Color-sensing-based bladder irrigation drop rate adaptive control system and method

Abstract

The application relates to the technical field of medical instrument control, and particularly discloses a bladder flushing drop speed self-adaptive control system and method based on color sensing, which comprises a color sensing module, a bubble interference identification and compensation module, a physiological state evaluation module and a drop speed self-adaptive control module. The color sensing module collects double-band light intensity signals sensitive to hemoglobin and sensitive to bubble scattering; the bubble interference identification and compensation module identifies and compensates bubble interference based on a reference band signal, and outputs corrected color characteristic values; the physiological state evaluation module evaluates the physiological state according to the characteristic values and outputs a target drop speed interval; and the drop speed self-adaptive control module adopts a conservative approaching strategy to calculate a set point, and adjusts the drop speed by PID control to drive an executing mechanism. The application can effectively inhibit bubble interference, realize smooth self-adaptive adjustment of the drop speed, and improve treatment safety.

Description

Technical Field

[0001] This invention belongs to the field of medical device control technology, specifically relating to an adaptive control system and method for bladder irrigation drip rate based on color sensing. Background Technology

[0002] In the field of medical monitoring and automated treatment, intelligent infusion systems are a key technological direction for achieving precise drug delivery and reducing the burden on medical staff. These systems integrate sensors and actuators to achieve real-time monitoring and closed-loop control of the infusion process, thereby improving the safety and effectiveness of treatment.

[0003] For bladder irrigation in post-urological surgery patients, precisely controlling the drip rate of the irrigation fluid to maintain stable intrabladder pressure, prevent blood clot formation, and reduce patient discomfort is an important specific application area in this field. Traditional bladder irrigation relies on medical staff manually adjusting the drip rate, which is not only labor-intensive but also makes it difficult to make timely and continuous adjustments based on changes in the color of the irrigation fluid (reflecting the presence of blood).

[0004] In existing technologies, some solutions have attempted to introduce color sensors to monitor changes in the color of the flushing fluid and automatically adjust the drip rate accordingly. However, these solutions face significant technical bottlenecks in practical applications: bubbles are inevitably generated when the flushing fluid flows through the pipeline. These bubbles cause strong reflection and scattering interference to light of specific wavelengths (especially the red light band used to determine blood concentration), resulting in severe distortion of the spectral data collected by the color sensor.

[0005] Misjudgments caused by air bubbles can directly trigger erroneous responses in the control system, resulting in abnormal and frequent fluctuations in the drip rate. These non-physiological and unstable flow rate changes can impact the patient's bladder mucosa, significantly increasing the risk of mucosal damage and secondary bleeding, thus greatly reducing the reliability of the automatic control system.

[0006] Therefore, how to effectively eliminate bubble interference in complex fluid environments, achieve stable and accurate perception of the color of the rinsing fluid, and establish reliable adaptive control logic accordingly has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0007] The purpose of this invention is to provide a color-sensing-based adaptive control system and method for bladder irrigation drip rate, in order to solve the technical contradiction in the prior art where air bubbles in the irrigation tubing cause severe optical interference to the color sensor, resulting in inaccurate drip rate control and frequent fluctuations, thereby increasing the risk of bladder mucosal damage to patients.

[0008] The technical solution of this invention is to provide a color-sensing-based adaptive control system for bladder irrigation drip rate. This system includes a color sensing module, a bubble interference recognition and compensation module, a physiological state assessment module, and a drip rate adaptive control module.

[0009] The color sensing module is used to acquire the spectral information of the bladder irrigation fluid. This module integrates at least two independent spectral acquisition units. The first spectral acquisition unit is configured to emit and receive incident light in a specific wavelength band, which is sensitive to the absorption characteristics of hemoglobin. The second spectral acquisition unit is configured to emit and receive incident light in a reference wavelength band, which is not sensitive to the absorption characteristics of hemoglobin but is sensitive to the light scattering effect caused by air bubbles in the fluid.

[0010] The optical paths of the two spectral acquisition units converge in the fluid detection area, ensuring that the spectral information they acquire comes from the same fluid sample at the same time point. The color sensing module synchronously outputs the acquired light intensity signal of a specific wavelength band with the light intensity signal of a reference wavelength band.

[0011] The bubble interference identification and compensation module, connected to the color sensing module, processes the received dual-band light intensity signal to identify and compensate for bubble interference, ultimately outputting a corrected color feature value reflecting the true blood concentration. This module includes a bubble interference identification submodule and a spectral signal compensation submodule. The bubble interference identification submodule calculates the dynamic change rate of the reference band light intensity signal in real time and compares it with a preset bubble interference judgment threshold.

[0012] When the dynamic change rate of the reference band light intensity signal exceeds the judgment threshold, significant bubble interference is determined to exist at the current detection time, and a bubble interference flag signal is generated. The spectral signal compensation submodule receives the specific band light intensity signal, the reference band light intensity signal, and the bubble interference flag signal. Its working logic is as follows: when the bubble interference flag signal is not activated, the original color feature value is directly calculated based on the specific band light intensity signal. When the bubble interference flag signal is activated, the compensation algorithm is started. This compensation algorithm constructs a bubble interference compensation coefficient based on the offset of the reference band light intensity signal relative to its reference value, and uses this coefficient to perform reverse correction on the specific band light intensity signal to eliminate the signal gain caused by bubble scattering, thereby calculating the corrected color feature value.

