Intelligent gastrointestinal decompression drainage monitoring device, system and method

Abstract

The invention relates to the technical field of medical instruments, and discloses an intelligent gastrointestinal decompression drainage monitoring device, system and method. The data acquisition module acquires drainage pipeline multi-modal monitoring data in real time, the multi-modal fusion module processes the data to generate a standardized drainage feature vector, the anomaly supervision module evaluates the anomaly probability and type, and the real-time analysis module performs multi-dimensional time sequence correlation analysis to generate a report when the anomaly occurs. And the feedback execution module transmits a report, triggers an alarm and adjusts parameters of the negative pressure device. The problems that traditional gastrointestinal decompression drainage monitoring is low in efficiency, large in error, incapable of achieving intelligent regulation and control and the like are solved, drainage data can be accurately monitored, abnormity can be found and processed in time, intelligent regulation and control of the negative pressure device are achieved, the safety and effectiveness of gastrointestinal decompression treatment are improved, and medical data management and sharing are facilitated.

Description

Technical Field

[0001] The present invention relates to the technical field of medical devices, and particularly to an intelligent gastrointestinal decompression drainage monitoring device, system and method. Background Art

[0002] Gastrointestinal decompression drainage is a commonly used means in clinical treatment, and is widely used in the treatment process of various diseases such as gastrointestinal obstruction and postoperative recovery. Its principle is to place a drainage tube in the patient's gastrointestinal tract, and with the help of a negative pressure suction device, draw out the gas, liquid, etc. in the gastrointestinal tract to the outside of the body, so as to reduce the pressure in the gastrointestinal tract, promote the recovery of gastrointestinal function, prevent the reflux of gastrointestinal contents, reduce the anastomotic tension, and contribute to wound healing.

[0003] However, there are many problems in the existing gastrointestinal decompression drainage technology. In terms of drainage data monitoring, the traditional monitoring method mainly relies on medical staff to manually observe and record information such as the volume, color, and properties of the drainage fluid at regular intervals. This method is inefficient and prone to human errors. Medical staff need to frequently move between each hospital bed, consuming a lot of time and energy. In a busy work, it is very likely that the recording is not timely and accurate, affecting the judgment of the patient's condition. Moreover, manual observation can only obtain relatively rough data and cannot accurately analyze the composition of the drainage fluid, making it difficult to detect early potential abnormalities.

[0004] From the perspective of abnormal judgment and treatment, the lack of an intelligent monitoring system makes it difficult for medical staff to detect abnormal situations in the drainage process in a timely manner. When problems such as drainage tube blockage, poor drainage, and bleeding occur, they are often not discovered until the symptoms are more obvious, which may lead to delay of the disease, increase the patient's pain and treatment risk. For example, the blockage of the drainage tube may cause the pressure in the gastrointestinal tract to rise again, affecting the treatment effect and even causing serious complications such as gastrointestinal perforation; and the appearance of abnormal bloody components in the drainage fluid, if not discovered and treated in time, may mean that the patient has bleeding. If the bleeding volume is large and not controlled in time, it will pose a threat to the patient's life safety.

[0005] In addition, the operating parameters of the existing negative pressure devices are usually set according to experience and cannot be dynamically adjusted according to the specific conditions of the patient and the real-time changes during the drainage process. The conditions and physical conditions of different patients vary, and their requirements for negative pressure are also different. Fixed negative pressure parameters may lead to poor drainage effects. For example, too large negative pressure may damage the gastrointestinal mucosa and cause bleeding; too small negative pressure cannot effectively drain, prolonging the patient's recovery time.

[0006] In the aspects of data management and medical informatization, the traditional manual recording method is not conducive to the long-term preservation, statistical analysis, and sharing of data. It is difficult for medical staff to quickly obtain the historical drainage data of patients for comparative analysis, which affects the judgment of the development trend of patients' conditions and the optimization of treatment plans. At the same time, it cannot be effectively integrated with the hospital's information management system, which is not conducive to the efficient utilization of medical resources and the improvement of the quality of medical services. Summary of the Invention

[0007] The purpose of the present invention is to provide an intelligent gastrointestinal decompression drainage monitoring device, system, and method to solve the problems proposed in the above background technology.

[0008] To achieve the above purpose, the present invention provides the following technical solution: An intelligent gastrointestinal decompression drainage monitoring system, the system includes:

[0009] A data acquisition module for real-time acquisition of multi-modal monitoring data of the drainage pipeline, including instantaneous flow monitoring values, cumulative flow monitoring curves, liquid property spectral signals, negative pressure intensity monitoring values, and temperature fluctuation monitoring values;

[0010] A multi-modal fusion module for performing signal separation and compensation processing on the instantaneous flow monitoring value, cumulative flow monitoring curve, liquid property spectral signal, negative pressure intensity monitoring value, and temperature fluctuation monitoring value to generate a standardized drainage feature vector;

[0011] An anomaly supervision module for evaluating the anomaly probability of the standardized drainage feature vector through a dynamic supervision network and outputting the drainage anomaly probability and anomaly type identifier;

[0012] A real-time analysis module for performing multi-dimensional time series correlation analysis on the drainage data based on the anomaly type identifier when the drainage anomaly probability exceeds a preset threshold, and generating an anomaly event report;

[0013] A feedback execution module for transmitting the anomaly event report to the client and triggering an alarm instruction, and at the same time adjusting the operating parameters of the negative pressure device according to the anomaly type identifier.

