it into effective signal component and noise signal component according to the signal characteristics. The noise reduction operation is performed by setting the dynamic threshold. The dynamic threshold calculation formula is: Where σ represents the standard deviation of the noise signal component, and N represents the signal length of the simulated flow signal. This threshold effectively eliminates noise signals and preserves the true flow signal. The dynamic calibration unit ensures the long-term accuracy of the sensor, automatically performing zero-point calibration every 12-24 hours. During calibration, a solenoid valve temporarily blocks the connection between the drainage catheter and the drainage bag, bringing the flow rate to zero. At this point, the zero-flow reference value corresponding to the ultrasonic flow sensor is recorded. Simultaneously, 32 samples are collected in this zero-flow state. Moving median filtering is used to eliminate outliers with amplitudes greater than 3ε0, ensuring the accuracy of zero-flow measurements. For example, during one calibration, a few of the 32 samples collected exhibited abnormalities. These were effectively eliminated through moving median filtering, ensuring the reliability of the zero-flow reference value.

[0064] The data acquisition and transmission module integrates an analog-to-digital converter (ADC) to synchronously acquire the corresponding pressure analog signal and flow analog signal to generate pressure data and flow data. It supports wireless dual-channel transmission using Wi-Fi as the main channel and Bluetooth as the backup channel, and uses AES-256 encryption and automatic retransmission mechanism for redundant transmission to the monitoring terminal.

[0065] In the embodiment of the present invention, the data acquisition and transmission module integrates a 16-bit high-performance and high-precision analog-to-digital converter ADC, which synchronously acquires pressure analog signals and flow analog signals at a sampling rate of 1000Hz and a signal-to-noise ratio of >90dB. In the actual acquisition process, the ADC converts the continuous analog signal into a discrete digital signal to generate corresponding pressure data and flow data. To ensure accurate data synchronization, the system aligns the pressure analog signal and the flow analog signal through a timestamp with an accuracy of ±0.1ms. For example, at a certain moment, the pressure analog signal and the flow analog signal are acquired at the same time, and the system assigns them an accurate timestamp to ensure the accuracy of subsequent data analysis. The transmission adopts a wireless dual-channel transmission method of primary channel Wi-Fi and backup channel Bluetooth, and adopts AES-256 encryption and automatic retransmission mechanism. AES-256 encryption technology encrypts the data with high strength to ensure the security of data during transmission. The automatic retransmission mechanism ensures that data is not lost. When an error occurs in data transmission, the system automatically resends the data. When the packet loss rate is greater than 5% due to electromagnetic interference, the system quickly switches to the backup channel Bluetooth for data transmission within ≤0.5 seconds. For example, in some areas of the hospital, the Wi-Fi signal is interfered with and the packet loss rate exceeds 5%. The system immediately switches to the Bluetooth channel to ensure the continuity of data transmission.

[0066] The intelligent monitoring and decision-making module uses pressure data and flow data to calculate drainage efficiency in real time; by configuring a random forest machine learning model and dynamically adjusting the pressure threshold according to the hospital information system, a multi-level alarm signal corresponding to cerebrospinal fluid drainage is generated based on the pressure threshold and drainage efficiency decision.

[0067] In the embodiment of the present invention, the intelligent monitoring decision module first processes the pressure data, dynamically sets the window length to smooth the instantaneous pressure fluctuations of the pressure data, and initially sets the window length to 10 sampling points (0.01s per sampling point). Based on this, the pressure change rate is calculated using the formula: Where ΔP represents the pressure change value, Δt represents the window length, if the pressure change rate (converted), the window length is shortened to 5 sampling points, otherwise it is maintained at 10 sampling points. For example, within a certain period of time, the pressure change rate is large, exceeding 5mmH2O / min, and the system automatically shortens the window length to more accurately capture pressure fluctuations. Then, by setting a sliding interval corresponding to 3-5 minutes, the pressure data after fluctuation smoothing is calculated by sliding average to obtain the sliding average pressure. During the calculation process, the system averages the pressure data in turn according to the time length of 3-5 minutes, effectively eliminating the impact of short-term fluctuations and reflecting the overall trend of pressure. Then, the instantaneous cerebrospinal fluid drainage volume at the end of the sliding interval corresponding to the time is extracted from the flow data. For example, at the end of the 3-5 minute sliding interval, the system accurately extracts the instantaneous drainage volume at that moment, and calculates the drainage efficiency in real time based on the ratio between the instantaneous drainage volume and the sliding average pressure. The formula is Where Q represents the instantaneous drainage volume, P avg Represents the sliding mean pressure. The drainage efficiency reflects the effect of cerebrospinal fluid drainage and provides an important basis for subsequent decision-making. In order to achieve accurate alarm, a random forest machine learning model is configured in the monitoring terminal. First, the patient's age, weight and historical data in the hospital information system are obtained, of which there are no less than 1,000 historical similar case samples. The initial pressure threshold is determined to be 70-200 mmH2O (that is, the initial threshold for adults) based on the patient's age and weight. The historical similar case samples are input into the random forest machine learning model for training. The model tree depth is set to 10. When the confidence between the historical similar cases is calculated to be >85%, the adjustment amplitude is ≤5 mm. H2O dynamically adjusts the pressure threshold and generates multi-level alarm signals corresponding to cerebrospinal fluid drainage based on the pressure threshold combined with the sliding average pressure and drainage efficiency. For example, when it is adjusted to 75-205 mmH2O, when a single sliding average pressure exceeds the pressure threshold and lasts for 5-10 seconds, the decision is triggered to generate a first-level prompt alarm signal corresponding to cerebrospinal fluid drainage, which is yellow in color, to remind medical staff to pay attention to the patient's condition. When the sliding average pressure exceeds the pressure threshold and the drainage efficiency drops sharply to zero, it is automatically judged that the drainage catheter is blocked, and the decision is immediately triggered to generate a second-level emergency alarm signal corresponding to cerebrospinal fluid drainage, which is red in color, to alert medical staff to take quick measures.