[0013] The physiological state assessment module, connected to the bubble interference recognition and compensation module, is used to assess the patient's real-time physiological bleeding status based on corrected color feature values ​​and generate a corresponding target drip rate range. This module incorporates a multi-state assessment model.

[0014] The model divides the continuous variation space of color feature values ​​into at least three discrete physiological state levels, including a stable state, a mild bleeding warning state, and a significant bleeding alarm state. Each physiological state level is associated with a preset target drip rate lower limit and a target drip rate upper limit, which together constitute the target drip rate interval for that state. The physiological state assessment module continuously receives the corrected color feature values ​​and inputs them into the multi-state assessment model for matching to determine the physiological state level at the current moment, and then outputs the corresponding target drip rate interval instruction.

[0015] The adaptive drip rate control module is connected to the physiological state assessment module and ultimately to the drip rate regulation actuator. This module generates a precise drip rate control signal based on the target drip rate range command. The adaptive drip rate control module includes a drip rate setpoint calculation submodule and a proportional-integral-derivative (PID) controller. The drip rate setpoint calculation submodule receives the target drip rate range command and obtains the current actual drip rate value fed back by an independent drip rate sensor. This submodule calculates the setpoint based on a conservative approach strategy.

[0016] Specifically, if the current actual drip rate is within the target drip rate range, the setpoint will be maintained at the current actual drip rate. If the current actual drip rate is lower than the lower limit of the target drip rate range, the setpoint will be gradually increased by a preset positive adjustment step size that is lower than the lower limit of the range until it enters the target range.

[0017] If the current actual drip rate is higher than the upper limit of the target drip rate range, the setpoint is gradually lowered by a preset negative adjustment step size that is greater than the upper limit of the range until it enters the target range. The proportional-integral-derivative controller receives the setpoint output by the drip rate setpoint calculation submodule and the actual drip rate value fed back by the drip rate sensor, calculates the deviation between the two, and generates a control signal based on the proportional, integral, and derivative parameters to drive the drip rate adjustment actuator to operate, so that the actual drip rate smoothly and stably tracks the change of the setpoint.

[0018] In one embodiment of the present invention, the method for calculating the dynamic change rate in the bubble interference identification submodule employs standard deviation calculation within a sliding time window. This submodule acquires a reference band light intensity signal sequence at a fixed sampling frequency and calculates the standard deviation of the signal within a time window of length N sampling points as a quantitative indicator of the dynamic change rate. When the standard deviation exceeds a preset bubble interference determination threshold, bubble interference is determined to exist.

[0019] Furthermore, the compensation algorithm in the spectral signal compensation submodule is implemented as follows: First, during system initialization or a stable period when no bubble interference is detected, the average value of the reference band light intensity signal is recorded as a baseline value. When the bubble interference flag signal is activated, the current reference band light intensity signal value is acquired in real time. The ratio of the current signal value to the baseline value is calculated to obtain the original interference coefficient.

[0020] Subsequently, a saturation function is applied to the original interference coefficient, limiting it to between a preset maximum and minimum compensation coefficient, thus generating the final bubble interference compensation coefficient for correction. Finally, the currently acquired specific band light intensity signal is divided by this bubble interference compensation coefficient to obtain the corrected specific band light intensity signal, and color feature values ​​are calculated accordingly.

[0021] In one embodiment of the present invention, the multi-state assessment model in the physiological state assessment module employs threshold comparison logic with hysteresis to prevent frequent state switching near the boundaries. This model sets two thresholds for the boundaries of each physiological state level: an entry threshold and an exit threshold. For example, to transition from a stable state to a mild bleeding warning state, a higher color feature value is set as the entry threshold. To return from the mild bleeding warning state to a stable state, the color feature value needs to decrease to a lower exit threshold. The entry threshold being higher than the exit threshold creates a hysteresis interval, effectively filtering out fluctuations in the assessment results caused by minor signal fluctuations.

[0022] Furthermore, the conservative approach strategy in the drip rate setpoint calculation submodule ensures that the absolute values ​​of both the positive and negative adjustment step sizes are no greater than 20% of the target drip rate range. This design guarantees that even if the drip rate needs to be adjusted, the change is gradual and gentle, avoiding abrupt changes in drip rate that may be caused by switching target ranges, thereby ensuring the stability of intrabladder pressure changes.

[0023] In one embodiment of the present invention, the system further includes a data recording and learning module. This module is connected to the bubble interference identification and compensation module and the physiological state assessment module, and is used to continuously record the historical frequency of bubble interference events, the corrected color feature value sequence, and the physiological state assessment results. The data recording and learning module periodically analyzes this historical data and dynamically optimizes the bubble interference judgment threshold and the state boundary threshold in the physiological state assessment model accordingly, enabling the system to adapt to individual differences among different patients and the slow physiological changes during long-term treatment.