[0014] Preferably, the execution steps of the data acquisition module include:

[0015] Collecting liquid property spectral signals through an infrared spectral sensor and extracting blood component characteristic values and non-blood component characteristic values through a signal separation algorithm;

[0016] Obtaining the instantaneous flow monitoring value and cumulative flow monitoring curve through a flow sensor and synchronously recording the temperature fluctuation monitoring value through a temperature sensor;

[0017] The negative pressure intensity monitoring value is monitored in real time through a negative pressure sensor, and the multi-modal monitoring data is aligned according to the time stamp and stored in the local cache queue.

[0018] Preferably, the execution steps of the multi-modal fusion module further include:

[0019] Perform noise suppression processing on the liquid property spectral signal to generate a denoised spectral signal;

[0020] Perform time window segmentation on the instantaneous flow rate monitoring value and the cumulative flow rate monitoring curve, and extract the flow rate fluctuation characteristic parameters;

[0021] Perform normalization processing on the negative pressure intensity monitoring value and the temperature fluctuation monitoring value to generate a joint environmental parameter matrix;

[0022] Input the denoised spectral signal, the flow rate fluctuation characteristic parameters, and the joint environmental parameter matrix into the feature fusion engine to generate the standardized drainage feature vector.

[0023] Preferably, the execution steps of the anomaly supervision module further include:

[0024] Construct an abnormal event database based on historical drainage data, where the abnormal event database includes an abnormal flow rate threshold interval, a blood component concentration threshold, and an environmental parameter association rule;

[0025] Perform feature matching on the standardized drainage feature vector through the dynamic supervision network, and calculate the deviation degree from the abnormal flow rate threshold interval, the probability of exceeding the blood component concentration limit, and the conflict index of the environmental parameter association rule;

[0026] Based on the weighted sum result of the deviation degree, the exceeding probability, and the conflict index, output the drainage anomaly probability and the anomaly type identifier.

[0027] Preferably, the execution steps of the real-time analysis module further include:

[0028] When the anomaly type identifier is a flow rate anomaly, extract the mutation points of the cumulative flow rate monitoring curve and calculate their frequency density;

[0029] When the anomaly type identifier is a blood component anomaly, perform trend fitting on the blood component characteristic value and identify its change slope;

[0030] When the anomaly type identifier is an environmental parameter conflict, retrieve the conflict items in the joint environmental parameter matrix with the preset rules and generate a conflict description label;

[0031] Integrate the mutation point frequency density, the change slope, or the conflict description label into the abnormal event report.

[0032] Preferably, the execution steps of the feedback execution module further include:

[0033] Invoking a preset negative pressure adjustment strategy according to the abnormal type identifier, including reducing the negative pressure intensity, pausing the drainage, or starting a cleaning program;

[0034] Transmitting the alarm instruction and the abnormal event report to a specified client through an encrypted channel, and synchronously updating the drainage log in the electronic nursing record sheet.

[0035] Preferably, the construction steps of the dynamic supervision network include:

[0036] Collecting normal event samples and abnormal event samples from historical drainage data, and extracting their corresponding standardized drainage feature vectors;

[0037] Training a random forest classification model based on the standardized drainage feature vectors to distinguish normal drainage events, abnormal blood component events, and abnormal flow events;

[0038] Online calibration of the random forest classification model through a sliding time window to generate the dynamic supervision network.

[0039] Preferably, the execution steps of the signal separation algorithm include:

[0040] Segmenting the liquid property spectral signal by bands and extracting the spectral intensity sequence of the target band;

[0041] Separating the independent signal components of the blood component and the non-blood component through independent component analysis;

[0042] Performing baseline drift compensation on the separated signal components to generate compensated blood component characteristic values and non-blood component characteristic values.

[0043] Preferably, the present invention further includes an intelligent gastrointestinal decompression drainage monitoring device, and the device includes:

[0044] An infrared spectrum sensor, a flow sensor, a negative pressure sensor, and a temperature sensor for real-time collection of multi-modal monitoring data of the drainage pipeline;

[0045] An embedded processor for executing the above intelligent gastrointestinal decompression drainage monitoring system.