[0068] The high-precision pressure sensor further includes:

[0069] Biocompatible titanium alloy shell with a parylene moisture barrier coating on the inner surface;

[0070] In the embodiment of the present invention, during the manufacturing process of the high-precision pressure sensor of the intelligent cerebrospinal fluid drainage monitoring system, the shell is made of corresponding biocompatible titanium alloy material with a tensile strength of ≥895MPa and a density of 4.43g / cm 3The titanium alloy raw material is processed into a shell cavity with a size of 3 cm long, 2 cm wide and 1 cm high by precision milling and EDM in a CNC machining center. The shell wall thickness is uniformly set to 0.3 mm. The inner surface is treated by physical vapor deposition (PVD) technology to coat the polyparaxylene moisture-proof layer. The vacuum degree is 5×10 -4 Within the Pa coating chamber, paraxylene is heated to 650°C, sublimated into a gaseous state, and then transported to the inner surface of the housing via a carrier gas. Deposited at a substrate temperature of 30°C, it forms a 5μm-thick moisture barrier. This barrier, tested with tape, achieves 5B adhesion, effectively shielding the sensor from external moisture, ensuring stable operation of the sensor's internal circuitry within a humidity range of 20%-90%, meeting the stringent biocompatibility and moisture resistance requirements of medical implants.

[0071] The self-test circuit detects the impedance change corresponding to the energized bridge every 15-30 minutes, and triggers a calibration request when the impedance change deviation is greater than 5%, wherein the calibration request is performed once every 7 days through the corresponding regular calibration mechanism, wherein the calibration operation is calibrated according to the calibration formula, P 实际 =k×P 原始 +b; where P 实际 is the calibrated pressure source, k is the calibration coefficient, P 原始 is the original pressure source, b is the offset;

[0072] In an embodiment of the present invention, the self-test circuit of the high-precision pressure sensor is built based on the corresponding microcontroller, and a built-in 16-bit ADC module is used for impedance acquisition. Every 15-30 minutes, the microcontroller applies a 5V constant voltage excitation to the energized bridge, and samples the bridge arm resistance 200 times through the ADC at a sampling rate of 1000SPS to calculate the average impedance value. Taking the initial total impedance of a sensor energized bridge as 10kΩ as an example, if the average impedance calculated in a certain detection changes to 10.6kΩ, the impedance change deviation is ((10.6-10) / 10×100%)=6%, which exceeds the 5% threshold. The microcontroller immediately sends a calibration request instruction to the system main controller through the interface. The calibration operation is performed by the system main controller once every 7 days. According to the calibration formula P 实际 =k×P 原始 +b; where P 实际 is the calibrated pressure source, k is the calibration coefficient, P 原始 is the original pressure source, b is the offset, when performing calibration, first connect the known pressure value to P 实际 =100mmH2O standard pressure source, collect the original pressure value P output by the sensor 原始 =98mmH2O, substitute into the formula to get 100=k×98+b; then connect P 实际 =150mmH2O standard pressure source, collect P原始 =146mmH2O, the calibration coefficient k = 1.2 and the offset b = -17.6 are obtained by solving the simultaneous equations. The calculated k and b values are written into the sensor EEPROM storage unit to realize the calibration correction of the pressure data and ensure that the measurement error is always controlled within the range of ±0.1mmH2O.

[0073] The optimized pressure transmission structure includes a 2mm diameter silicone buffer diaphragm, which has been verified by CFD simulation to reduce the measurement error caused by cerebrospinal fluid drainage to ≤0.5mmH2O.

[0074] In the embodiment of the present invention, the silicone buffer diaphragm in the pressure conduction optimization structure is made of medical-grade silicone rubber (Shore hardness 20A), which is formed by precision injection molding. The diameter is strictly controlled to 2mm±0.02mm and the thickness is 0.3mm. During the CFD simulation verification, a three-dimensional simulation model of cerebrospinal fluid drainage was constructed based on ANSYS Fluent software, and the cerebrospinal fluid density was set to 1005kg / m 3 , the dynamic viscosity is 0.0015Pa·s, the inlet flow velocity is 0.05m / s, and the silicone buffer diaphragm is assembled at the contact surface of the drainage catheter and the pressure sensor for simulation calculation. By setting the LES (large eddy simulation) turbulence model, the pressure fluctuation amplitude on the diaphragm surface is calculated. After 100 simulations of different working conditions, the results show that the maximum pressure measurement error is 0.48mmH2O, which meets the design requirement of ≤0.5mmH2O. In actual application, when cerebrospinal fluid flows through the drainage catheter, the silicone buffer diaphragm buffers the fluid pulsation through its own elastic deformation and smoothly transmits the pressure signal to the sensor sensing chip. For example, when the cerebrospinal fluid flow rate suddenly changes by 0.02m / s, the measurement error of the ordinary sensor reaches 1.2mmH2O, while the measurement error of the sensor equipped with this optimized structure is only 0.35mmH2O, which significantly improves the accuracy and stability of pressure measurement.

