Thoracic surgery postoperative drainage monitoring data processing method and system

Abstract

The invention discloses a thoracic surgery postoperative drainage monitoring data processing method and a thoracic surgery postoperative drainage monitoring data processing system. The method comprises the following steps: synchronously acquiring thoracic respiration dynamic characteristics of a patient and fluid dynamic characteristics of a thoracic closed drainage system through a data acquisition module; the central processing server establishes a time sequence mapping relationship between the two, performs feature fusion analysis, and calculates to obtain a thoracic cavity-pipeline coupling efficiency index; performing logic judgment on the postoperative recovery state of the patient on the basis of the index, and identifying a chest-pipeline fluid transmission state including breathing work limitation, drainage channel blockage and pleural cavity air leakage; and dynamically generating a grading equipment response strategy including adjusting training intensity and triggering pipeline maintenance early warning or analgesia intervention according to an identification result. According to the method, accurate recognition and closed-loop management of postoperative complications are achieved by quantitatively evaluating the linkage relation between respiration driving and drainage response.

Description

Technical Field

[0001] This invention relates to the field of computer data processing technology, and in particular to a method and system for processing postoperative drainage monitoring data in thoracic surgery. Background Technology

[0002] Following thoracic surgery, the effective operation of the closed thoracic drainage system and the recovery of the patient's respiratory function are two key factors determining prognosis. In current clinical nursing models, monitoring these two aspects is often independent and relies on manual experience. Healthcare professionals typically assess respiratory status by observing chest rise and fall, and judge drainage progress by visually observing the fluctuations in the water-seal bottle's fluid level and the escape of air bubbles. This traditional monitoring method severs the causal link between respiratory motion as the physical driving force and fluid level fluctuations as the fluid response.

[0003] Due to a lack of technical means to analyze the linkage between respiration and drainage, it is difficult to accurately determine the root cause of abnormal fluid level fluctuations in clinical practice. For example, when a decrease in fluid level fluctuations is observed in a water-seal bottle, medical staff may find it difficult to immediately distinguish whether the drainage tube is physically blocked by blood clots, causing an interruption in pressure transmission, or whether the patient's reluctance to breathe deeply due to postoperative pain is leading to insufficient driving force. This ambiguity in judgment can easily lead to the misuse of nursing interventions, such as incorrectly performing drainage tube compression on patients whose breathing is limited solely by pain, increasing their suffering; or failing to detect early incomplete tube blockage in a timely manner, delaying the lung re-expansion process. Furthermore, current postoperative management lacks a closed-loop feedback mechanism based on objective data, making it difficult to dynamically adjust rehabilitation training plans according to the patient's real-time chest cavity-tube coupling status. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for processing postoperative drainage monitoring data in thoracic surgery, in order to solve the problems pointed out in the background art.

[0005] In a first aspect, the present invention provides a method for processing postoperative drainage monitoring data in thoracic surgery, which is applied to a postoperative drainage monitoring data processing system in thoracic surgery. The system includes a data acquisition module, a central processing server, and a bedside interactive terminal. The method includes the following steps: The data acquisition module acquires real-time monitoring data of patients after thoracic surgery, including the patient's thoracic respiratory dynamics data and the fluid dynamics data of the closed thoracic drainage system. The method further includes the following steps: The central processing server establishes a time-synchronous mapping relationship between the respiratory dynamics feature data and the fluid dynamics feature data, performs feature fusion analysis, and calculates the thoracic cavity-tubular coupling efficiency index. The central processing server makes a logical judgment on the patient's current postoperative recovery status based on the chest cavity-tubular coupling efficiency index, and identifies the chest cavity-tubular fluid transmission status, which includes at least the low-amplitude response state of the driving source, the pressure conduction blocking state, and the non-periodic fluid escape state. The central processing server dynamically generates graded device response strategies based on the identified thoracic cavity-tubular fluid transmission status. These graded device response strategies include adjusting the intensity of rehabilitation training, triggering tubular maintenance warnings, or generating analgesia pump parameter adjustment prompts.

[0006] Optionally, the method further includes a preoperative service path planning step: Acquire the patient's basic medical data upon admission, including chest CT images and pulmonary function test indicators; A three-dimensional model of the thoracic cavity is constructed based on the chest CT images, and the theoretical lung volume change curve is predicted by combining the lung function test indicators. Based on the theoretical lung volume change curve, an initial postoperative rehabilitation training plan is generated and stored in the patient's electronic health record.

[0007] Optionally, the method further includes a service correction step for the subjective pain dimension: During the acquisition of the real-time monitoring data, in response to the detection of abnormal respiratory characteristics, the bedside interactive terminal is triggered to receive the patient's subjective pain score feedback. The subjective pain score feedback is correlated with the respiratory dynamics characteristic data to calculate the pain-respiratory inhibition coefficient. When the pain-respiratory depression coefficient exceeds a preset threshold, the priority of the analgesia pump parameter adjustment prompt signal in the graded device response strategy is automatically increased.

[0008] Optionally, the method further includes a discharge follow-up service management step: At the patient's discharge stage, historical trend data of the thoracic cavity-duct coupling efficiency index during the patient's hospitalization are summarized; The historical trend data was analyzed using machine learning algorithms to predict the lung re-expansion completion time after the patient's discharge. Based on the prediction results, a personalized home rehabilitation guidance plan is automatically generated, and follow-up visit reminders are regularly pushed to patients via mobile devices.

[0009] Optionally, establishing the time-synchronized mapping relationship between the respiratory dynamics feature data and the fluid dynamics feature data specifically includes: Extract the time points of the thoracic expansion wave peaks from the respiratory dynamics feature data; Extract the time points of the wave peaks of the liquid surface fluctuations in the water-sealed bottle from the fluid dynamics characteristic data; Calculate the phase delay difference between the two peak time points and use the phase delay difference as a correction factor for evaluating the efficiency of pleural cavity pressure transmission; The real-time monitoring data is time-series aligned using the correction factor to generate a synchronization feature vector.

[0010] Optionally, the calculation of the thoracic cavity-vascular coupling efficiency index specifically includes: The pressure transmission gain is obtained by calculating the ratio of the rate of change of thoracic expansion amplitude to the rate of change of liquid level fluctuation amplitude in the water-sealed bottle within the synchronization time window. Combining the phase delay difference and the pressure conduction gain, the thoracic cavity-tubular coupling efficiency index is output using a preset fuzzy logic algorithm; The pleural-vascular coupling efficiency index is used to characterize the effectiveness of converting the patient's respiratory work into changes in negative pleural pressure.