[0024] This invention also provides a color-sensing-based adaptive control method for bladder irrigation drip rate, applied to the aforementioned system. The method includes the following steps: Step 1: Synchronously acquire the specific wavelength light intensity signal and the reference wavelength light intensity signal of the bladder irrigation outflow through the color sensing module.

[0025] Step 2: Process the dual-band signal using the bubble interference identification and compensation module. First, bubble interference is identified based on the dynamic change rate of the reference band light intensity signal. Then, upon identification of interference, a compensation coefficient is constructed using the reference band signal to correct the light intensity signal in the specific band, and the anti-interference color characteristic value is calculated.

[0026] Step 3: Through the physiological state assessment module, the color feature values ​​obtained in Step 2 are mapped to the preset physiological state level, and the corresponding target drip rate range instruction is output.

[0027] Step 4: Control is executed through the drip rate adaptive control module. First, based on the target drip rate range command and the current actual drip rate, a conservative approach strategy is used to calculate the drip rate setpoint. Then, a proportional-integral-derivative control algorithm is used to drive the actuator to adjust the drip rate, ensuring that the actual drip rate smoothly tracks the setpoint.

[0028] Step 5: During operation, the system continuously records relevant data and periodically optimizes internal parameter thresholds to complete the adaptive learning process.

[0029] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention fundamentally solves the core interference problem of bubbles in color sensing by deploying a dual-spectrum acquisition unit and an innovative bubble interference recognition and compensation algorithm. Utilizing a reference wavelength sensitive to bubble scattering as a "reconnaissance signal," interference events can be reliably detected. Upon detection of interference, instead of simply discarding the data or triggering an alarm, a compensation coefficient based on the reference signal is constructed to perform real-time mathematical correction on the interfered blood concentration signal. This allows usable color information reflecting the true physiological state to be extracted even in the presence of bubbles. This ensures the continuity and accuracy of the control system's sensing input, laying a solid foundation for stable control.

[0030] 2. This invention introduces a target drip rate range control strategy based on physiological state levels, combined with a conservative approach to setpoint calculation method. This design transforms the direct and potentially drastically changing color-drip rate mapping into a gradual, range-oriented adjustment process. The system prioritizes maintaining the current drip rate within a reasonable range, adjusting slowly only when necessary in limited steps. This mechanism effectively avoids frequent drip rate oscillations caused by minor fluctuations in sensor signals or changes in physiological state boundaries, ensuring extreme stability of the infusion flow rate, minimizing the risk of fluid impact on the bladder mucosa, and improving the safety and comfort of treatment.

[0031] 3. This invention enhances the robustness and individual adaptability of the system by integrating a state assessment with hysteresis and a parameter self-learning module. The hysteresis comparison logic eliminates the jitter in state assessment near the threshold, making control decisions more stable. Data recording and learning functions enable the system to adaptively fine-tune the disturbance identification threshold and physiological state threshold based on long-term operating data, thereby better matching the individual physiological characteristics and treatment stages of different patients. This achieves an upgrade from static control to dynamic adaptive control, improving the system's universality and reliability in different clinical scenarios. Attached Figure Description

[0032] Figure 1 This is a schematic diagram of the overall technical solution architecture of the color-sensing-based adaptive control system for bladder irrigation drip rate proposed in this invention. Figure 2 This is a schematic diagram of the core principle framework of the bubble interference identification and compensation module in this invention; Figure 3 This is a logical flowchart of the mapping between physiological state assessment and target drip rate interval in this invention; Figure 4 This is a schematic diagram of the conservative approach control strategy of the drip rate adaptive control module in this invention; Figure 5 This is a schematic diagram of the data flow and adaptive learning loop between multiple modules in the system of this invention. Detailed Implementation

[0033] Example 1: This invention provides a color-sensing-based adaptive control system for bladder irrigation drip rate. Please refer to the appendix. Figures 1 to 5 The system physically constitutes a closed-loop control system. Its core function is to adjust the drip rate of bladder irrigation fluid in real time, accurately and smoothly in response to the color change caused by bleeding in the patient's bladder irrigation outflow. It also has a strong ability to resist optical interference from air bubbles.

[0034] The system achieves these goals through a series of precisely coordinated functional modules, including a color sensing module, a bubble interference recognition and compensation module, a physiological state assessment module, a drip rate adaptive control module, and a data recording and learning module. These modules are connected and exchange data through well-defined data interfaces and communication protocols, ensuring the continuity and real-time nature of information flow from perception, processing, decision-making to execution.

[0035] The color sensing module is the system's front-end sensor, and its core task is to collect the spectral information of the bladder irrigation fluid. Please refer to the appendix. Figure 1This module is integrated into the transparent detection chamber of the bladder irrigation outflow line, ensuring that the irrigation fluid can be effectively measured as it flows through the detection area. The color sensing module integrates at least two spectral acquisition units that are completely independent in terms of optical design and spectral response.