[0075] The self-test circuit also includes measuring the impedance response difference ΔR of the energized bridge under 0.1mA / 1mA dual current excitation and calculating the corresponding temperature drift compensation amount Where α represents the temperature drift coefficient of the material corresponding to the energized bridge, R0 represents the internal resistance of the energized bridge, and a calibration request can also be triggered when ΔT>2°C.

[0076] The ultrasonic flow sensor further includes:

[0077] The anti-interference unit uses wavelet transform technology, including Daubechies4 wavelet basis function, to separate the effective signal component and the noise signal component corresponding to the flow simulation signal to complete the noise reduction operation;

[0078] In an embodiment of the present invention, the flow analog signal collected by the ultrasonic flow sensor is susceptible to interference from the external environment. The anti-interference unit uses wavelet transform technology and selects the Daubechies4 wavelet basis function to perform noise reduction processing on the flow analog signal. When the ultrasonic flow sensor starts working, it continuously collects the flow analog signal every second and transmits the signal to the anti-interference unit in real time. Taking a flow analog signal containing 1000 data points collected at a time as an example, the anti-interference unit inputs the signal into a processing program written in Python and uses the PyWavelets library to call the Daubechies4 wavelet basis function. First, the signal is decomposed into different frequency bands by a three-layer wavelet decomposition. During the decomposition process, the Daubechies4 wavelet basis function performs a piecewise convolution operation on the signal based on its specific filter coefficients, decomposing the original signal into an approximate component and a detail component. The approximate component retains the low-frequency characteristics of the signal and corresponds to the actual flow signal trend; the detail component contains high-frequency characteristics, where the noise signal is mostly present. Then, the detail component is processed by setting a threshold to calculate the standard deviation of the detail component. The dynamic threshold is determined according to the formula Wherein σ represents the standard deviation corresponding to the noise signal component, N represents the signal length 1000 corresponding to the flow simulation signal, assuming that the calculated standard deviation is 0.5, the dynamic threshold is approximately 7.07, the coefficients of the detail component whose absolute value is less than the threshold are set to zero, and the coefficients greater than the threshold are retained. After the processing is completed, the Daubechies4 wavelet basis function is used to perform 3-layer wavelet reconstruction, and the processed approximate component and detail component are recombined to obtain the effective flow signal after noise reduction. After this processing, the noise caused by factors such as electromagnetic interference and equipment vibration is effectively removed, making the flow signal smoother and more accurate, providing reliable data for subsequent flow analysis.

[0079] The dynamic calibration unit automatically performs zero-point calibration every 12-24 hours to temporarily block the connection between the drainage catheter and the drainage bag through the solenoid valve during the calibration process, and records the zero flow reference value ε0 corresponding to the ultrasonic flow sensor. At the same time, 32 sampling values are collected in the zero flow state as a moving median filter to eliminate abnormal values with an amplitude > 3ε0.

[0080] In an embodiment of the present invention, a dynamic calibration unit is used to automatically perform a zero-point calibration operation every 12-24 hours. When the calibration starts, the system sends an instruction to the solenoid valve through the control circuit, and the solenoid valve immediately acts to block the connection between the drainage catheter and the drainage bag, so that the cerebrospinal fluid cannot flow, thereby forming a zero flow state. For example, the ultrasonic flow sensor continuously collects signals and collects 32 data points at a sampling frequency of 100 Hz within the next 12-24 hours to obtain 32 sampling values. Taking the 32 sampling values [2, -1, 0, 1, 4, 2, -2, 3, 5, 1, 0, -3, 2, 1, 0, -1, 2, 3, -4, 1, 0, 2, -2, 3, 1, 0, -1, 2, 1, 0, -2, 3] collected in a certain calibration as an example, the moving median filtering method is used to process these data. The window size is set to 5. Starting from the first data point, 5 consecutive data points [2, -1, 0, 1, 4] are taken in sequence to calculate their The median is 1, and this median is used as the first filtered data point. Then the window is moved back one data point to [-1, 0, 1, 4, 2], and the median is calculated to be 1, which is used as the second filtered data point. And so on. During the filtering process, outliers with an amplitude greater than 3 (such as 5 and -4 in the above data) are removed to prevent them from affecting the zero flow reference value. After completing the filtering process of 32 sampling values, the average value of the remaining data is calculated and used as the zero flow reference value of the ultrasonic flow sensor. For example, if the average value of the remaining data after filtering is 0.5, the zero flow reference value is determined to be 0.5 mL / h. After the calibration is completed, the system controls the solenoid valve to open and restore normal cerebrospinal fluid drainage. Through this zero point calibration operation every 12-24 hours, the zero drift error of the sensor is effectively eliminated, ensuring the accuracy and reliability of the ultrasonic flow sensor measurement results, so that the measured cerebrospinal fluid flow data can truly reflect the actual drainage situation.