[0011] Optionally, the step of identifying the pleural-tubular fluid transport status specifically includes: When the thoracic cavity-tubular coupling efficiency index shows that the thoracic expansion amplitude is normal but the fluctuation amplitude of the liquid level in the water seal bottle is significantly lower than the standard value, it is determined to be the pressure transmission blockage state; When the chest cavity-tube coupling efficiency index is in the high score range, but the retrospective analysis finds that the absolute values ​​of the patient's chest expansion amplitude and the fluctuation amplitude of the water seal bottle are significantly lower than the preset benchmark value, and the two maintain a high linear correlation, it is determined to be the low amplitude response state of the driving source. When a continuous bubble escape characteristic outside of a breathing cycle is detected in the fluid dynamics feature data, it is determined to be the non-periodic fluid escape state.

[0012] Optionally, the step of dynamically generating the hierarchical device response strategy specifically includes: In response to the pressure transmission blockage state, a drainage tube squeezing nursing task order is generated, and a level one audible and visual alarm is sent to the nurse station; If the low-amplitude response state of the driving source is determined to be non-painful limitation based on the pain score, an enhanced breathing training game instruction is generated to encourage the patient to take deep breaths through augmented reality feedback. For the non-periodic fluid escape state, the current rehabilitation training load parameters are automatically locked, and an emergency ward round request from a thoracic surgeon is generated.

[0013] Optionally, the data acquisition module acquires the real-time monitoring data in the following ways: The patient's chest wall micro-motion signals were collected using a millimeter-wave radar sensor deployed at the head of the bed, and the respiratory dynamics characteristic data were calculated. A machine vision module deployed on a flow guide frame is used to capture dynamic video of the liquid surface in a water-sealed bottle. The sequence of liquid column height changes is extracted using image semantic segmentation technology and used as the fluid dynamics feature data.

[0014] Secondly, an embodiment of the present invention provides a postoperative drainage monitoring data processing system for thoracic surgery, comprising: The data acquisition module is configured to acquire real-time monitoring data of patients after thoracic surgery. The real-time monitoring data includes the patient's thoracic respiratory dynamics characteristics and the fluid dynamics characteristics of the closed thoracic drainage system. A central processing server is configured to establish a time-synchronous mapping relationship between the respiratory dynamics feature data and the fluid dynamics feature data, and to perform feature fusion analysis to calculate the thoracic cavity-tubular coupling efficiency index; based on the thoracic cavity-tubular coupling efficiency index, the current postoperative recovery status of the patient is logically determined to identify the thoracic cavity-tubular fluid transmission status; and according to the identified thoracic cavity-tubular fluid transmission status, a graded device response strategy is dynamically generated. A bedside interactive terminal is configured to display the response strategy of the triage device and receive patient feedback. The graded equipment response strategy includes adjusting the intensity of rehabilitation training, triggering pipeline maintenance warnings, or generating analgesia pump parameter adjustment prompts.

[0015] The present invention has achieved the following beneficial effects: This invention constructs a time-synchronized mapping relationship between thoracic respiratory dynamics and drainage system fluid dynamics, creating a thoracic cavity-tubular coupling efficiency index. This effectively solves the problem of data silos between respiratory and drainage monitoring in traditional nursing care. This index can accurately quantify the efficiency of converting the patient's respiratory work into effective negative pressure in the pleural cavity. Therefore, when abnormal fluid level fluctuations occur, it automatically distinguishes between pressure transmission failure due to physical channel obstruction and insufficient driving force due to physiological respiratory restriction. This significantly improves the specificity of clinical monitoring and judgment, providing medical staff with objective extubation indications and complication warnings, and avoiding unnecessary tubing operations or treatment delays caused by misjudgment.

[0016] The present invention establishes an intelligent closed-loop service system from monitoring and identification to hierarchical intervention. The system not only improves the monitoring comfort of patients by using non-contact technology, but can also dynamically generate differentiated service strategies according to different identified thoracic-catheter fluid transmission states. For painful respiratory limitation, the system can automatically increase the priority of analgesic intervention and initiate augmented reality breathing training in combination with pain scores; for drainage obstruction, the system can accurately locate and trigger a nursing warning. This precise intervention mode based on pathophysiological status helps to improve the effectiveness and compliance of postoperative rehabilitation training for patients, reduce the risks of complications such as atelectasis, pleural effusion, and delayed pneumothorax, and thus shorten the hospitalization period of patients.

[0017] Other features and advantages of the present invention will be described in the following specification, and, in part, will be obvious from the specification, or will be understood by implementing the present invention. The objectives and other advantages of the present invention can be achieved and obtained by the structures specifically pointed out in the written specification and the accompanying drawings.

[0018] The technical solutions of the present invention will be further described in detail below through the accompanying drawings and embodiments. Brief Description of the Drawings

[0019] The accompanying drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the present invention, and do not constitute a limitation to the present invention. In the accompanying drawings: Figure 1 is a schematic diagram of a data processing system for postoperative drainage monitoring in thoracic surgery according to an embodiment of the present invention; Figure 2 is a flowchart of a data processing method for postoperative drainage monitoring in thoracic surgery according to an embodiment of the present invention. Detailed Embodiments

[0020] The following describes the preferred embodiments of the present invention with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to explain and illustrate the present invention, and are not used to limit the present invention.

[0021] Embodiment 1: This embodiment details the overall hardware architecture, deployment environment, and detailed technical parameters of each component of a data processing system for postoperative drainage monitoring in thoracic surgery. This system is designed to solve the technical problems in current postoperative care in thoracic surgery, such as the lack of linkage monitoring between the recovery status of patients' respiratory function and the working status of the thoracic drainage system, the absence of quantitative evaluation methods, and the lag in complication warning. As Figure 1As shown, the system constructs an intelligent medical service closed-loop environment based on the collaborative work of the Internet of Things, edge computing and cloud computing. At the physical level, it mainly covers the hospital's thoracic surgery intensive care unit, general wards, nurse station monitoring center and the hospital's internal private cloud server cluster.

[0022] First, a high-precision non-contact data acquisition module is deployed in each bed unit within the ward. This acquisition module is a multi-source heterogeneous sensor array designed to acquire medical-grade physiological and physical parameters without disturbing the patient's rest or increasing their physical burden (e.g., without the need for cables or electrodes). Specifically, for acquiring the patient's chest wall respiratory dynamics data, this embodiment uses a millimeter-wave radar sensor. Compared to traditional contact breathing straps or patch electrodes, millimeter-wave radar can penetrate the patient's clothing and bedding in a non-invasive and unrestricted manner to capture minute displacement changes in the chest wall. Preferably, this millimeter-wave radar adopts a frequency-modulated continuous wave (FMCW) system, with its operating frequency set in the 77GHz to 81GHz band. This band has an extremely short wavelength, providing millimeter-level range resolution and extremely high Doppler velocity sensitivity, suitable for capturing minute chest wall fluctuations caused by respiration. The radar's antenna array is designed with a multiple-transmit, multiple-receive (MIMO) architecture, such as a three-transmit, four-receive or four-transmit, four-receive configuration, to achieve high-precision angular resolution and three-dimensional point cloud imaging of the chest wall. Preferably, during installation, the radar device is fixed within a range of approximately 0.8 to 1.5 meters from the patient's chest using a medical-grade ABS universal bracket. This can be achieved by mounting it on an IV stand, equipment belt, or dedicated cantilever bracket at the bedside, with its beam center axis precisely pointing towards the area from the patient's sternal manubrium to the xiphoid process. To eliminate interference from other moving objects in the ward (such as the movement of caregivers, vibrations from the IV pump, or curtain swaying caused by air conditioning), the radar device incorporates a digital signal processor (DSP) that runs an adaptive beamforming algorithm. Through digital beamforming technology, the detection area is strictly locked within the bedside area, forming a virtual electronic fence for respiratory monitoring.