[0036] The first spectral acquisition unit is configured to emit and receive incident light in a specific wavelength band. This specific wavelength band is carefully selected, with its center wavelength typically located in the spectral region where hemoglobin exhibits significant absorption characteristics, such as the green light band from 540 nm to 580 nm. Therefore, the light intensity signal in this band is negatively correlated with the concentration of free hemoglobin in the effluent; that is, the higher the hemoglobin concentration, the more light is absorbed, and the weaker the reflected or transmitted light signal received by the detector.

[0037] The second spectral acquisition unit is configured to emit and receive incident light in a reference wavelength band. This reference wavelength band is selected based on the principle that it is extremely insensitive to the absorption characteristics of hemoglobin, but its wavelength falls within the sensitive range where tiny bubbles in the fluid produce a strong Mie scattering effect on light, such as the near-infrared band above 700 nanometers. When bubbles are present in the pipeline, the light in this band will be significantly scattered, causing drastic fluctuations in the light intensity signal received by the detector.

[0038] The optical paths of the two spectral acquisition units are precisely calibrated so that their optical paths intersect precisely at the center point of the fluid detection area. This means that at any sampling moment, the spectral information acquired by the first spectral acquisition unit and the second spectral acquisition unit comes from the exact same fluid micro-element sample, which fundamentally ensures the strict synchronization of the dual-band signals in time and space.

[0039] The color sensing module internally includes analog signal conditioning circuitry and an analog-to-digital converter, capable of converting the analog current signal generated by the photodetector into a high-resolution digital signal. This module synchronously samples and outputs two digital quantities at a fixed sampling frequency, e.g., 10 times per second: a light intensity signal in a specific wavelength band and a light intensity signal in a reference wavelength band. These two signals are transmitted in real-time to the subsequent bubble interference identification and compensation module via a standard digital communication interface, such as a serial peripheral interface or a controller area network bus.

[0040] The bubble interference recognition and compensation module is the core of the system's signal processing. It is responsible for de-interference processing of the raw dual-band light intensity signal from the color sensing module, outputting color characteristic values ​​that accurately reflect blood concentration. Please refer to the appendix. Figure 2This module logically comprises two sequentially executed sub-modules: a bubble interference identification sub-module and a spectral signal compensation sub-module. The bubble interference identification sub-module continuously receives a sequence of reference band light intensity signals from the color sensing module. Its core task is to quantify the fluctuation level of the reference band signal and determine whether there are significant bubbles sufficient to interfere with color judgment at the current moment. To achieve this goal, this sub-module employs a signal dynamic change rate calculation method based on a sliding time window.

[0041] Specifically, the system acquires the reference band light intensity signal at a fixed sampling frequency and maintains a first-in, first-out (FIFO) data buffer. This buffer stores the signal values ​​of the most recent N sampling points, where N is typically 20, corresponding to a 2-second time window. The bubble interference identification submodule periodically calculates the standard deviation of all signal values ​​within this sliding window, using this standard deviation as a quantitative indicator of the dynamic rate of change. This indicator physically reflects the fluctuation amplitude of the reference light intensity over a short period. The bubble interference identification submodule has a preset bubble interference judgment threshold, which is obtained through experimental calibration and represents the normal upper limit of reference signal fluctuation under stable flow conditions with no bubbles or only very small bubbles.

[0042] In each calculation cycle, the submodule compares the calculated standard deviation with the preset threshold. If the calculated standard deviation is less than or equal to the threshold, it is determined that there is no significant bubble interference at the current moment, and the bubble interference flag signal is set to logic 0 or invalid. If the calculated standard deviation is greater than the threshold, it is determined that there is significant bubble interference at the current moment, and the bubble interference flag signal is immediately set to logic 1 or valid. This flag signal, along with the specific band light intensity signal and the reference band light intensity signal at the current moment, is sent to the spectral signal compensation submodule.

[0043] The spectral signal compensation submodule receives three inputs: a specific wavelength band light intensity signal, a reference wavelength band light intensity signal, and a bubble interference flag signal. This submodule executes the core compensation algorithm, whose internal logic contains two branch paths, with the path executed determined by the state of the bubble interference flag signal. When the bubble interference flag signal is invalid, it indicates that the current optical environment is stable and there is no bubble interference.

[0044] The spectral signal compensation submodule directly calculates the original color feature value using the currently acquired light intensity signal of a specific wavelength band. The calculation of the color feature value can be a simple normalization of the light intensity signal of the specific wavelength band, such as dividing it by the background light intensity value of clear water measured during system initialization, or a more complex calibration formula that converts the light intensity into an approximate hemoglobin concentration. When the bubble interference flag signal is active, it indicates that bubble interference has been detected. At this time, the compensation algorithm must be activated to correct the distortion caused by bubbles to the specific wavelength band signal. The execution of the compensation algorithm follows a series of coherent steps.

[0045] First, the system requires a reference band light intensity signal as a benchmark for comparison. This benchmark is obtained during the system's power-on initialization phase by acquiring the reference band signal over a continuous period, such as 30 seconds, under stable, bubble-free flow conditions and calculating its arithmetic mean. In subsequent operation, if the system does not detect bubble interference for a longer period, such as 60 seconds, it will update this benchmark with the average value of the reference signal from the most recent stable period to accommodate possible slow drift of ambient light.