[0081] The noise reduction operation is performed by setting a corresponding dynamic threshold, wherein the dynamic threshold is specifically Wherein σ represents the standard deviation corresponding to the noise signal component, and N represents the signal length corresponding to the flow simulation signal.

[0082] The synchronous acquisition specifically involves aligning the corresponding pressure analog signal and flow analog signal through timestamps with an accuracy of ±0.1ms.

[0083] When the wireless dual-channel transmission mode is subjected to electromagnetic interference and causes the corresponding packet loss rate to be greater than 5%, the wireless dual-channel transmission mode switches to the backup channel within 0.5 seconds or less.

[0084] The intelligent monitoring decision module includes:

[0085] By dynamically setting the corresponding window length and smoothing the instantaneous pressure fluctuations of the pressure data based on the window length, the pressure data after fluctuation smoothing is generated;

[0086] In an embodiment of the present invention, when the intelligent cerebrospinal fluid drainage monitoring system is running, the data acquisition and transmission module obtains pressure data at a sampling rate of 100 Hz and transmits it to the intelligent monitoring decision module. In this module, Python language is combined with the numpy library for pressure data processing. The initial window length is 10 sampling points, and the corresponding time length is 0.01 seconds (because the sampling rate is 100 Hz). Taking 100 pressure data points collected continuously from a patient as an example, numbered from 1 to 100, the pressure change rate within the window of the 1st to 10th data points is first calculated. Assuming that the pressure value of the first data point is 100 mmH2O and the pressure value of the 10th data point is 105 mmH2O, according to the formula Where ΔP = 105 - 100 = 5 mmH2O, and Δt is the window time corresponding to 0.01 seconds. After conversion to minutes, the pressure change rate is 5 / 0.01 / 60 = 30,000 mmH2O / min, which is greater than 5 mmH2O / min. At this time, the window length is shortened to 5 sampling points, and the data within the shortened window is smoothed using the mean filtering method. For example, for the 1st to 5th data points, the pressure values are 100, 102, 101, 103, and 104 mmH2O, respectively. The calculated average value is (100 + 102 + 101 + 103 + 104) ÷ 5 = 102 mmH2O. This average value is used as the pressure data at this moment after smoothing. Subsequent data are processed in this way in turn, and finally the pressure data after fluctuation smoothing is generated.

[0087] By setting a sliding interval corresponding to 3-5 minutes and performing a sliding average calculation on the pressure data after fluctuation smoothing based on the sliding interval, the sliding average pressure P is obtained. avg ;

[0088] In the embodiment of the present invention, by reading the corresponding pressure data after fluctuation smoothing, the pandas library is used in the Python environment to set a sliding interval corresponding to 3-5 minutes. Since the sampling rate is 100 Hz, there are 3×60×100 and 5×60×100 data points in 3-5 minutes, that is, 18,000 to 30,000 data points. Taking this batch of data as an example, starting from the first data point, the 1st to 18,000 or 30,000 data points are taken as the first sliding interval, and the average value of the pressure data in this interval is calculated. Assuming that The total pressure data is 3,060,000 mmH2O, so the sliding average pressure is 3,060,000 ÷ 30,000 = 102 mmH2O. Then move the sliding interval backward by one data point, take the 2nd to 18,001st or 30,001th data point as the new sliding interval, repeat the above calculation process, calculate the average pressure of each sliding interval in turn, and store the results in a new column. Finally, a data set containing all sliding average pressure values is obtained. The data in this file can effectively reflect the average level of pressure within 3-5 minutes, providing a stable data foundation for subsequent analysis.

[0089] The instantaneous drainage volume Q of the cerebrospinal fluid is extracted from the corresponding flow data through the time end point corresponding to the sliding interval;

[0090] In an embodiment of the present invention, data is extracted from the flow data file (assuming it is "raw flow data.csv") stored in the data acquisition and transmission module, and the flow data and the sliding average pressure data are aligned according to the timestamp by using Python's pandas library. Taking a 5-minute sliding interval as an example, the time end of the interval is the moment corresponding to the 30,000th sampling point. In the corresponding file, the flow data value corresponding to the moment is found. Assuming that the flow data at this moment is displayed as 20 mL / h, the instantaneous drainage volume corresponding to the cerebrospinal fluid is extracted as 20 mL / h, and the instantaneous drainage volume corresponding to the time end of each sliding interval is recorded in a new list, and is associated with the corresponding sliding average pressure data for storage to form a data set containing information of both.

[0091] According to the instantaneous drainage volume Q and the sliding average pressure P avg The drainage efficiency is calculated in real time based on the ratio between

[0092] In the embodiment of the present invention, the drainage efficiency is obtained by reading the sliding average pressure and instantaneous drainage volume data through simple numerical calculation in Python. According to the formula For example, if the sliding mean pressure in a certain record is 100 mmH2O and the instantaneous drainage volume is 15 mL / h, then the drainage efficiency is η = 15 ÷ 100 = 0.15 mL / (h·mmH2O). The calculation is performed on all records in the data set in sequence, and each calculated drainage efficiency value is added to a new column. Finally, a complete data set containing the sliding mean pressure, instantaneous drainage volume, and drainage efficiency is generated. These data can reflect the efficiency of cerebrospinal fluid drainage in real time and provide an important reference for medical staff.