[0046] Preferably, the present invention further includes an intelligent gastrointestinal decompression drainage monitoring method, and the method includes:

[0047] Real-time collection of instantaneous flow rate, cumulative flow rate, liquid property spectrum, negative pressure intensity, and temperature data of the drainage pipeline;

[0048] Perform signal separation, compensation, and fusion processing on the multimodal data to generate a standardized drainage feature vector;

[0049] Evaluate the anomaly probability and type of the standardized drainage feature vector through a dynamic supervision network;

[0050] When an anomaly is detected, generate an anomaly event report and trigger an alarm, and at the same time adjust the operating parameters of the negative pressure device.

[0051] Compared with the prior art, the beneficial effects of the present invention are:

[0052] In terms of data monitoring, by integrating multiple sensors, such as infrared spectroscopy sensors, flow sensors, negative pressure sensors, and temperature sensors, multimodal monitoring data of the drainage pipeline can be obtained in real time, including instantaneous flow monitoring values, cumulative flow monitoring curves, liquid property spectral signals, negative pressure intensity monitoring values, and temperature fluctuation monitoring values. This comprehensive and real-time data acquisition method greatly improves the accuracy and timeliness of data compared with traditional manual observation and recording. Medical staff can obtain this data at any time without frequent manual recording, saving a large amount of time and effort and avoiding human errors. Moreover, multimodal data can more comprehensively reflect the actual situation during the drainage process. For example, the liquid property spectral signals collected by the infrared spectroscopy sensor, after being processed by the signal separation algorithm, can accurately extract the characteristic values of blood components and non-blood components, helping medical staff to detect abnormal changes in the composition of the drainage fluid earlier and providing a more accurate basis for disease diagnosis.

[0053] In terms of anomaly supervision and analysis functions, the multimodal fusion module in the system performs signal separation and compensation processing on the multimodal monitoring data to generate a standardized drainage feature vector, and then the anomaly supervision module evaluates the anomaly probability through a dynamic supervision network and outputs the drainage anomaly probability and anomaly type identifier. When an anomaly is detected, the real-time analysis module performs multi-dimensional time series correlation analysis on the drainage data based on the anomaly type identifier to generate an anomaly event report. This series of operations can accurately judge and deeply analyze at the first time when an anomaly occurs in the drainage process. For example, when the anomaly type identifier is a flow anomaly, the system can accurately extract the mutation points of the cumulative flow monitoring curve and calculate its frequency density to judge the severity and change trend of the flow anomaly; when the anomaly type identifier is a blood component anomaly, perform trend fitting on the blood component characteristic values to identify its change slope and timely detect potential bleeding risks. This intelligent anomaly monitoring and analysis mechanism can help medical staff discover problems earlier, take corresponding treatment measures in time, avoid delay of the disease condition, and effectively reduce the treatment risk of patients.

[0054] For the regulation of the negative pressure device, the feedback execution module adjusts the operating parameters of the negative pressure device according to the abnormal type identifier. When situations such as abnormal flow or abnormal blood components occur, the system can automatically call the preset negative pressure adjustment strategies, such as reducing the negative pressure intensity, pausing the drainage, or starting the cleaning program. This function realizes the intelligent regulation of the negative pressure device, which can dynamically adjust the negative pressure parameters according to the specific conditions of the patient and the real-time changes during the drainage process, avoiding adverse effects on the patient caused by excessive or insufficient negative pressure. For example, when it is detected that the drainage tube is blocked and the flow rate is too small, the system automatically pauses the drainage and starts the cleaning program to dredge the drainage tube and ensure the smooth progress of the drainage process; when abnormal blood components are found and there may be a bleeding risk, the negative pressure intensity is timely reduced to reduce the damage to the gastrointestinal mucosa and ensure the safety of the patient.

[0055] In terms of medical informatization and data management, the feedback execution module transmits the abnormal event report to the client and triggers an alarm instruction, while synchronously updating the drainage log in the electronic nursing record sheet. This not only facilitates the medical staff to timely understand the patient's drainage situation but also realizes the digital management of data. All drainage data can be stored for a long time, facilitating the medical staff to consult the patient's historical drainage data at any time for comparative analysis, so as to more accurately judge the development trend of the patient's condition and provide strong support for formulating personalized treatment plans. In addition, the system can be integrated with the hospital's information management system to realize the sharing and efficient utilization of medical resources and improve the overall quality of medical services. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 is the working principle diagram of the intelligent gastrointestinal decompression drainage monitoring system described in the present invention;

[0057] Figure 2 is the flowchart of the specific processing of the multi-modal fusion module;

[0058] Figure 3 is the flowchart of the abnormal supervision module for evaluating abnormalities;

[0059] Figure 4 is the flowchart of the abnormal processing of the real-time analysis module. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0060] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0061] Please refer to Figures 1-4, the present invention provides a technical solution: an intelligent gastrointestinal decompression drainage monitoring system, aiming to achieve comprehensive and accurate monitoring of the gastrointestinal decompression drainage process, timely detect abnormal situations and take effective measures. The specific implementation solution is as follows:

[0062] Data acquisition module: Through the collaborative work of multiple sensors, it obtains multi-modal monitoring data of the drainage pipeline in real time. Among them, the infrared spectroscopy sensor is used to collect the spectral signal of the liquid property, the flow sensor obtains the instantaneous flow monitoring value and the cumulative flow monitoring curve, the temperature sensor synchronously records the temperature fluctuation monitoring value, and the negative pressure sensor monitors the negative pressure intensity monitoring value in real time. These data are stored in the local cache queue after being aligned by time stamps, providing raw data support for subsequent analysis.