[0023] To collect fluid dynamics data for closed thoracic drainage systems, this embodiment deploys a dedicated intelligent machine vision module in the bedside drainage bottle placement area. Because postoperative drainage bottles (typically water-seal bottles) in thoracic surgery are placed low (usually hung by the bedside or placed on a floor drainage stand), and because the bottles are transparent and contain liquid, they are easily affected by lighting conditions. This vision module includes a high dynamic range (HDR), high-resolution industrial-grade RGB camera and a dedicated 940nm infrared supplemental light. The 940nm wavelength was chosen because it is invisible light and will not produce a "red glow" phenomenon when turned on at night, ensuring that nurses can clearly observe the drainage bottle status without affecting the patient's sleep. The camera's field of view (FOV) is specially designed to be approximately 90 to 110 degrees, fully covering the liquid level observation window of the long glass tube (water-seal tube) of conventional single- or double-bottle drainage systems, the area where air bubbles escape from the bottom of the water-seal chamber, and the drainage fluid collection chamber area. To accommodate different brands and sizes of drainage bottles, this module is equipped with a fast autofocus motor and macro shooting function, with a minimum focusing distance of 5 cm. The camera device is clamped to the drainage rack via a flexible serpentine bracket or a strong magnetic base, ensuring that the camera and drainage bottle remain relatively stationary. To improve the stability of image recognition, the camera device integrates a three-axis accelerometer (IMU) for real-time monitoring of the camera's attitude angle. In the event of an accidental collision (such as a cleaner accidentally bumping into it while cleaning) causing the camera to shift or tip over, the system can detect this through IMU data and automatically trigger a field-of-view anomaly alarm, prompting medical staff to perform calibration.

[0024] The data acquisition module transmits pre-processed data to the central processing server in real time via a dedicated medical Wi-Fi 6 (802.11ax) LAN or a 5G slice network within the ward. For data transmission protocols, a high-speed real-time streaming protocol based on User Datagram Protocol (UDP) is used for raw radar data to ensure low latency of respiratory waveforms; for control commands, status heartbeat packets, and alarm information, a Message Queuing Telemetry Transport Protocol (MQTT) is used to ensure the reliability and order of command arrival. Simultaneously, to strictly prevent the leakage of patient privacy data, all transmitted data packets are encrypted with AES-256-bit high-strength encryption, and an SSL / TLS secure tunnel is established in the transmission link to ensure data security at the air interface.

[0025] The central processing server is deployed in the hospital's data center or private cloud platform, meeting the national information security level protection level three standard. The server cluster is equipped with high-performance graphics processing unit (GPU) accelerator cards for parallel processing of radar signal FFT transformation, point cloud clustering, and deep learning image analysis tasks for machine vision. The server runs a real-time operating system or containerized cluster, responsible for data access gateway, feature extraction engine, time-series alignment module, multimodal fusion algorithm module, clinical decision support system (CDSS), and business logic services.

[0026] In addition to the data acquisition and processing terminals, the system also includes a bedside interactive terminal located at the patient's bedside. This terminal uses a 10-inch to 15.6-inch medical-grade capacitive touchscreen display, suspended from the bedside via a swing arm for easy operation by the patient while lying down. The terminal integrates a microphone array and stereo speakers, supporting far-field voice wake-up and interaction. It not only provides a window for patients to view their recovery progress and receive postoperative education and medical order reminders, but also allows the system to output intervention commands (such as guided breathing training games and pain assessment questionnaires). Simultaneously, at the nurses' station, the system is equipped with a central monitoring screen and a mobile nursing handheld terminal (PDA) to display in real-time a heat map of the chest cavity-tubular coupling efficiency index, a risk warning list, and nursing task work orders for all patients in the ward.

[0027] Example 2: This embodiment details the working principle of the data acquisition module in acquiring real-time monitoring data of patients after thoracic surgery, such as... Figure 2 As shown, in particular, how to extract accurate respiratory dynamics and fluid dynamics data from raw physical signals.

[0028] In the acquisition of chest wall respiratory dynamics data, after the millimeter-wave radar device is activated, it transmits continuous linear frequency-modulated pulses to the patient's chest. The electromagnetic waves reflect back to the radar receiving antenna after contacting the chest wall. Due to the periodic, minute displacements of the chest wall during respiration (typically between 2mm and 10mm), the phase and frequency of the reflected echo undergo a Doppler shift relative to the transmitted signal. Upon receiving the radar echo signal, the edge computing unit or central processing server first performs signal preprocessing, including removing the DC component and windowing (e.g., Hanning window) to suppress spectral leakage. Next, a fast Fourier transform (FFT) is performed in the range dimension to convert the time-domain signal into a range-frequency spectrum, thus obtaining a range profile. In the range profile, the system uses an adaptive constant false alarm rate (CFAR) detection algorithm to extract target point clusters representing chest wall motion from complex background noise (such as bed rail reflections). Subsequently, within the locked range cells, the system extracts the phase signal. Due to the extremely short wavelength of millimeter waves, phase changes are highly sensitive to displacement. The system uses a phase unwinding algorithm to restore the phase changes to a continuous chest wall displacement time-series curve. To eliminate the micro-flicker signals (frequency approximately 1Hz-1.5Hz) caused by heartbeats and low-frequency drift caused by random body movements, the system utilizes a cascaded digital filter bank, including a high-pass filter with a cutoff frequency of 0.05Hz and a low-pass filter with a cutoff frequency of 0.8Hz, to obtain a clean chest wall respiratory motion waveform. Based on this waveform, the feature extraction engine further calculates the following key metrics: Respiratory rate (RR, respiratory counts per minute), respiratory depth (waveform peak-to-trough difference, which characterizes tidal volume trend after body size normalization), inspiratory-to-expiratory ratio (I:E ratio, the ratio of inspiratory time to expiratory time, used to assess obstructive ventilatory dysfunction), and respiratory stability index (calculated based on the coefficient of variation of the duration and amplitude of adjacent respiratory cycles).