[0093] By configuring a random forest machine learning model in the monitoring terminal and dynamically adjusting the pressure threshold according to the patient's age, weight and historical data in the hospital information system, the pressure threshold is combined with the sliding average pressure P avg The drainage efficiency η is used to determine the multi-level alarm signals corresponding to the cerebrospinal fluid drainage.

[0094] In an embodiment of the present invention, a random forest machine learning model is configured based on the Python scikit-learn library on the server of the monitoring terminal, and the patient's age, weight and historical data are extracted from the database of the hospital information system. The historical data contains no less than 1,000 historical similar case samples, and the data storage format is a CSV file. The data is read by using the pandas library, and the initial pressure threshold range is determined to be 70-200 mmH2O according to the patient's age and weight. For example, for a patient aged 50 and weighing 60 kg, the initial pressure threshold is initially set to 120 mmH2O. The relevant features in the historical case data (such as age, weight, pressure value, drainage status, etc.) are used as input, and the corresponding pressure threshold is used as output and input into the random forest machine learning model for training. The tree depth is set to 10. During the training process, the model establishes each pressure threshold by learning from the historical data. The relationship between the characteristics and the pressure threshold. When the confidence level between similar historical cases is calculated to be >85%, the pressure threshold is dynamically adjusted according to an adjustment range of ≤5mmH2O. For example, after training and calculation, it is found that the confidence level in cases similar to the patient reaches 90%, and the pressure threshold can be adjusted to 118mmH2O according to the model calculation. Then the corresponding adjustment is made. Based on the adjusted pressure threshold, the decision is made in combination with the corresponding sliding average pressure and drainage efficiency. When a single sliding average pressure exceeds the pressure threshold and lasts for 5-10 seconds, a first-level prompt alarm signal is triggered and the color is set to yellow. When the sliding average pressure exceeds the pressure threshold and the drainage efficiency drops sharply to zero, it is automatically judged that the drainage catheter is blocked, and a second-level emergency alarm signal is immediately triggered and the color is set to red. The alarm signal is issued through the display screen and sound prompt of the monitoring terminal, and the alarm information is sent to the mobile terminal device of the medical staff to ensure timely handling of abnormal situations.

[0095] The window length corresponding to the dynamic setting is specifically the window length corresponding to the initial setting of 10 sampling points, and the corresponding pressure change rate is calculated based on the window length. Where ΔP represents the pressure change value, Δt represents the window length, and the corresponding pressure change rate is determined at the same time. When , the corresponding window length is shortened to 5 sampling points, otherwise it is kept at 10 sampling points.

[0096] The process of dynamically adjusting the pressure threshold is specifically as follows:

[0097] Obtaining the patient's age, weight, and historical data from the hospital information system, wherein the historical data includes a sample of similar historical cases (n ≥ 1000);

[0098] In an embodiment of the present invention, a database connection program written in Python is used on a monitoring terminal server of an intelligent cerebrospinal fluid drainage monitoring system to establish a connection with the hospital information system database using the pymysql library. An SQL query statement is set: SELECT age, weight, medical_history FROM patient_table WHERE condition = 'cerebrospinal_fluid_drainage'. Patient data related to cerebrospinal fluid drainage is filtered out from the patient data table patient_table in the hospital information system, where age represents the patient's age, weight represents the patient's weight, and medical_history contains historical case information. After executing the query statement, the acquired data is stored locally on the server in CSV format, and the acquired data is filtered and organized to extract historical similar case samples. Taking a certain data acquisition in actual operation as an example, a total of 2,000 patient data are acquired. Through manually set similarity judgment rules (such as consistency of key information such as symptom type and treatment plan), 1,500 historical similar case samples that meet the requirements are screened out, ensuring the validity and pertinence of the historical data and providing sufficient data support for the subsequent determination and adjustment of the pressure threshold.

[0099] The corresponding initial pressure threshold is determined to be 70-200 mmH2O according to the patient's age and weight in the hospital information system;

[0100] In this embodiment of the present invention, the patient's age and weight data are obtained from the previous step and processed using the Python pandas library. A function is created to input the patient's age and weight, and the initial pressure threshold is determined according to a preset rule. The specific rule is that when the patient is under 18 years old and weighs 20-40 kg, the initial pressure threshold is set to 70 mmH2O; if the weight is 40-60 kg, the initial pressure threshold is set to 80 mmH2O, and so on. When the patient is 18-60 years old, the initial pressure threshold increases by 1 for every 10 kg increase in weight. 0mmH2O and no more than 200mmH2O; when the patient is older than 60 years old, the initial pressure threshold is reduced by 10mmH2O based on the calculation of 18-60 years old, but not less than 70mmH2O. Taking a 40-year-old patient weighing 70kg as an example, according to the rules, the initial pressure threshold is 80+(70-60)÷10×10=90mmH2O. For each patient in the data set, the above calculation process is executed in sequence by writing a loop statement, and finally a complete data set containing the patient's age, weight and corresponding initial pressure threshold is obtained.

[0101] By configuring a random forest machine learning model in the monitoring terminal with a tree depth of 10, and inputting historical similar case samples n≥1000 into the corresponding random forest machine learning model for model training, and at the same time comparing and calculating the confidence between historical similar cases >85%, the corresponding pressure threshold is dynamically adjusted according to the corresponding adjustment range ≤5mmH2O.