[0046] When bubble interference occurs, the algorithm acquires the current reference band light intensity signal value in real time. It calculates the ratio of the current signal value to the reference value to obtain the original interference coefficient. This ratio reflects the amplification factor of the reference light intensity relative to its normal level due to bubble scattering. Because the size and distribution of bubbles are random, this original interference coefficient may fluctuate within a wide range. To prevent overcompensation or undercompensation, the algorithm processes a preset saturation function. The saturation function defines the upper and lower limits of the compensation coefficient, i.e., the maximum and minimum compensation coefficients.

[0047] For example, the maximum compensation coefficient can be set to 1.5, and the minimum compensation coefficient can be set to 0.8. The algorithm limits the original interference coefficient to this range, generating the final bubble interference compensation coefficient. Finally, a crucial correction operation is performed: the specific wavelength light intensity signal acquired at the current moment is divided by this bubble interference compensation coefficient. The physical meaning is that, assuming the bubble causes approximately proportional light scattering enhancement to both wavelengths, by dividing by the enhancement factor of the reference wavelength, the scattering gain of the bubble on the specific wavelength signal can be reversed, thus restoring the specific wavelength light intensity signal that would exist without the bubble.

[0048] The color feature value calculated using the corrected light intensity signal of a specific wavelength band is the anti-interference color feature value. This algorithm can be expressed mathematically as follows: Let the light intensity signal of the specific wavelength band be... The reference band light intensity signal is The reference band baseline value is The saturation function is Then the bubble interference compensation coefficient Corrected specific band light intensity signal Subsequently, based on Calculate the color feature value C. Through this series of operations, the bubble interference recognition and compensation module can still output a relatively accurate and usable color feature value even when bubbles are present, greatly improving the robustness of the system's perception.

[0049] The physiological state assessment module is connected after the bubble interference recognition and compensation module. Its function is to map continuously changing color feature values ​​to discrete, clinically meaningful physiological state levels, and generate control commands accordingly. Please refer to the appendix. Figure 3 The core of this module is the multi-state assessment model. This model divides the entire possible range of color feature value C into at least three non-overlapping but continuously connected intervals, each interval corresponding to a specific physiological state level.

[0050] These three levels are typically defined as: stable state, mild bleeding warning state, and significant bleeding alarm state. A stable state corresponds to a very low color characteristic value, indicating that the outflow is clear or contains only trace amounts of blood, and the patient has no active bleeding or the bleeding has stopped. A mild bleeding warning state corresponds to a moderate color characteristic value, indicating slight active bleeding that requires attention and may necessitate appropriately increasing the flushing rate to prevent blood clot formation. A significant bleeding alarm state corresponds to a very high color characteristic value, indicating significant active bleeding that requires immediate and increased flushing to maintain tubing patency and alerting healthcare personnel.

[0051] Each physiological state level is not associated with a single drip rate value, but rather with a target drip rate range, which is defined by the target drip rate lower limit V. min And the target drip rate upper limit V max Common definitions are used. For example, the target drip rate range for a stable state might be 20 to 30 drops per minute; the range for a mild bleeding warning state might be 35 to 50 drops per minute; and the range for a significant bleeding alarm state might be 60 to 80 drops per minute. These range values ​​are preset based on clinical experience and stored in the module's non-volatile memory. The real-time assessment logic of the physiological state assessment module is not a simple instantaneous threshold comparison, but rather employs threshold comparison logic with hysteresis to prevent frequent switching of assessment results due to minor signal fluctuations near state boundaries.

[0052] The model sets two sets of thresholds for the boundary between every two adjacent states: a higher entry threshold and a lower exit threshold. For example, for the boundary between a stable state and a mild bleeding warning state, a color feature value entry threshold is set. When the real-time color feature value C drops below Rise and exceed At this point, the state transitions from a stable state to a mild bleeding warning state. However, to return from the mild bleeding warning state to the stable state, the color feature value C needs to decrease and fall below a lower exit threshold. ,in . and The difference between them forms a hysteresis interval. As long as C falls within this hysteresis interval, the physiological state assessment module will maintain the previous state determination and will not switch.

[0053] This mechanism effectively filters noise, ensuring the stability of the state assessment results. The physiological state assessment module continuously receives corrected color feature values ​​C from the bubble interference identification and compensation module, compares them with a preset hysteresis threshold in the multi-state assessment model, and determines the physiological state level at the current moment. Once the state is determined, the module immediately outputs the corresponding target drip rate range instruction, which includes the target drip rate lower limit V for the current state. min And the target drip rate upper limit V max .