[0029] In addition, by utilizing the angular resolution capability of MIMO radar, the system can also extract the displacement waveforms of the left and right chest separately. By calculating the correlation coefficient and amplitude ratio of the bilateral waveforms, an index of chest wall movement symmetry can be generated, which is of great clinical significance for judging unilateral atelectasis, pleural effusion or diaphragmatic paralysis.

[0030] Next, the process of acquiring hydrodynamic feature data for the closed thoracic drainage system is explained. The machine vision module continuously captures video streams of the water-sealed bottle (typically at 30fps or 60fps) and transmits them to the server. The server-side image processing engine first uses a deep learning-based object detection network (such as a pruned version of YOLOv5 or MobileNet-SSD) to automatically locate key structural regions of the water-sealed bottle in the video frames, including the water column tube (long glass tube), the bottom of the water seal chamber, and the liquid level markings. After localization, the system uses a combination of traditional computer vision techniques and deep learning for refined feature extraction. Specifically, for the water column tube region, the system uses the Canny edge detection operator combined with the Hough Transform to accurately identify the tube wall boundaries within the region of interest (ROI), limiting the liquid level search range. Then, using grayscale projection or gradient maximum methods, the concave bottom contour of the liquid surface at the interface between the liquid level and air inside the tube is accurately located. To improve localization accuracy, the algorithm employs subpixel subdivision technology, increasing the measurement resolution of the liquid level height to the 0.1 mm level. The system tracks the pixel coordinate changes of the liquid surface contour in the vertical direction in real time, generating a liquid surface fluctuation time-series curve. To convert pixel coordinates into physical height (millimeters of water column), the system automatically identifies pre-printed standard scale lines or specific markers (such as QR codes) on the drainage bottle after initial installation or each position change, and calculates the ratio coefficient between pixels and physical length. By analyzing the liquid surface fluctuation time-series curve, the system can extract the liquid surface fluctuation amplitude (the height difference between the highest point at the end of inspiration and the lowest point at the end of expiration, directly representing the range of changes in negative pressure in the pleural cavity), fluctuation frequency, and the drift trend of the fluctuation baseline (reflecting changes in the amount of pleural effusion or changes in the position of the drainage tube). For the bottom area of ​​the water seal chamber, the system uses frame difference or optical flow to monitor for bubble generation. When a moving target is detected, the system further analyzes its morphological characteristics (such as roundness and area) and texture features to distinguish between real bubbles and reflections caused by liquid sloshing. If a bubble is confirmed, the system will record the time of bubble formation, duration, and density of the bubble cluster (bubble flux per unit time) to quantify the degree of leakage.

[0031] Example 3: This embodiment focuses on describing the principle of establishing a time-synchronous mapping relationship between respiratory dynamics characteristic data and fluid dynamics characteristic data, as well as the calculation principle of the thoracic cavity-tubular coupling efficiency index.

[0032] In clinical physiology and fluid dynamics models, thoracic cavity movement is the driving source (input signal), changes in intrapleural negative pressure are the intermediate process, and fluctuations in the fluid level of the water-sealed bottle are the final response (output signal). A strict physical causal chain and signal transmission delay exist among these three. However, in engineering implementation, radar data acquisition systems and visual acquisition systems typically operate in independent hardware clock domains, and network transmission delays are subject to random jitter. Directly comparing two sets of data may create the illusion of misalignment between respiratory peaks and fluid level peaks on the time axis, leading to incorrect monitoring judgments (e.g., misdiagnosis as tubing blockage or delay). Therefore, high-precision time alignment is essential.

[0033] The system first uses the Network Time Protocol (NTP) to calibrate the timing of the radar and camera. At the data processing end, the server establishes a unified virtual timeline container. When radar and video data frames arrive at the server, the system parses the source acquisition timestamp (not the arrival timestamp) carried in the data packet header. Using linear interpolation or cubic spline interpolation algorithms, the two sets of non-uniformly sampled data are resampled to the same frequency (e.g., uniformly set to 50Hz), thus achieving a one-to-one correspondence between data points.

[0034] After completing the data hard alignment, the system begins feature fusion analysis to calculate the soft latency. First, the system extracts the inspiratory peak of the respiratory waveform (i.e., the moment of maximum chest expansion) as the reference clock signal. In an ideal, unobstructed drainage system, when the thoracic cavity expands to its maximum, the negative pressure in the pleural cavity reaches its maximum (absolute value), at which point the fluid level in the water-seal bottle should also be drawn to its highest point. However, due to the length of the drainage tube (usually 1-1.5 meters), its diameter, the viscous resistance of the fluid (gas or liquid) within the tube, and the compliance of the pleural tissue, the peak time of the fluid level fluctuation varies. Often lags behind The system calculates the time difference between these two moments. This is the phase delay difference. This parameter reflects the damping characteristics of the pressure wave along its propagation path. Before extracting the peak time point, the system first performs a validity pre-check: calculating the effective value of the amplitude of the water-sealed bottle's liquid surface fluctuation. If this effective value is lower than a preset quiescent threshold (e.g., 2mm), it is determined that the drainage system is in a fluid-static state at the current moment. The system directly marks the phase delay difference in subsequent calculations as infinite and skips the peak extraction step to prevent algorithmic errors caused by signal loss. Before extracting the phase delay difference, the system first performs a signal validity pre-check. The standard deviation of the water-sealed bottle's liquid surface fluctuation sequence is calculated. ,like If the pulse delay is below a preset silence threshold (e.g., 0.5 mm), the drainage system is determined to be in a state of pressure conduction blockage or no drive. In this case, the phase delay difference is directly assigned to the preset maximum penalty value (e.g., 2000 ms), and subsequent peak extraction steps are skipped to prevent algorithm logic errors. For signals that pass the pre-detection, to improve computational robustness under weak respiration, this system preferably uses sliding window cross-correlation analysis instead of single peak matching. Specifically, the system extracts the respiratory displacement sequence within the synchronization time window. and liquid surface fluctuation sequence Calculate the cross-correlation function between the two. Within a preset physiological delay range (0-1000ms), search for... Time shift that reaches the global maximum value ,Should This is the precise phase delay difference.