[0102] In the embodiment of the present invention, a random forest machine learning model is configured based on the scikit-learn library in the Python environment of the monitoring terminal server to create a model instance using the RandomForestRegressor class, set n_estimators = 100 (the number of trees), max_depth = 10 (the tree depth), and random_state = 42 (the random seed ensures that the results are reproducible) to extract historical similar case sample data, and use the patient's age, weight and other features as the input feature matrix X, and the corresponding pressure initial threshold as the output label y, and by using the train_test_split function The data set is divided into training set and test set in the ratio of 8:2, that is, X_train, X_test, y_train, y_test = train_test_split (X, y, test_size = 0.2, random_state = 42), and the training set X_train and y_train are input into the random forest model for training. Execute model.fit (X_train, y_train). After the training is completed, the test set X_test is used for prediction to obtain the predicted pressure threshold y_pred. The mean square error (MSE) and determination coefficient (R) between the predicted value and the actual value are calculated. 2 ) to evaluate model performance. During the model training process, for each group of historically similar cases, the similarity between cases is calculated, and the cosine similarity algorithm is used to calculate the cosine value between the case feature vectors. If the cosine value is greater than 0.85 (corresponding to a confidence level >85%), the cases are considered similar. For similar cases, the pressure threshold is dynamically adjusted according to the difference between the model prediction result and the initial pressure threshold, according to the rule of an adjustment range of ≤5mmH2O. For example, the initial pressure threshold of a patient is 90mmH2O, and the model prediction value is 93mmH2O. The difference between the two is 3mmH2O, which meets the adjustment range requirement. The patient's pressure threshold is then adjusted to 93mmH2O. Through continuous iterative training and adjustment, the pressure threshold is made more suitable for the individual patient's condition, providing an accurate judgment basis for the multi-level alarm of cerebrospinal fluid drainage.

[0103] The pressure threshold combined with the sliding average pressure P avg The drainage efficiency η decision generates multi-level alarm signals corresponding to cerebrospinal fluid drainage, including:

[0104] When the single sliding mean pressure P avg When the pressure exceeds the threshold and lasts for 5-10 seconds, the decision is triggered to generate a yellow first-level prompt alarm signal corresponding to cerebrospinal fluid drainage;

[0105] In an embodiment of the present invention, the corresponding sliding average pressure data and the pressure threshold data of the corresponding patient are read in real time by a program written in Python in the monitoring terminal of the intelligent cerebrospinal fluid drainage monitoring system. The program samples data at a frequency of 100 Hz to ensure that pressure changes are captured in time. Taking a patient as an example, the pressure threshold is 90 mmH2O after dynamic adjustment. When reading the data, the program compares the sliding average pressure at each sampling moment with the pressure threshold. Suppose that at a certain moment, the sliding average pressure monitored is 91 mmH2O, 92 mmH2O, 93 mmH2O, etc. The program starts a timer to record the duration of the sliding average pressure exceeding the pressure threshold. Every 0.01 second (a sampling interval at a sampling frequency of 100 Hz) a check is performed. Whether the sliding average pressure still exceeds the pressure threshold, when the duration reaches 5-10 seconds, the program immediately executes the alarm signal generation instruction, and the alarm signal is triggered through the hardware interface of the monitoring terminal. The program sends an electrical signal to the alarm indicator light control panel connected to the monitoring terminal. The control panel is pre-configured to light up the yellow indicator light when it receives a specific signal; at the same time, the program calls the system's audio playback interface to play a pre-recorded prompt sound, such as "Please note, the cerebrospinal fluid drainage pressure is abnormal." In addition, the program will also encapsulate the alarm information in JSON format and send it to the medical staff's mobile terminal device via the Wi-Fi network. The information content includes patient ID, alarm type (level 1 prompt alarm), current sliding average pressure value and duration, etc., to ensure that medical staff can obtain abnormal conditions of the patient's cerebrospinal fluid drainage in a timely manner.

[0106] When the sliding mean pressure P avg When the pressure threshold is exceeded and the drainage efficiency η drops sharply to zero, it is automatically determined that the drainage catheter is blocked, and an immediate decision is made to trigger the generation of a secondary emergency alarm signal corresponding to cerebrospinal fluid drainage, which is red.

[0107] In an embodiment of the present invention, the corresponding sliding mean pressure and drainage efficiency data are also monitored in real time in the program of the monitoring terminal. When the sliding mean pressure is monitored to exceed the pressure threshold of the corresponding patient (for example, the pressure threshold of a patient is 90 mmH2O, and the sliding mean pressure is monitored to be 95 mmH2O), the program further pays attention to the change of the drainage efficiency. The program samples and analyzes the drainage efficiency data at a frequency of 1 Hz. Assume that at a certain moment, the drainage efficiency is 5 mL / (h·mmH2O), and then 4 mL / (h·mmH2O), 3 mL / (h·mmH2O), 2 mL / (h·mmH2O) are monitored in sequence. When the drainage efficiency is monitored to drop suddenly to 0 mL / (h·mmH2O), the program automatically determines that the drainage catheter is blocked according to the preset judgment logic. At this time, the program immediately triggers The second-level emergency alarm signal sends a high-level signal to the red alarm indicator light through the hardware control circuit of the monitoring terminal, causing it to flash rapidly; at the same time, it drives the high-decibel buzzer to continuously emit a sharp alarm sound. At the software level, the program encapsulates the alarm information with a higher priority and sends it to the medical staff's mobile terminal devices and the hospital's central monitoring system simultaneously through the Wi-Fi network and the backup Bluetooth channel. The alarm information includes details such as patient ID, alarm type (second-level emergency alarm), current sliding average pressure value, drainage efficiency value, and suspected cause of fault (drainage catheter obstruction). After receiving the alarm information, the hospital's central monitoring system will quickly locate the patient's location on the electronic map and highlight the relevant information on the monitoring screen so that medical staff can take quick measures to deal with the emergency situation of cerebrospinal fluid drainage catheter obstruction.