[0063] Multi-modal fusion module: It performs signal separation and compensation processing on the instantaneous flow monitoring value, cumulative flow monitoring curve, liquid property spectral signal, negative pressure intensity monitoring value and temperature fluctuation monitoring value obtained by the data acquisition module. Through a series of processes, a standardized drainage feature vector is generated, and the multi-modal data is fused into a unified feature expression for subsequent analysis.

[0064] Abnormal supervision module: It uses a dynamic supervision network to evaluate the abnormal probability of the standardized drainage feature vector. This network constructs an abnormal event database through learning historical drainage data, including information such as abnormal flow threshold intervals, blood component concentration thresholds and environmental parameter association rules. During the evaluation process, it calculates the deviation degree, over-limit probability and conflict index with these rules, and finally outputs the drainage abnormal probability and abnormal type identifier.

[0065] Real-time analysis module: When the drainage abnormal probability exceeds the preset threshold, it performs multi-dimensional time series correlation analysis on the drainage data according to the abnormal type identifier. For different abnormal types, such as flow abnormality, blood component abnormality or environmental parameter conflict, different analysis methods are adopted, key features are extracted, and an abnormal event report is generated.

[0066] Feedback execution module: It transmits the abnormal event report to the client and triggers an alarm instruction to remind medical staff to handle it in time. At the same time, it adjusts the operating parameters of the negative pressure device according to the abnormal type identifier, such as calling preset negative pressure adjustment strategies, including reducing the negative pressure intensity, pausing drainage or starting a cleaning program, etc., to ensure the safety and effectiveness of the drainage process.

[0067] The present invention will be further described below in conjunction with Embodiments 1 to 6:

[0068] Embodiment 1:

[0069] In actual application scenarios, the workflow of the data acquisition module is crucial. Taking the gastrointestinal decompression drainage operation in a hospital as an example, to accurately obtain multi-modal monitoring data, various sensors are installed on the drainage pipeline. Among them, an infrared spectroscopy sensor is used to collect spectral signals of the liquid properties. Since the composition of the drainage fluid is complex and the spectral signals are easily interfered, a signal separation algorithm is adopted to process the collected spectral signals. During the processing, the spectral signals are first segmented by bands, dividing the entire spectral range into multiple small bands, and the spectral intensity sequence of the target band is extracted. This target band is determined based on clinical experience and research on the composition of the drainage fluid, and it can effectively reflect the characteristics of the bloody and non-bloody components in the drainage fluid. Then, the independent component analysis method is used to separate the independent signal components of the bloody and non-bloody components. Finally, to eliminate the baseline drift generated during the signal acquisition and processing, baseline drift compensation is performed on the separated signal components to generate the compensated characteristic values of the bloody and non-bloody components.

[0070] The flow sensor is responsible for obtaining the instantaneous flow monitoring value and the cumulative flow monitoring curve. At the same time, the temperature sensor synchronously records the temperature fluctuation monitoring value. These two sensors work together to ensure the synchrony and accuracy of the flow and temperature data. The electromagnetic flow sensor can be used as the flow sensor, and its working principle is based on the law of electromagnetic induction. When the conductive drainage fluid flows in the magnetic field, an induced electromotive force is generated inside the sensor, and the flow rate of the drainage fluid can be calculated by measuring the magnitude of this electromotive force. The thermistor temperature sensor can be selected as the temperature sensor, and the temperature of the drainage fluid is measured according to the characteristic that the resistance value of the thermistor changes with temperature.

[0071] The negative pressure sensor monitors the negative pressure intensity monitoring value in real time. The multi-modal monitoring data collected by these sensors are stored in the local cache queue after being aligned by time stamps. The role of the time stamp is to mark each data with a time tag, which is convenient for accurately processing and analyzing the data in chronological order later, ensuring the consistency and coherence of the data. After being stored in the local cache queue, the data can be processed and further analyzed in a timely manner, providing accurate data support for subsequent modules.

[0072] Example 2:

[0073] The multi-modal fusion module plays a crucial role in connecting the upper and lower parts of the entire monitoring system. For the spectral signals of the liquid properties, since various noises are inevitably mixed in during the acquisition process, these noises may interfere with the subsequent extraction and analysis of signal characteristics. Therefore, noise suppression processing is required. The wavelet transform denoising algorithm can be used. The wavelet transform can decompose the signal into different frequency sub-bands. By processing the coefficients of each sub-band, the coefficients of the sub-band where the noise is located are removed or attenuated, so as to achieve the purpose of noise suppression and generate the denoised spectral signals.