[0035] Simultaneously, the system calculates the pressure conduction gain. The system selects a stable respiratory cycle window (e.g., 30 seconds) and calculates the normalized root mean square value of the chest expansion amplitude within that window. ) and the normalized root mean square value of the liquid level fluctuation amplitude ( The ratio of the two. This refers to the pressure conduction gain. To accurately reflect the attenuation of respiratory drive energy after transmission through the tubing, the system employs a physical dimension unification and benchmark calibration method. First, using the calibration parameters described in Example 2, the phase signal acquired by the radar and the pixel signal extracted visually are uniformly converted into physical displacement (millimeters). Second, during the initialization phase (or when the tubing is confirmed to be unobstructed), the system records the benchmark conduction coefficient (…). In real-time monitoring, the pressure transmission gain is defined as: Regarding the reference transmission coefficient To avoid baseline deviations caused by tubing compression or weak breathing in the initial postoperative state, this system introduces a dynamic calibration mechanism based on compression and repositioning. Upon system startup, or after the machine vision module detects significant deformation and rapid rebound of the drainage tube (corresponding to a standard compression procedure performed by the nurse), the system automatically triggers a calibration time window (e.g., 30 seconds). Because the negative pressure generated by the compression operation temporarily clears physical blockages in the tubing, the fluid transport state at this time is closest to the ideal value. The system captures the average chest expansion amplitude within this time window. Average amplitude of liquid level fluctuation Calculate the ratio of the two. The system parameters are updated. If no squeezing action is detected for an extended period, the system will use a preset standard coefficient from the database based on a model with the same pipe diameter and length as a temporary reference. This algorithm preserves the absolute amplitude characteristics of the signal, ensuring that when the drainage channel is blocked ( When the value is reduced, the Gain value shows a significant decrease.

[0036] Based on the phase delay difference calculated above and pressure conduction gain The system constructs a multi-dimensional evaluation model to generate a thoracic cavity-tubular coupling efficiency index. This model is a fuzzy logic-based inference system. Specifically, this embodiment discloses the detailed configuration of the fuzzy logic algorithm, employing the Mamdani inference model: Input variable fuzzification: Set the domain of the phase delay difference to [0, 1000] milliseconds, and define three fuzzy sets: {"synchronization" (0-150ms, trapezoidal membership function), "delay" (150-500ms, triangular membership function), "blocking lag" (>500ms, trapezoidal membership function)}.

[0037] Set the universe of discourse for pressure conduction gain to [0, 2.0], and define three fuzzy sets: {"attenuation" (0-0.4), "matching" (0.4-1.2), "amplification" (>1.2)}.

[0038] In this embodiment, the parameters of the trapezoidal membership function are set as follows (taking phase delay difference as an example): the vertex parameters [a,b,c,d] of the synchronization set are set to [-50,0,100,150]ms; the vertex parameters of the blocking hysteresis set are set to [500,600,2000,2000]ms. If the system cannot detect an effective liquid surface peak, the maximum delay is directly assigned.

[0039] Furthermore, the membership function of the phase delay difference The definition is as follows: Synchronization set (trapezoidal distribution): when hour, ;when It decays rapidly to 0.

[0040] Blocked lag set (Sigmoid or trapezoidal distribution): when hour, .

[0041] This specific numerical definition is based on the hydrodynamic characteristics of clinical thoracic surgical drainage, ensuring the robustness of the algorithm in identifying tubing blockages.

[0042] Fuzzy rule base: The system pre-configures the following core logic rules: Rule 1: IF phase delay difference IS “synchronous” AND pressure conduction gain IS “matched” THENCE IIS “superior” (output value 85-100); Rule 2: IF phase delay difference IS “blocking hysteresis” OR pressure conduction gain IS “attenuation” THENCEIS “poor” (output value 0-40); Rule 3: IF phase delay difference IS “delay” AND pressure conduction gain IS “match” THENCE IIS “good” (output value 60-85).

[0043] Table 1 below lists the preset core fuzzy control rule base in this embodiment.

[0044] Table 1. Fuzzy rule base for thoracic cavity-vascular coupling effectiveness index (CEI)

[0045] Defuzzification: The centroid method is used to calculate the abscissa of the geometric center of the output fuzzy set, and a unique precise value is obtained as the CEI index.

[0046] This index reflects whether the patient's respiratory work is effectively converted into negative pressure fluctuations in the pleural cavity, and is a core digital indicator for assessing the patency of the drainage system and the lung's re-expansion capacity. Compared to the traditional method of relying solely on nurses' visual observation of water column fluctuations, this index can quantify minute obstruction trends.

[0047] Example 4: This embodiment details the specific logical determination process for identifying three typical thoracic cavity-tubular fluid transport states (low-amplitude response state of the driving source, pressure conduction blocking state, and non-periodic fluid escape state) based on the thoracic cavity-tubular coupling efficiency index and original feature data.

[0048] The first case: Identification of the pressure conduction blocking state.

[0049] In the early postoperative period, blood clots or fibrin blockage of the drainage tube, or compression and twisting of the drainage tube, are extremely high-risk complications. If blockage occurs, pressure changes within the pleural cavity cannot be effectively transmitted to the water-seal bottle. At this time, the system's logic engine will detect the following combination of characteristics: Radar data showed that the patient's chest wall breathing movements were normal, and even compensatory enhancement (increased work of breathing, faster frequency) occurred due to shortness of breath. However, machine vision data showed that the fluctuation of the liquid level in the water-sealed bottle was minimal or even static, resulting in a significant decrease in the calculated pressure conduction gain and a sharp drop in the CEI index (usually below 40 points). Furthermore, the system also analyzes phase delay. If partial blockage occurs (such as a moving blood clot in the tubing), it will manifest as increased fluid resistance, leading to an abnormally large and unstable phase delay. When the system detects the above characteristic pattern continuously over multiple respiratory cycles (e.g., 5 consecutive cycles), it determines a pressure conduction blockage and triggers the highest level alarm. To rule out false positives (such as an obstructed drainage bottle), the system simultaneously checks the image quality of the camera, only issuing an alarm after confirming there is no obstruction.

[0050] The second scenario: Identification of the low-amplitude response state of the driving source.

[0051] This is typically caused by patient pain fear, residual sedative medication, respiratory muscle weakness, or atelectasis. In this situation, the radar-monitored chest expansion is minimal (shallow and rapid breathing), and simultaneously, due to the weak driving force (respiratory movement), the fluctuation amplitude of the fluid level in the water seal bottle also decreases. The logic engine will detect a highly positively correlated linear decrease in both amplitudes, while the phase delay remains within the normal range (indicating that the tubing is physically patent, but the source of power is insufficient). At this time, the CEI index may remain at a moderate level (e.g., 50-70 points) because the gain ratio is acceptable, but the absolute physical values ​​are all below the preset clinical baseline. To further distinguish between pain-related and non-pain-related respiratory limitation, the system will access the pain score data (VAS score) from the bedside interactive terminal. If the pain score is high (e.g., >4 points) and breathing is weak, the system determines it as pain-related respiratory limitation; if the pain score is low but breathing is weak, it is determined as non-pain-related respiratory limitation (e.g., muscle weakness, recovery period from general anesthesia, or poor compliance).

[0052] The third case: Identification of non-periodic fluid escape states.