[0108] The second embodiment of the present invention also provides an intelligent cerebrospinal fluid drainage monitoring method, such as Figure 3 As shown, the method is implemented based on the intelligent cerebrospinal fluid drainage monitoring system as described above, and the intelligent cerebrospinal fluid drainage monitoring method includes:

[0109] S01: Real-time acquisition and dynamic calibration of the corresponding pressure analog signal are performed through a high-precision pressure sensor;

[0110] S02: Using an ultrasonic flow sensor to collect data in real time and perform wavelet denoising and temperature compensation to output a flow simulation signal corresponding to cerebrospinal fluid drainage;

[0111] S03: The output pressure analog signal and flow analog signal are synchronously collected by a high-performance and high-precision analog-to-digital converter ADC to generate corresponding pressure data and flow data, and then transmitted to the monitoring terminal;

[0112] S04: The sliding average processing is performed on the monitoring terminal to calculate the sliding average pressure and drainage efficiency, and the pressure threshold is dynamically adjusted by configuring the random forest machine learning model. Based on the pressure threshold combined with the sliding average pressure and drainage efficiency decision, a multi-level alarm signal corresponding to cerebrospinal fluid drainage is generated.

[0113] The present invention is therefore intended to be illustrative and non-restrictive in all respects, with the scope of the invention being defined by the appended claims rather than the foregoing description, and all changes that come within the meaning and range of equivalents of the application documents are intended to be embraced therein.

[0114] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. Intelligent cerebrospinal fluid drainage monitoring system, characterized by: The system includes a drainage catheter, a drainage bag and a monitoring terminal. The system also integrates a high-precision pressure sensor, an ultrasonic flow sensor, a data acquisition and transmission module and an intelligent monitoring and decision-making module. The intelligent monitoring and decision-making module is composed of the monitoring terminal and the hospital information system. The high-precision pressure sensor is a piezoresistive sensor and is installed 3-5 cm from the end of the drainage catheter corresponding to the ventricle to output a pressure analog signal through an energized bridge conversion; The ultrasonic flow sensor uses a piezoelectric ceramic ultrasonic transducer and is installed in the straight section of the drainage catheter corresponding to 5-10 cm downstream of the high-precision pressure sensor. The flow-temperature compensation relationship formula v = 1402.5 + 4.9T - 0.05T is used. 2 Perform dynamic compensation to output flow analog signal, where v is flow rate and T is temperature; The data acquisition and transmission module integrates an analog-to-digital converter to synchronously acquire the corresponding pressure analog signal and flow analog signal to generate pressure data and flow data, and supports wireless dual-channel transmission using Wi-Fi as the main channel and Bluetooth as the backup channel, while adopting encryption and automatic retransmission mechanism for redundant transmission to the monitoring terminal; The intelligent monitoring and decision-making module uses pressure data and flow data to calculate drainage efficiency in real time; by configuring a random forest machine learning model and dynamically adjusting the pressure threshold according to the hospital information system, a multi-level alarm signal corresponding to cerebrospinal fluid drainage is generated based on the pressure threshold and drainage efficiency decision.

2. The intelligent cerebrospinal fluid drainage monitoring system according to claim 1, characterized in that: The high-precision pressure sensor further includes: Biocompatible titanium alloy shell with a parylene moisture barrier coating on the inner surface; The self-check circuit detects the impedance change corresponding to the energized bridge every 15-30 minutes, and triggers a calibration request when the deviation of the impedance change is greater than 5%, wherein the calibration operation is performed through the calibration formula, P 实际 =k×P 原始 +b; Among them, P 实际 is the calibrated pressure source, k is the calibration coefficient, P 原始 is the original pressure source, b is the offset; The optimized pressure transmission structure includes a 2mm diameter silicone buffer diaphragm, which has been verified by CFD simulation to reduce the measurement error caused by cerebrospinal fluid drainage to ≤0.5mmH2O.

3. The intelligent cerebrospinal fluid drainage monitoring system according to claim 2, characterized in that: The self-test circuit also includes measuring the impedance response difference ΔR of the energized bridge under 0.1mA / 1mA dual current excitation and calculating the corresponding temperature drift compensation amount Where α represents the temperature drift coefficient of the material corresponding to the energized bridge, R0 represents the internal resistance of the energized bridge, and a calibration request can also be triggered when ΔT>2°C.