[0054] The drip rate adaptive control module is the system's execution decision-making center. It is responsible for generating the control signal that ultimately drives the drip rate adjustment actuator based on the target instructions given by the physiological state assessment module and the current actual situation. Please refer to the appendix. Figure 1 and attached Figure 4 This module connects to the physiological state assessment module and ultimately to the drip rate regulation actuator via a power drive circuit. This actuator is typically a linear pinch valve or roller pump driven by a stepper motor or servo motor. The drip rate adaptive control module contains two key sub-modules: a drip rate setpoint calculation sub-module and a proportional-integral-derivative (PID) controller. The drip rate setpoint calculation sub-module receives two inputs: one is a target drip rate range command from the physiological state assessment module, including V... min and V max The other path is the current actual drip rate feedback value from an independent drip rate sensor. This drip rate sensor typically uses an infrared pair to detect the droplets falling inside the dripper, providing accurate drops per minute. The core task of the drip rate setpoint calculation submodule is to calculate the desired drip rate setpoint V based on a conservative approximation strategy. setpoint The detailed rules of the conservative approach strategy are as follows: First, determine the current actual drip rate value. The relationship with the target drip rate range. If Greater than or equal to V min And less than or equal to V max If the current actual drip rate is already within the target range, then no adjustment is needed. To maintain stability to the greatest extent possible, the setpoint V is adjusted. setpoint Set to equal to the current .if Less than V min If the current dripping rate is too slow, below the lower limit of the target range, then the rate needs to be gradually increased. At this point, the setpoint V... setpoint The calculation method is as follows: V setpoint = + .in It is a preset positive adjustment step size, the absolute value of which is strictly limited and required not to exceed the target drip rate range V. max - V min 20%. For example, if the interval is 30 to 50 drops per minute, and the span is 20, then The maximum rate is no more than 4 drops per minute. This method of incremental acceleration ensures a smooth process. If, after one adjustment, the new... Still below V min Then, in the next control cycle, it will continue to use... Increase the step size by the set point until the actual drip rate enters the target range. If Greater than V max If the current dripping rate is too fast, exceeding the upper limit of the target range, then it needs to be gradually reduced. At this point, the set point... The calculation method is as follows: = - .in, The preset negative adjustment step size is also limited to no more than 20% of the target drip rate range. A gradual decrease in drip rate is achieved through small step reductions. The essence of this strategy lies in the fact that it does not aim to instantly reach the midpoint of the target range or a fixed value, but rather to guide and maintain the actual drip rate within a reasonable range. Furthermore, the adjustment process is extremely gentle and gradual, completely avoiding any sudden jumps in drip rate that might be caused by changes in physiological state levels. This is crucial for maintaining stable intrabladder pressure.

[0055] The proportional-integral-derivative (PID) controller is a classic control algorithm implementation unit. It receives the setpoint from the drip rate setpoint calculation submodule. and the actual value from the drip rate sensor Calculate the deviation between the two. = - Based on this deviation, the controller generates a control signal through a comprehensive calculation involving proportional, integral, and derivative operations. .

[0056] The proportional term provides a fast response proportional to the deviation; the integral term eliminates steady-state error and ensures long-term accuracy; the derivative term predicts trends, suppresses overshoot, and improves system stability. Control signal After digital-to-analog conversion and power amplification, the motor driving the drip rate regulating actuator, such as a pinch valve, changes the degree of compression on the pipeline, thereby linearly adjusting the flow resistance and ultimately changing the drip rate in the dripper. The entire control loop operates at a fixed frequency, such as 5 times per second, allowing the actual drip rate to smoothly, stably, and without overshoot track the slow changes in the setpoint.

[0057] The data recording and learning module is the intelligent unit that enables the system to achieve long-term self-adaptation and individualized optimization. Please refer to the appendix. Figure 5 This module connects to the bubble interference identification and compensation module and the physiological state assessment module, continuously collecting and recording key operational data from these modules. The recorded data primarily includes: timestamp sequences, dynamic rate of change index sequences of reference band light intensity signals, bubble interference indicator signal sequences, corrected color feature value sequences, physiological state assessment result sequences, and actual drip rate sequences. This data is stored in local non-volatile memory in the form of a circular buffer or a timestamped database.

[0058] The data recording and learning module does not passively record data; it periodically initiates a data analysis and parameter optimization routine, for example, every 24 hours or after each bladder irrigation treatment cycle. This routine performs two core self-learning tasks. The first task is to optimize the bubble interference detection threshold. The module will statistically analyze the distribution characteristics of the dynamic change rate of the reference signal when these events occur, based on all events marked as bubble interference in past cycles.

[0059] Simultaneously, the fluctuation level of the reference signal is analyzed during periods not marked as interference. Through machine learning algorithms such as cluster analysis or statistical analysis, the bubble interference detection threshold can be dynamically adjusted to better adapt to current tubing characteristics, flushing fluid properties, and potential ambient light background noise, reducing the false positive rate while maintaining sensitivity. The second task is to optimize the state boundary thresholds in the physiological state assessment model. The module analyzes the historical distribution of color feature values ​​and, in conjunction with events that healthcare workers may manually intervene in or annotate, such as recording moments of increased bleeding, to assess the reasonableness of the currently preset entry and exit thresholds.