[0074] The instantaneous flow rate monitoring values and the cumulative flow rate monitoring curve are segmented by time windows, and the size of the time window is determined according to the actual situation and the data analysis requirements. For example, set the time window to 5 minutes. Within each time window, extract the flow rate fluctuation characteristic parameters, such as the maximum value, minimum value, average value, standard deviation, etc. of the flow rate. These parameters can more precisely reflect the fluctuation of the flow rate over a period of time and provide richer information for subsequent analysis.

[0075] For the negative pressure intensity monitoring values and the temperature fluctuation monitoring values, since their numerical ranges and dimensions are different, in order to facilitate unified analysis and fusion, normalization processing is required. The purpose of normalization processing is to map data in different ranges to the same interval. Usually, the method adopted is to map the data to the [0,1] interval. Assume that the negative pressure intensity monitoring value is P, its minimum value is P min , and the maximum value is P max , and the normalized negative pressure intensity value P norm The calculation formula is: Similarly, for the temperature fluctuation monitoring value T, its minimum value is T min , and the maximum value is T max , and the normalized temperature value T norm The calculation formula is: The normalized negative pressure intensity values and temperature values are combined into a joint environmental parameter matrix, which comprehensively reflects the environmental parameter situation during the drainage process.

[0076] Finally, the denoised spectral signal, the flow rate fluctuation characteristic parameters, and the joint environmental parameter matrix are input into the feature fusion engine. The feature fusion engine can adopt a neural network model, such as a multi-layer perceptron (MLP). The neural network can automatically extract the features in the data by learning a large amount of sample data and fuse these features together to generate a standardized drainage feature vector. This standardized drainage feature vector synthesizes the key information of multi-modal monitoring data and provides strong data support for subsequent anomaly supervision and analysis.

[0077] Example 3:

[0078] The anomaly supervision module is the core part of the monitoring system for judging whether the drainage state is normal.

[0079] Construct an abnormal event database based on historical drainage data, which is sourced from a large number of clinical cases. From these data, the abnormal flow threshold range, the concentration threshold of blood components, and the environmental parameter association rules are analyzed and summarized. For example, through statistical analysis of the flow data of a large number of normal and abnormal drainage cases, the abnormal flow threshold range is determined. When the instantaneous flow monitoring value or the cumulative flow monitoring curve exceeds this range, there may be a flow abnormality. For the concentration threshold of blood components, the concentration ranges of blood components in normal and abnormal situations are determined through laboratory tests and spectral data analysis of the drainage fluid samples. The environmental parameter association rules are obtained by studying the relationship between the monitored values of negative pressure intensity and temperature fluctuations, as well as their correlation with the normal drainage state.

[0080] Perform feature matching on the standardized drainage feature vectors through a dynamic supervision network. When constructing the dynamic supervision network, normal event samples and abnormal event samples in the historical drainage data are collected, and their corresponding standardized drainage feature vectors are extracted. Based on these standardized drainage feature vectors, a random forest classification model is trained. The random forest classification model consists of multiple decision trees, and each decision tree is trained based on different sample subsets and feature subsets. Finally, the results of multiple decision trees are integrated for classification. Through this model, normal drainage events, blood component abnormal events, and flow abnormal events can be distinguished. During the actual operation process, the random forest classification model is calibrated online through a sliding time window to adapt to the dynamic changes of the data. The size of the sliding time window can be adjusted according to the actual situation. For example, it can be set to 10 minutes, and the model is calibrated with new data every 10 minutes to ensure the accuracy and adaptability of the model.

[0081] During the feature matching process, calculate the deviation degree from the abnormal flow threshold range. The calculation method of the deviation degree can be to calculate the ratio of the distance between the current flow monitoring value and the boundary of the abnormal flow threshold range to the width of the threshold range. Suppose the current instantaneous flow monitoring value is Q, and the abnormal flow threshold range is [Q min ,Q max . The calculation formula for the deviation degree D is: (when Q < Q min ) or (when Q > Q max ). At the same time, calculate the probability of exceeding the concentration of blood components. By comparing the current blood component characteristic value with the concentration threshold of blood components, the probability of exceeding the limit is calculated using a probability statistical model. Finally, calculate the conflict index of the environmental parameter association rules. The conflict index can be obtained by calculating the degree of difference between the current combined environmental parameter matrix and the preset environmental parameter association rules. Based on the weighted sum result of the deviation degree, the probability of exceeding the limit, and the conflict index, the drainage abnormality probability and the abnormal type identifier are output. The formula for the weighted sum is: Abnormality probability P abnormal = w1D + w2Poverlimit +w3C, where w1, w2, and w3 are weight coefficients determined according to the importance of different parameters, and P overlimit is the probability of the concentration of blood components exceeding the limit, and C is the conflict index of the environmental parameter association rule.