4. The intelligent cerebrospinal fluid drainage monitoring system according to claim 1, characterized in that: The ultrasonic flow sensor further includes: The anti-interference unit uses wavelet transform technology, including Daubechies4 wavelet basis function, to separate the effective signal component and the noise signal component corresponding to the flow simulation signal to complete the noise reduction operation; A dynamic calibration unit automatically performs zero-point calibration every 12-24 hours to temporarily block the connection between the drainage catheter and the drainage bag via a solenoid valve during the calibration process, and records the zero flow reference value ε0 corresponding to the ultrasonic flow sensor. At the same time, 32 sampling values are collected under the zero flow state as a moving median filter to eliminate abnormal values with an amplitude greater than 3ε0; The noise reduction operation is performed by setting a corresponding dynamic threshold, wherein the dynamic threshold is specifically Wherein σ represents the standard deviation corresponding to the noise signal component, and N represents the signal length corresponding to the flow simulation signal.

5. The intelligent cerebrospinal fluid drainage monitoring system according to claim 1 is characterized by: The synchronous acquisition specifically involves aligning the corresponding pressure analog signal and flow analog signal through timestamps with an accuracy of ±0.1ms; When the wireless dual-channel transmission mode is subjected to electromagnetic interference and causes the corresponding packet loss rate to be greater than 5%, the wireless dual-channel transmission mode switches to the backup channel within 0.5 seconds or less.

6. The intelligent cerebrospinal fluid drainage monitoring system according to claim 1, characterized in that: The intelligent monitoring decision module includes: By dynamically setting the corresponding window length and smoothing the instantaneous pressure fluctuations of the pressure data based on the window length, the pressure data after fluctuation smoothing is generated; By setting a sliding interval corresponding to 3-5 minutes and performing a sliding average calculation on the pressure data after fluctuation smoothing based on the sliding interval, the sliding average pressure P is obtained. avg ; The instantaneous drainage volume Q of the cerebrospinal fluid is extracted from the corresponding flow data through the time end point corresponding to the sliding interval; According to the instantaneous drainage volume Q and the sliding average pressure P avg The drainage efficiency is calculated in real time based on the ratio between By configuring a random forest machine learning model in the monitoring terminal and dynamically adjusting the pressure threshold according to the patient's age, weight and historical data in the hospital information system, the pressure threshold is combined with the sliding average pressure P avg The drainage efficiency η is used to determine the multi-level alarm signals corresponding to the cerebrospinal fluid drainage.

7. The intelligent cerebrospinal fluid drainage monitoring system according to claim 6, characterized in that: The window length corresponding to the dynamic setting is specifically the window length corresponding to the initial setting of 10 sampling points, and the corresponding pressure change rate is calculated based on the window length. Where ΔP represents the pressure change value, Δt represents the window length, and the corresponding pressure change rate is determined at the same time. When , the corresponding window length is shortened to 5 sampling points, otherwise it is kept at 10 sampling points.

8. The intelligent cerebrospinal fluid drainage monitoring system according to claim 6, characterized in that: The process of dynamically adjusting the pressure threshold is specifically as follows: Obtaining the patient's age, weight, and historical data from the hospital information system, wherein the historical data includes a sample of similar historical cases (n ≥ 1000); The corresponding initial pressure threshold is determined to be 70-200 mmH2O according to the patient's age and weight in the hospital information system; By configuring a random forest machine learning model in the monitoring terminal with a tree depth of 10, and inputting historical similar case samples n≥1000 into the corresponding random forest machine learning model for model training, and at the same time comparing and calculating the confidence between historical similar cases >85%, the corresponding pressure threshold is dynamically adjusted according to the corresponding adjustment range ≤5mmH2O.

9. The intelligent cerebrospinal fluid drainage monitoring system according to claim 6, characterized in that: The pressure threshold combined with the sliding average pressure P avg The drainage efficiency η decision generates multi-level alarm signals corresponding to cerebrospinal fluid drainage, including: When the single sliding mean pressure P avg When the pressure exceeds the threshold and lasts for 5-10 seconds, the decision is triggered to generate a yellow first-level prompt alarm signal corresponding to cerebrospinal fluid drainage; When the sliding mean pressure P avg When the pressure threshold is exceeded and the drainage efficiency η drops sharply to zero, it is automatically determined that the drainage catheter is blocked, and an immediate decision is made to trigger the generation of a secondary emergency alarm signal corresponding to cerebrospinal fluid drainage, which is red.

10. An intelligent cerebrospinal fluid drainage monitoring method, characterized by: The method is implemented based on the intelligent cerebrospinal fluid drainage monitoring system according to any one of claims 1 to 9 above, and the intelligent cerebrospinal fluid drainage monitoring method includes: S01: Real-time acquisition and dynamic calibration of the corresponding pressure analog signal are performed through a high-precision pressure sensor; S02: Using an ultrasonic flow sensor to collect data in real time and perform wavelet denoising and temperature compensation to output a flow simulation signal corresponding to cerebrospinal fluid drainage; S03: The output pressure analog signal and flow analog signal are synchronously collected by a high-performance and high-precision analog-to-digital converter ADC to generate corresponding pressure data and flow data, and then transmitted to the monitoring terminal; S04: The sliding average processing is performed on the monitoring terminal to calculate the sliding average pressure and drainage efficiency, and the pressure threshold is dynamically adjusted by configuring the random forest machine learning model. Based on the pressure threshold combined with the sliding average pressure and drainage efficiency decision, a multi-level alarm signal corresponding to cerebrospinal fluid drainage is generated.

Images