[0060] For example, if the system detects that color feature values ​​frequently fluctuate around a certain threshold, causing state instability, but no corresponding physiological changes are observed clinically, the learning algorithm may appropriately increase the hysteresis width. Alternatively, for a specific patient whose baseline body fluid color may differ from that of the average person, the learning algorithm can slowly shift the entire threshold set to make the state assessment more consistent with the patient's individualized physiological baseline.

[0061] The optimized new parameter thresholds are securely verified and updated in the configuration storage of the bubble interference recognition submodule and the physiological state assessment module. This process forms a closed-loop learning cycle from data to knowledge to parameter adjustment, enabling the system to adapt more and more accurately to the specific patients and treatment environments it serves over time, evolving from a static, general system into a dynamic, personalized system.

[0062] The entire system operates through a collaborative workflow of the aforementioned modules. The color sensing module serves as the starting point, continuously providing dual-band optical signals. The bubble interference identification and compensation module processes these signals in real time, eliminating bubble interference and outputting reliable color feature values. The physiological state assessment module transforms these feature values ​​into clinically meaningful physiological states and target drip rate ranges. The drip rate adaptive control module then calculates the setpoint in a conservative and smooth manner based on this range and the current actual drip rate, and drives the actuator through precise proportional-integral-derivative control to ultimately achieve accurate and stable drip rate adjustment.

[0063] Meanwhile, the data recording and learning module works silently in the background, collecting data from the entire process and periodically optimizing key parameters to continuously improve system performance. All data exchanges between modules are completed through strictly defined communication protocols, including data frame formats, checksums, time synchronization information, and error retransmission mechanisms, ensuring the integrity and reliability of data transmission in the complex electromagnetic environment of medical electrical equipment. The system also has a comprehensive anomaly handling mechanism. For example, when the color sensor module loses its signal, the drip rate sensor malfunctions, or the actuator stalls, it triggers different levels of local alarms and automatically switches to a safe mode, such as maintaining the current drip rate or switching to the lowest safe drip rate. Simultaneously, it sends detailed fault codes and status information to the nurse station monitoring system, ensuring absolute patient safety.

[0064] Example 2: Based on the complete system architecture described in Example 1, this example provides an alternative implementation scheme for specific clinical scenarios or with different hardware configurations. The core feature of this example is the adaptive deepening and expansion of the spectral acquisition unit configuration of the color sensing module and the structure of the physiological state assessment model, aiming to cope with more complex interference from body fluid components or to provide more refined physiological state resolution capabilities.

[0065] In this embodiment, while retaining the basic architecture of the dual-spectrum acquisition unit described in Embodiment 1, the color sensing module upgrades the spectral resolution of the first spectral acquisition unit, namely the specific band acquisition unit. This unit no longer uses a single-band broadband light source and filter combination, but instead employs a miniaturized spectrometer assembly or a tunable narrowband light source. It can acquire the light intensity signals of bladder irrigation fluid at multiple discrete wavelength points. These wavelength points are optimized to have different absorption characteristics for oxyhemoglobin, deoxyhemoglobin, and other potentially interfering substances such as bilirubin and drug metabolites.

[0066] For example, the system can collect light intensity at multiple characteristic wavelengths, such as 540 nm, 577 nm, and 660 nm. Multi-wavelength signals provide richer spectral information, enabling the system to run more advanced algorithms to distinguish blood color from other causes of liquid discoloration. Correspondingly, the compensation algorithm in the bubble interference recognition and compensation module also needs to be expanded. Its basic logic still utilizes a reference band, typically near-infrared, that is sensitive to bubble scattering but unaffected by blood component absorption as the benchmark for interference detection and compensation. However, when bubble interference is detected, the compensation operation will simultaneously apply to signals from multiple specific wavelength channels.

[0067] In practice, the scattering enhancement effect of bubbles on various wavelengths can be considered to have wavelength-dependent coefficients, which can be obtained through pre-calibration or estimation based on the Mie scattering model. The compensation algorithm calculates the compensation coefficients corresponding to each wavelength based on the change of the current reference band signal relative to the baseline value, and corrects each specific wavelength signal. Subsequently, using the corrected multi-wavelength signals, a comprehensive color feature value is calculated through a pre-trained multiple linear regression model or lookup table.