[0082] Example 4:

[0083] When the abnormal supervision module detects abnormal drainage and outputs the abnormal type identifier, the real-time analysis module will conduct targeted analysis according to different abnormal types. When the abnormal type identifier is traffic anomaly, it is necessary to extract the mutation points of the cumulative traffic monitoring curve and calculate their frequency density. The cumulative traffic monitoring curve reflects the change of the total amount of drainage fluid over time, and the mutation points may indicate abnormal situations during the drainage process, such as blocked drainage tubes or sudden changes in the drainage speed. By calculating the first-order difference of the curve, when the difference result exceeds a certain threshold, the point is determined as a mutation point. Assume that the cumulative traffic monitoring curve is Q(t), the time interval is Δt, and the first-order difference formula is: The mutation point threshold is θ. When |ΔQ(t)|>θ, the t point is a mutation point. Calculate the frequency density of the mutation points. The calculation method of the frequency density is the ratio of the number of mutation points to the time range within a certain time range. For example, within 1 hour, the number of mutation points is counted as n, and the time range is T = 60 minutes, then the frequency density By analyzing the frequency density of the mutation points, the severity and frequency of the traffic anomaly can be judged.

[0084] When the abnormal type identifier is blood component anomaly, perform trend fitting on the blood component characteristic values to identify its change slope. The change of the blood component characteristic values reflects the change of the content of blood components in the drainage fluid, and the trend can be predicted through trend fitting. The method of linear regression can be used for trend fitting. Assume that the blood component characteristic value is X(t) and the time is t, and the linear regression model is X(t)=a + bt, where a is the intercept and b is the change slope. Solve the values of a and b by the least squares method to obtain the change slope b. The positive or negative and magnitude of the change slope reflect the change trend of the blood component characteristic values. If the slope is positive and large, it indicates that the content of blood components is increasing rapidly, and there may be abnormal situations such as bleeding.

[0085] When the abnormal type identifier is environmental parameter conflict, retrieve the conflict items in the combined environmental parameter matrix that conflict with the preset rules, and generate conflict description labels. The preset rules are determined based on clinical experience and research on normal drainage environmental parameters. For example, it is stipulated that the negative pressure intensity is within a certain range and the temperature is also within a corresponding reasonable range. When the negative pressure intensity value or temperature value in the combined environmental parameter matrix exceeds the preset range, or the relationship between the two does not conform to the preset rules, it is determined that there are conflict items. Generate conflict description labels such as "too high negative pressure, abnormal temperature", etc. These labels can intuitively describe the situation of environmental parameter conflicts, facilitating medical staff to understand the abnormal situation. Finally, integrate the mutation point frequency density, change slope, or conflict description labels into the abnormal event report, so that the abnormal event report can comprehensively and accurately reflect the characteristics and details of the abnormal situation.

[0086] Example 5:

[0087] The feedback execution module in the entire monitoring system is responsible for promptly communicating abnormal information to relevant personnel and adjusting the negative pressure device to ensure the safety and effectiveness of the drainage process. When receiving the abnormal type identifier, call the preset negative pressure adjustment strategy according to the abnormal type identifier. If the abnormal type identifier is flow rate abnormality and the flow rate is too large, it may cause discomfort or injury to the patient. At this time, the strategy of reducing the negative pressure intensity can be called to reduce the drainage speed. Assume that the current negative pressure intensity is P1 and the reduced negative pressure intensity is P2. According to clinical experience and equipment performance, determine the reduction amplitude. For example, P2 = P1 - ΔP, where ΔP is the negative pressure reduction value set according to the actual situation. If the flow rate abnormality is due to blockage of the drainage tube resulting in too small a flow rate, the drainage may be paused and a cleaning procedure may be initiated to unclog the drainage tube. The cleaning procedure can inject cleaning liquid into the drainage tube and use the pressure and flushing effect of the cleaning liquid to remove the blockage.

[0088] For abnormal blood components, if the content of blood components is too high, it may be necessary to adjust the negative pressure device to avoid further bleeding or adverse effects on the wound. Similarly, the strategy of reducing the negative pressure intensity can be adopted, and the specific adjustment amplitude is determined according to the severity of the abnormal blood components and clinical experience.

[0089] When transmitting the alarm instruction and the abnormal event report to the client, to ensure the security and confidentiality of the data, it is transmitted through an encrypted channel. The encrypted channel can adopt a secure encryption algorithm, such as the AES (Advanced Encryption Standard) algorithm. Encrypt the data in the abnormal event report, and then transmit it to the specified client through the network. After receiving the data, the client uses the corresponding key to decrypt it to obtain the content of the abnormal event report. At the same time, synchronously update the drainage log in the electronic nursing record sheet, recording information such as the occurrence time, abnormal type, and treatment measures of the abnormal event, which is convenient for medical staff to view and trace at any time. In this way, medical staff can timely understand the abnormal situations during the drainage process and take corresponding measures to handle them, ensuring the safety and treatment effect of patients.