[0053] This indicates alveolar rupture, bronchopleural fistula, or air leakage at the lung resection site. The key characteristics are primarily reflected in machine vision data. The system detects a non-periodic, continuous bubble escape signal in the water seal chamber area. Unlike normal coughing and exhalation (usually brief and accompanied by vigorous chest movement), pathological air leakage often manifests as a continuous, fine stream of bubbles, even during quiet breathing or breath-holding. The system calculates the bubble escape flux (an estimated total volume of bubbles per unit time) and considers chest movement. If a bubble flow is detected, and the baseline of the fluid level fluctuation shows a downward trend (indicating increased intrapleural air accumulation and loss of negative pressure), the system classifies it as a severe non-periodic fluid escape state.

[0054] To improve the robustness of the judgment, the system also introduces a confidence assessment mechanism. For each judgment result, the system calculates a weighted confidence score (0-1) based on the signal-to-noise ratio of the data, the significance of the features, and the consistency of historical trends. Only when the confidence score exceeds 0.85 (this threshold can be adjusted according to clinical needs) will the system officially generate a monitoring judgment label and push it to medical staff; otherwise, it will be marked as suspected and medical staff will be prompted to conduct manual review.

[0055] Example 5: This embodiment describes the specific execution method of the system dynamically generating a graded equipment response strategy based on the identified thoracic cavity-tubular fluid transmission status.

[0056] Based on the pressure conduction blockage determination result, the system-generated service strategy aims to quickly restore tubing patency. The system first issues a red audible and visual alarm via the healthcare workstation and clearly indicates the possible location of the blockage in a pop-up window (estimated based on waveform distortion analysis using a fluid dynamics model). Simultaneously, the system automatically generates a drainage tube compression nursing task sheet, which is pushed to the responsible nurse's handheld PDA via the hospital's internal communication software. This task sheet details the standard procedures for the compression operation (e.g., "squeeze spirally from the proximal end to the distal end"). After the nurse completes the operation, the system automatically enters a treatment evaluation mode, frequently monitoring fluid level fluctuations over the next 5 minutes. If the fluctuation amplitude recovers, the system automatically de-alarms and records the nursing care as successful; if there is still no improvement, the system escalates the alarm level and calls a doctor to perform flushing or tube replacement.

[0057] For low-amplitude response state determinations of the driving source, especially painful respiratory limitation, the system's strategy focuses on analgesia and motivation. The system sends suggestions to the physician to adjust the analgesia regimen (such as "suggest increasing background infusion dose" or "assess whether nerve block is needed"), and simultaneously launches an augmented reality breathing training game on the bedside interactive terminal. This game uses real-time respiratory depth data from radar feedback as the game control input. For example, the screen displays an animation of blowing up a balloon or candle; the deeper the patient inhales, the larger the balloon becomes. This visual and gamified feedback mechanism can divert the patient's attention from pain and intuitively motivate them to increase tidal volume. The system dynamically adjusts the game difficulty (i.e., the required respiratory depth threshold) based on the patient's real-time performance. For non-painful respiratory limitation, the system generates targeted physical intervention instructions, such as prompting the nurse to assist the patient with percussion and expectoration, or instructing the patient to use a breathing trainer (three-ball device). The system records the patient's use of the breathing trainer via camera and uses AI visual algorithms to automatically identify the height of the float's ascent, quantifying the training effect and generating daily reports for physician review.

[0058] For non-periodic fluid escape state assessments, the system's strategy focuses on safety monitoring and workload management. The system automatically locks the current rehabilitation training plan, suspending all high-intensity chest expansion exercises or violent coughing inductions to prevent the leak from widening. Simultaneously, the system generates an emergency ward round request from a thoracic surgeon and saves video clips before and after the leak, serving as crucial evidence for the doctor's monitoring and assessment. During the doctor's intervention, the system continuously monitors the changing trend of the bubble escape rate; if an exponential increase in the rate is detected, a department-wide broadcast alarm is immediately triggered, indicating the risk of tension pneumothorax.

[0059] Example 6: This embodiment details the steps involved in planning the preoperative service pathway.

[0060] After the patient is admitted, the system first obtains the patient's chest CT image data (DICOM format) through the hospital's PACS (Picture Archiving and Communication System). The central processing server uses 3D reconstruction algorithms (such as Marching Cubes) to perform high-precision 3D segmentation and reconstruction of lung tissue, tracheal tree, blood vessels, and thoracic skeleton. In the 3D model, the system can accurately calculate the total lung volume and estimate the residual lung volume after lesion resection.

[0061] Simultaneously, the system connects to a pulmonary function testing system to obtain physiological indicators such as the patient's forced vital capacity (FVC) and forced expiratory volume in one second (FEV1). Based on biological soft tissue finite element analysis (FEA) technology, a pre-set constitutive model of the hyperelastic material of the lung parenchyma (such as the Neo-Hookean model, with Young's modulus set to 1-5 kPa) is used. The system maps the pulmonary function indicators onto a three-dimensional lung model to simulate the lung lobe deformation and airway resistance distribution of the patient under different respiratory intensities.

[0062] Based on the simulation results, the system constructs a theoretical lung volume change curve for the patient after surgery. This curve represents the expected lung re-expansion process from day 1 to day 7 post-surgery under ideal rehabilitation conditions. Using this as a benchmark, the system generates an initial post-operative rehabilitation training plan. This plan uses quantifiable indicators, such as: "Post-operative day 1 goal: chest expansion of 1.5 cm and fluid level fluctuation of 3 cm with each inhalation, with a cumulative training time of 30 minutes per day; Post-operative day 3 goal: chest expansion of 2.5 cm with each inhalation." This personalized plan is stored in the patient's electronic health record and serves as the benchmark parameter for daily rehabilitation guidance via the bedside interactive terminal. By comparing actual post-operative monitoring data with the pre-operative planning benchmark, the system can accurately evaluate whether the rehabilitation progress is ahead of schedule or behind schedule, thereby dynamically adjusting the subsequent training intensity.

[0063] Example 7: This embodiment focuses on explaining the underlying logic of pain correction mechanisms.

[0064] Step S1: Abnormal Trigger. The system first continuously monitors the patient's respiratory dynamics. When the system detects that the patient's respiratory depth (tidal volume proxy index) has decreased by more than 30% of the baseline value within 3 consecutive respiratory cycles (i.e., respiratory limitation characteristics have appeared), the central processing server determines that the trigger condition has been met.

[0065] Step S2: Interactive Guidance. The system immediately wakes up the bedside interactive terminal, emits a gentle prompt tone, and automatically pops up a slider, asking the patient in voice: "Do you feel your pain is getting worse? Please slide the slider."

[0066] Step S3: Difference Calculation. After the patient responds, the system obtains the current VAS score ( The system retrieves the historical score from the last evaluation. ), calculate the change in rating .