Claims

1. A color-sensing-based adaptive control system for bladder irrigation drip rate, characterized in that, include: The color sensing module is used to collect spectral information of the bladder irrigation fluid. The bubble interference identification and compensation module is connected to the color sensing module and is used to process the received dual-band light intensity signal to identify and compensate for bubble interference. The physiological state assessment module is connected to the bubble interference recognition and compensation module, which is used to assess the patient's real-time physiological bleeding status based on the corrected color feature values ​​and generate the corresponding target drip rate range. The drip rate adaptive control module is connected to the physiological state assessment module and finally to the drip rate adjustment actuator, which is used to generate a precise drip rate control signal according to the target drip rate range instruction; The bubble interference identification and compensation module includes a bubble interference identification submodule and a spectral signal compensation submodule. The bubble interference identification submodule calculates the dynamic change rate of the reference band light intensity signal in real time and compares it with a preset bubble interference judgment threshold. When the dynamic change rate of the reference band light intensity signal exceeds the judgment threshold, it determines that there is significant bubble interference at the current detection time and generates a bubble interference flag signal. The spectral signal compensation submodule receives the specific band light intensity signal, the reference band light intensity signal, and the bubble interference flag signal. Its working logic is that when the bubble interference flag signal is not activated, it directly calculates the original color feature value based on the specific band light intensity signal. When the bubble interference flag signal is activated, it starts the compensation algorithm. The compensation algorithm constructs a bubble interference compensation coefficient based on the offset of the reference band light intensity signal relative to its reference value, and uses the coefficient to perform reverse correction on the specific band light intensity signal to eliminate the signal gain caused by bubble scattering, thereby calculating the corrected color feature value. The physiological state assessment module has a built-in multi-state assessment model, which divides the continuous change space of color feature values ​​into at least three discrete physiological state levels, including stable state, mild bleeding warning state, and significant bleeding alarm state. Each physiological state level is associated with a preset target drip rate lower limit and target drip rate upper limit to jointly constitute the target drip rate interval for that state. The physiological state assessment module continuously receives the corrected color feature values ​​and inputs them into the multi-state assessment model for matching to determine the physiological state level at the current moment, and then outputs the corresponding target drip rate interval instruction. The adaptive drip rate control module includes a drip rate setpoint calculation submodule and a proportional-integral-derivative (PID) controller. The drip rate setpoint calculation submodule receives the target drip rate range instruction and obtains the current actual drip rate value fed back by an independent drip rate sensor. This submodule calculates the setpoint according to a conservative approach strategy. Specifically, if the current actual drip rate value is within the target drip rate range, the setpoint is maintained at the current actual drip rate value. If the current actual drip rate value is lower than the lower limit of the target drip rate range, the setpoint is gradually increased by a preset positive adjustment step size less than the lower limit of the range until it enters the target range. If the current actual drip rate value is higher than the upper limit of the target drip rate range, the setpoint is gradually decreased by a preset negative adjustment step size greater than the upper limit of the range until it enters the target range. The PID controller receives the setpoint output by the drip rate setpoint calculation submodule and the actual drip rate value fed back by the drip rate sensor, calculates the deviation between the two, and generates a control signal based on the proportional, integral, and derivative parameters to drive the drip rate adjustment actuator to make the actual drip rate smoothly and stably track the change of the setpoint.

2. The color-sensing-based adaptive bladder irrigation drip rate control system according to claim 1, characterized in that, The color sensing module integrates at least two independent spectral acquisition units. The first spectral acquisition unit is configured to emit and receive incident light of a specific wavelength that is sensitive to the absorption characteristics of hemoglobin. The second spectral acquisition unit is configured to emit and receive incident light of a reference wavelength that is not sensitive to the absorption characteristics of hemoglobin but is sensitive to the light scattering effect caused by bubbles in the fluid. The optical paths of the two spectral acquisition units intersect in the fluid detection area to ensure that the spectral information they acquire comes from the same fluid sample at the same time point. The color sensing module outputs the acquired specific wavelength light intensity signal and the reference wavelength light intensity signal synchronously.

3. The color-sensing-based adaptive bladder irrigation drip rate control system according to claim 2, characterized in that, The method for calculating the dynamic change rate in the bubble interference identification submodule is to use the standard deviation within a sliding time window. This submodule acquires a reference band light intensity signal sequence at a fixed sampling frequency and calculates the standard deviation of the signal within a time window of length N sampling points as a quantitative indicator of the dynamic change rate. When the standard deviation exceeds the preset bubble interference judgment threshold, it is determined that bubble interference exists.

4. The color-sensing-based adaptive bladder irrigation drip rate control system according to claim 3, characterized in that, The compensation algorithm in the spectral signal compensation submodule is specifically implemented as follows: during system initialization or a stable period when no bubble interference is detected, the average value of the reference band light intensity signal is recorded as the baseline value. When the bubble interference flag signal is activated, the current reference band light intensity signal value is acquired in real time; the original interference coefficient is obtained by calculating the ratio of the current signal value to the reference value. The original interference coefficient is processed by a saturation function to limit it between the preset maximum and minimum compensation coefficients to generate the final bubble interference compensation coefficient. The corrected specific band light intensity signal is obtained by dividing the currently acquired specific band light intensity signal by the bubble interference compensation coefficient, and the color feature value is calculated accordingly.

5. The color-sensing-based adaptive bladder irrigation drip rate control system according to claim 4, characterized in that, The multi-state assessment model in the physiological state assessment module adopts threshold comparison logic with hysteresis. This model sets two thresholds for the boundary of each physiological state level, namely the entry threshold and the exit threshold. The entry threshold is higher than the exit threshold to form a hysteresis interval, so as to prevent the assessment results from frequently switching due to small signal fluctuations near the boundary.

6. The color-sensing-based adaptive bladder irrigation drip rate control system according to claim 5, characterized in that, The conservative approach strategy in the drip rate setpoint calculation submodule has an absolute value of no more than 20% of the target drip rate range for both the positive and negative adjustment step sizes.

Images