[0090] Example 6:

[0091] The construction of the dynamic supervision network is a key link in the abnormal supervision module. First, collect normal event samples and abnormal event samples from historical drainage data, and these samples come from a large number of clinical cases. During the collection process, ensure the diversity and representativeness of the samples, covering different conditions and different drainage stages. Extract the corresponding standardized drainage feature vectors from these samples. The standardized drainage feature vectors are obtained after being processed by the multi-modal fusion module, which integrates the key information of various monitoring data.

[0092] Train a random forest classification model based on the standardized drainage feature vectors. The random forest classification model consists of multiple decision trees. When each decision tree is trained, a part of the samples are randomly selected from the sample set as the training data, and at the same time, a part of the features are randomly selected from the feature set for splitting. This can increase the diversity and generalization ability of the model and avoid overfitting. Through training, the random forest classification model can distinguish normal drainage events, abnormal events of blood components, and abnormal events of flow rate. During the training process, continuously adjust the parameters of the model, such as the number of decision trees, the maximum depth, etc., to improve the accuracy of the model.

[0093] After the training is completed, perform online calibration on the random forest classification model through a sliding time window. The size of the sliding time window is determined according to the actual situation, for example, set to 15 minutes. Every 15 minutes, use the newly collected drainage data within a period of time as new samples to calibrate the model. During the calibration process, recalculate the parameters of the model to make it adapt to the dynamic changes of the data and improve the adaptability and accuracy of the model to new data.

[0094] The signal separation algorithm plays an important role in the data acquisition module. When processing the spectral signals of the liquid characteristics, the spectral signals are first segmented by wavelength bands. According to the research on the composition of the drainage fluid and clinical experience, the target wavelength bands that can effectively reflect the characteristics of the bloody and non-bloody components are determined, and the spectral intensity sequences of the target wavelength bands are extracted. Then, the independent component analysis method is used to separate the independent signal components of the bloody and non-bloody components. The independent component analysis method is a statistical-based signal processing method. It assumes that the mixed signal is a linear combination of multiple independent source signals, and by finding a linear transformation matrix, the mixed signal is separated into independent source signals. Finally, baseline drift compensation is performed on the separated signal components. Since baseline drift may occur during signal acquisition and transmission, resulting in a change in the zero point of the signal and affecting the accurate analysis of signal characteristics. Through the baseline drift compensation algorithm, the signal is corrected to generate the compensated characteristic values of the bloody and non-bloody components.

[0095] The intelligent gastrointestinal decompression drainage monitoring device consists of an infrared spectral sensor, a flow sensor, a negative pressure sensor, a temperature sensor, and an embedded processor. These sensors are responsible for real-time acquisition of multi-modal monitoring data of the drainage pipeline, and the embedded processor executes the functions of each module of the intelligent gastrointestinal decompression drainage monitoring system. The embedded processor can use a high-performance microprocessor, which has powerful data processing capabilities and computing speeds, and can quickly process the data collected by the sensors and run the algorithms and programs of the monitoring system.

[0096] The intelligent gastrointestinal decompression drainage monitoring method includes the following steps: real-time acquisition of the instantaneous flow rate, cumulative flow rate, liquid characteristic spectrum, negative pressure intensity, and temperature data of the drainage pipeline, and real-time acquisition of data is achieved through sensors installed on the drainage pipeline. Signal separation, compensation, and fusion processing are performed on the multi-modal data to generate a standardized drainage characteristic vector, and this step is consistent with the work process of the multi-modal fusion module. The abnormal probability and type of the standardized drainage characteristic vector are evaluated through a dynamic supervision network, and the constructed dynamic supervision network is used for abnormal evaluation. When an abnormality is detected, an abnormal event report is generated and an alarm is triggered, and at the same time, the operating parameters of the negative pressure device are adjusted, which corresponds to the functions of the real-time analysis module and the feedback execution module. Through this series of steps, comprehensive monitoring and management of the gastrointestinal decompression drainage process are realized, ensuring the treatment effect and safety of patients.

[0097] It should be noted that in this text, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprising", "including" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device.

[0098] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent gastrointestinal decompression drainage monitoring system, characterized in that: include: The data acquisition module is used to obtain multi-modal monitoring data of the drainage pipeline in real time, including instantaneous flow monitoring value, cumulative flow monitoring curve, liquid property spectral signal, negative pressure intensity monitoring value and temperature fluctuation monitoring value; A multimodal fusion module, used for performing signal separation and compensation processing on the instantaneous flow monitoring value, the cumulative flow monitoring curve, the liquid property spectral signal, the negative pressure intensity monitoring value and the temperature fluctuation monitoring value, and generating a standardized drainage feature vector; An abnormal supervision module is used to evaluate the abnormal probability of the standardized drainage feature vector through a dynamic supervision network, and output the drainage abnormal probability and abnormal type identification; A real-time analysis module, configured to perform multi-dimensional time series correlation analysis on the drainage data based on the abnormal type identifier and generate an abnormal event report when the drainage abnormality probability exceeds a preset threshold; The feedback execution module is used to transmit the abnormal event report to the client and trigger an alarm instruction, and adjust the operating parameters of the negative pressure device according to the abnormal type identifier.