[0067] Step S4: Coefficient Generation. The pain-respiratory inhibition coefficient is calculated using a normalized and nonlinearly corrected model. : in, The relative decay rate of respiratory depth (dimensionless). This represents the change in pain scores (absolute value). This is a preset sensitivity adjustment factor (typically 0.3). The mathematical model introduces an exponential function as the denominator, which avoids the bias when... The computational divergence risk is close to zero, which, on the other hand, aligns with the nonlinear characteristics of biological perception in the Weber-Fechner Law. When the calculated... When the pain level exceeds a preset threshold (e.g., 1.2), pain is determined to be the dominant factor causing respiratory limitation.

[0068] Through this event-triggered interaction logic, the present invention avoids the unreasonable requirement for patients to perform high-frequency continuous scoring, and intervenes in the assessment only at critical moments when there is a suspected change in the breathing pattern, thereby establishing the clinical feasibility of the function while ensuring the validity of the data.

[0069] The pain coefficient is derived by calculating the cross-correlation between the rate of change in pain scores and the rate of decrease in respiratory depth. The specific logic is as follows: The system continuously monitors the patient's subjective rating input and objective respiratory waveform. If the system detects that whenever the patient increases their pain score, their respiratory depth immediately shows a significant truncated drop (i.e., abruptly stopping halfway through inspiration, exhibiting a cliff-like waveform), and their respiratory rate increases, then the calculated pain-respiratory inhibition coefficient is high. This indicates that pain is the primary factor limiting breathing, and the patient's complaint is credible. Conversely, if the patient gives a very high pain score (e.g., 10), but radar monitoring shows that they can still maintain a normal respiratory rhythm and depth with a smooth waveform, it suggests that the patient may have excessive anxiety or a high pain tolerance. In this case, the coefficient is low.

[0070] Based on this objective coefficient, the system can more intelligently adjust analgesia strategies. For patients with high coefficients, the system strongly recommends that doctors increase analgesia; for patients with low coefficients, the system suggests prioritizing psychological counseling or placebo therapy to avoid the respiratory depression or addiction caused by excessive opioid use. In addition, the system also incorporates heart rate variability (HRV) as an auxiliary objective indicator for assessing pain stress, further improving the accuracy of pain assessment.

[0071] Example 8: This embodiment describes the management of discharge follow-up services.

[0072] As the patient is about to be discharged, the system automatically performs a comprehensive data mining task. It retrieves all CEI (Chronic Energy Indicator), respiratory waveforms, and training achievement rates data from the patient's hospitalization period. Using a Long Short-Term Memory (LSTM) network, a deep learning model well-suited for processing time-series data, it predicts the patient's future lung function recovery trend. The prediction results are specified as the estimated lung recruitment completion time and the estimated time to full labor capacity. Specifically, an LSTM network model is constructed, whose input layer receives time-series feature vectors. The model includes the following specific dimensions: 1) Hourly average cavity-to-tubular coupling efficiency index (CEI); 2) Respiratory rate coefficient of variation (RRV); 3) Pressure conduction gain; 4) Pixel flux of air bubble escape per unit time; and 5) Frequency of analgesia pump compressions. The model's time step is set to 24 hours, and the probability of lung recruitment completion is output through a fully connected layer.

[0073] Based on the prediction results, the system generates a detailed home rehabilitation guidance plan. This plan will be pushed to the patient's mobile app or WeChat mini-program. The plan includes video tutorials that match the training movements performed during the patient's hospitalization, ensuring the accuracy of the movements.

[0074] Furthermore, the system establishes a closed-loop remote follow-up system based on mobile devices. Patients can upload videos of themselves taking deep breaths at home using their mobile phone camera, or wear a simple home respiratory counter (based on the IMU principle). The AI ​​algorithm on the backend of the system analyzes this uploaded data to determine if there are any risks associated with home rehabilitation (such as delayed pneumothorax or recurrence of pleural effusion). Once an abnormal trend is detected (such as a gradually increasing respiratory rate for three consecutive days, or a patient complaining of shortness of breath), the system will immediately send a follow-up appointment reminder to the patient and simultaneously notify the attending physician's follow-up workstation.

[0075] Example 9: To ensure medical-grade data reliability, the system incorporates a rigorous self-test procedure upon startup. Each time it starts, the millimeter-wave radar performs background noise calibration to detect any co-channel interference sources in the surrounding environment; the machine vision module uses standard color blocks on a water-sealed bottle to calibrate white balance and exposure parameters. If the self-test fails (e.g., the camera is obstructed, the radar signal-to-noise ratio is too low, or the network connection is unstable), the system will prevent entry into monitoring mode and report an error to the nurses' station to prevent erroneous data from misleading medical decisions.

[0076] During operation, the system also features an abnormal data interruption mechanism. If a non-physical jump in data is detected within a very short time (e.g., a 20 cm jump in liquid level within 0.1 seconds, which significantly deviates from the preset physiological / physical model range), the system will determine it as a sensor malfunction or data transmission error, automatically discard the data frame, and initiate a smooth interpolation algorithm to fill the gap, while simultaneously recording a fault log. This is to prevent false alarms caused by occasional sensor malfunctions.

[0077] Example 10: In the data acquisition module, although the preferred embodiment uses millimeter-wave radar to monitor chest wall movement, those skilled in the art should understand that other non-contact or contact sensors can also achieve the same function. For example, a time-of-flight (ToF) depth camera or a structured light 3D camera can be used to capture chest wall undulations; piezoelectric ceramic breathing straps or flexible fabric strain sensors worn on the patient's chest can also be used to acquire respiratory signals. Similarly, for monitoring drainage bottles, in addition to visible light cameras, ultrasonic level sensors, capacitive level sensors, or laser triangulation sensors can also be used to monitor the fluid level. Any physical quantities that characterize chest wall movement and changes in drainage fluid level are within the scope of this application's technical solution.

[0078] In terms of algorithm implementation, although the preferred embodiment employs fuzzy logic and deep learning, the core of this application lies in establishing the correlation between respiration and drainage. Therefore, using traditional statistical methods (such as Pearson correlation coefficient analysis), adaptive thresholding, or Kalman filter state estimation based on physical models to implement the above logical judgment does not fundamentally deviate from the technical concept of this application.

[0079] While this application uses post-thoracic surgery as an example, its principles are equally applicable to other medical scenarios requiring closed drainage, such as ventricular drainage management in neurosurgery and abdominal drainage management in general surgery. The system can be readily adapted by simply adjusting the corresponding parameter thresholds based on the fluid dynamic characteristics of different drainage routes.

[0080] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A thoracic surgery postoperative drainage monitoring data processing method applied to a thoracic surgery postoperative drainage monitoring data processing system, the system comprising a data acquisition module, a central processing server and a bedside interactive terminal, the method comprising the following steps: obtaining real-time monitoring data of a thoracic surgery postoperative patient through the data acquisition module, the real-time monitoring data comprising thoracic respiration dynamics characteristic data of the patient and fluid dynamics characteristic data of a closed thoracic drainage system; characterized in that the method further comprises the following steps: the central processing server establishes a time synchronization mapping relationship between the respiration dynamics characteristic data and the fluid dynamics characteristic data, and performs characteristic fusion analysis to calculate a thoracic-tube coupling efficiency index; the central processing server performs logical judgment on the current postoperative recovery state of the patient based on the thoracic-tube coupling efficiency index, identifies the thoracic-tube fluid transmission state, and the thoracic-tube fluid transmission state at least includes a low-amplitude response state of a driving source, a pressure conduction blockage state and a non-periodic fluid escape state; the central processing server dynamically generates a hierarchical device response strategy according to the identified thoracic-tube fluid transmission state, and the hierarchical device response strategy includes adjusting the rehabilitation training intensity, triggering the tube maintenance early warning or generating the analgesic pump parameter adjustment prompt signal.

2. The method of claim 1, wherein, The method further comprises a preoperative service path planning step: obtaining the patient's admission basic diagnosis and treatment data, the diagnosis and treatment data including chest CT image and lung function test indicators; constructing a thoracic three-dimensional model based on the chest CT image, combining the lung function test indicators to predict the theoretical lung volume change curve of the patient after the operation; generating an initialized postoperative rehabilitation training plan according to the theoretical lung volume change curve, and storing the plan in the patient's electronic health record.

3. The method of claim 1, wherein the method further comprises: The method further comprises a subjective pain dimension service correction step: in the process of obtaining the real-time monitoring data, in response to monitoring abnormal respiration characteristics, triggering the bedside interactive terminal to receive the patient's subjective pain score feedback; correlation analysis of the subjective pain score feedback and the respiration dynamics characteristic data is performed to calculate a pain-respiration suppression coefficient; when the pain-respiration suppression coefficient exceeds a preset threshold, the priority of the analgesic pump parameter adjustment prompt signal in the hierarchical device response strategy is automatically raised.

4. The data processing method for postoperative drainage monitoring in thoracic surgery according to claim 1, characterized in that, The method further comprises a discharge follow-up service management step: in the patient discharge stage, the historical trend data of the thoracic-tube coupling efficiency index during the patient's hospitalization is summarized; using a machine learning algorithm to analyze the historical trend data to predict the lung recruitment completion time after the patient is discharged; according to the prediction result, an individualized home rehabilitation guidance scheme is automatically generated, and the patient is regularly pushed for re-examination reminders through a mobile terminal.

5. The data processing method for postoperative drainage monitoring in thoracic surgery according to claim 1, characterized in that, The establishment of the time synchronization mapping relationship between the respiration dynamics characteristic data and the fluid dynamics characteristic data specifically comprises: extracting the thoracic expansion peak time point in the respiration dynamics characteristic data; extracting the water seal bottle liquid level fluctuation peak time point in the fluid dynamics characteristic data; calculating a phase delay difference between two of the wave peak time points, and taking the phase delay difference as a correction factor for evaluating the pressure transmission efficiency of the pleural cavity; performing time series alignment on the real-time monitoring data using the correction factor, and generating a synchronized feature vector.

6. A method of processing post-thoracic surgery drainage monitoring data according to claim 5, characterized in that, The calculation obtains a chest-tube coupling performance index, specifically comprising: calculating a ratio of a chest expansion amplitude change rate to a water seal bottle liquid surface fluctuation amplitude change rate in a synchronized time window to obtain a pressure conduction gain; combining the phase delay difference and the pressure conduction gain, and using a preset fuzzy logic algorithm to output the chest-tube coupling performance index; The chest-tube coupling performance index is used to represent the effectiveness of the conversion of respiratory work into negative pressure changes in the pleural cavity.

7. The post-operative drainage monitoring data processing method for thoracic surgery of claim 1, wherein, The step of identifying the chest-tube fluid transmission state specifically comprises: When the chest-tube coupling performance index shows that the chest expansion amplitude is normal but the water seal bottle liquid surface fluctuation amplitude is significantly lower than the standard value, it is determined that the pressure conduction is blocked; When the chest-tube coupling performance index is in a high partition interval, but the backtracking analysis shows that the absolute values of the chest expansion amplitude and the water seal bottle liquid surface fluctuation amplitude of the patient are significantly lower than the preset reference value, and both maintain a high linear correlation, it is determined that the driving source is in a low-amplitude response state; When a continuous bubble escape feature that is not in a respiratory cycle is detected in the fluid dynamics feature data, it is determined that the fluid is escaping in a non-periodic manner.

8. A post-operative drainage monitoring data processing method for thoracic surgery according to claim 7, characterized in that, The step of dynamically generating a hierarchical device response strategy specifically comprises: For the pressure conduction blocked state, a drainage tube squeezing nursing task order is generated, and a first level audible and visual alarm is sent to the nurses' station; For the driving source low-amplitude response state, if it is determined to be non-painful restriction in combination with the pain score, a strengthened respiratory training game instruction is generated, and an augmented reality feedback is used to encourage the patient to take deep breaths; For the non-periodic fluid escape state, the current rehabilitation training load parameter is automatically locked, and an emergency ward round request for a thoracic surgeon is generated.

9. The post-operative drainage monitoring data processing method for thoracic surgery of claim 1, wherein, The data acquisition module specifically acquires the real-time monitoring data in the following manner: A millimeter wave radar sensor deployed at the bedside is used to collect the micro-motion signals of the patient's chest wall, and the respiratory dynamics feature data is calculated and obtained; A machine vision module deployed on the drainage rack is used to capture the liquid surface dynamic video of the water seal bottle, and the liquid column height change sequence is extracted as the fluid dynamics feature data through image semantic segmentation technology.

10. A post-thoracic surgery drainage monitoring data processing system, characterized by, The system comprises: A data acquisition module configured to acquire real-time monitoring data of a postoperative patient in thoracic surgery, the real-time monitoring data comprising respiratory dynamics feature data of the patient's chest and fluid dynamics feature data of a closed thoracic drainage system; A central processing server configured to establish a time synchronization mapping relationship between the respiratory dynamics feature data and the fluid dynamics feature data, and perform feature fusion analysis to calculate a chest-tube coupling performance index; based on the chest-tube coupling performance index, logically determine the current postoperative recovery state of the patient, and identify the chest-tube fluid transmission state; dynamically generate a hierarchical device response strategy according to the identified chest-tube fluid transmission state. A bedside interactive terminal configured to display the hierarchical device response strategy and receive patient feedback; The hierarchical device response strategy comprises adjusting rehabilitation training intensity, triggering pipeline maintenance early warning, or generating a pain pump parameter adjustment prompt signal.

Images