2. The intelligent gastrointestinal decompression drainage monitoring system according to claim 1, characterized in that: The execution steps of the data acquisition module include: The liquid property spectral signal is collected by infrared spectral sensor, and the blood component characteristic value and non-blood component characteristic value are extracted by signal separation algorithm; The instantaneous flow monitoring value and the cumulative flow monitoring curve are obtained through the flow sensor, and the temperature fluctuation monitoring value is synchronously recorded through the temperature sensor; The negative pressure intensity monitoring value is monitored in real time by a negative pressure sensor, and the multimodal monitoring data is stored in a local cache queue after being aligned by timestamp.

3. The intelligent gastrointestinal decompression drainage monitoring system according to claim 2, characterized in that: The execution steps of the multimodal fusion module also include: Performing noise suppression processing on the liquid property spectral signal to generate a denoised spectral signal; The instantaneous flow monitoring value and the cumulative flow monitoring curve are divided into time windows to extract flow fluctuation characteristic parameters; Normalizing the negative pressure intensity monitoring value and the temperature fluctuation monitoring value to generate a joint environmental parameter matrix; The denoised spectral signal, flow fluctuation characteristic parameters and joint environmental parameter matrix are input into a feature fusion engine to generate the standardized drainage feature vector.

4. The intelligent gastrointestinal decompression drainage monitoring system according to claim 2, characterized in that: The execution steps of the abnormal supervision module also include: Building an abnormal event database based on historical drainage data, wherein the abnormal event database includes abnormal flow threshold intervals, blood component concentration thresholds, and environmental parameter association rules; Performing feature matching on the standardized drainage feature vector through the dynamic supervision network, calculating the degree of deviation from the abnormal flow threshold interval, the probability of exceeding the limit of blood component concentration, and the conflict index of the environmental parameter association rule; Based on the weighted sum of the deviation degree, the limit-exceeding probability and the conflict index, the drainage anomaly probability and the anomaly type identifier are output.

5. The intelligent gastrointestinal decompression drainage monitoring system according to claim 2, characterized in that: The execution steps of the real-time analysis module also include: When the abnormal type is identified as flow abnormality, extracting the mutation point of the cumulative flow monitoring curve and calculating its frequency density; When the abnormal type is identified as abnormal blood components, trend fitting is performed on the characteristic values ​​of the blood components to identify the change slope thereof; When the abnormal type is identified as an environmental parameter conflict, searching the combined environmental parameter matrix for conflicting items with preset rules, and generating a conflict description tag; The mutation point frequency density, change slope or conflict description label is integrated into the abnormal event report.

6. The intelligent gastrointestinal decompression drainage monitoring system according to claim 2, characterized in that: The execution steps of the feedback execution module also include: Invoke a preset negative pressure adjustment strategy according to the abnormal type identifier, including reducing the negative pressure intensity, suspending drainage, or starting a cleaning program; The alarm instruction and the abnormal event report are transmitted to the designated client through an encrypted channel, and the drainage log in the electronic nursing record is updated synchronously.

7. The intelligent gastrointestinal decompression drainage monitoring system according to claim 2, characterized in that: The steps of constructing the dynamic supervision network include: Collect normal event samples and abnormal event samples from historical drainage data, and extract their corresponding standardized drainage feature vectors; Training a random forest classification model based on the standardized drainage feature vector to distinguish normal drainage events, blood component abnormal events, and flow abnormal events; The random forest classification model is calibrated online through a sliding time window to generate the dynamic supervision network.

8. The intelligent gastrointestinal decompression drainage monitoring system according to claim 2, characterized in that: The execution steps of the signal separation algorithm include: Performing band segmentation on the liquid property spectral signal to extract the spectral intensity sequence of the target band; Independent component analysis was used to separate the independent signal components of blood components and non-blood components; Baseline drift compensation is performed on the separated signal components to generate compensated blood component characteristic values ​​and non-blood component characteristic values.

9. An intelligent gastrointestinal decompression drainage monitoring device, characterized in that: include: Infrared spectrum sensor, flow sensor, negative pressure sensor and temperature sensor are used to collect multi-modal monitoring data of drainage pipelines in real time; An embedded processor, used to execute the intelligent gastrointestinal decompression drainage monitoring system as described in any one of claims 1-8.

10. An intelligent gastrointestinal decompression drainage monitoring method, characterized in that: include: Real-time collection of instantaneous flow, cumulative flow, liquid property spectrum, negative pressure intensity and temperature data of the drainage pipeline; Performing signal separation, compensation and fusion processing on the multimodal data to generate a standardized drainage feature vector; Evaluating the abnormal probability and type of the standardized drainage feature vector through a dynamic supervision network; When an abnormality is detected, an abnormal event report is generated and an alarm is triggered, and the operating parameters of the negative pressure device are adjusted.

Images