A continuous negative pressure wound drainage control system for diabetic foot wounds

Abstract

The present application relates to the technical field of negative pressure control, in particular to a continuous negative pressure drainage control system for diabetic foot wound. The system obtains weight data and pressure data of the drainage process, divides time windows and calculates the average of drainage volume, pressure consumption and unit drainage pressure consumption value; extracts a preset number of unit drainage pressure consumption values to form a sequence, converts it into an undirected adjacency matrix using the visibility graph algorithm, and obtains the latest connection weight of the node to reflect the real resistance by combining the weight transfer algorithm; the time window with weight continuously greater than the preset threshold is taken as the shutdown time window to prevent instantaneous blockage misjudgment; the historical time window is traced back, the reciprocal of the weight is taken as the penalty factor for weighted average, the safe maximum negative pressure limit value without pressure error is backstepped, and the total drainage volume is accumulated; finally, the next cycle is controlled based on this. The present application realizes dynamic adaptive matching of the amount of drug and suction negative pressure, prevents drug liquid overflow and protects granulation tissue.

Description

Technical Field

[0001] This invention relates to the field of negative pressure control technology for drug delivery, and specifically to a continuous drug delivery negative pressure drainage control system for diabetic foot wounds. Background Technology

[0002] In the clinical treatment of diabetic foot ulcers, wound drainage typically employs a continuous cycle of medication infusion and negative pressure aspiration to clean the wound and promote healing. As treatment progresses, new granulation tissue gradually grows and fills the original physical defects, causing the effective cavity volume capable of holding medication to continuously shrink in a dynamic physiological evolution. Existing drainage devices generally operate with a pre-set fixed medication volume and a constant upper limit of negative pressure. This fixed medication volume cannot adapt to the shrinking wound volume; excess medication easily overflows the dressing and soaks the surrounding delicate skin, posing a risk of secondary infection. Furthermore, towards the end of the drainage process, when the free medication in the wound has been drained, if the device continues to maintain the predetermined high negative pressure, it will force the sponge dressing to undergo severe contraction and deformation, applying excessive mechanical compression and traction to the newly formed capillaries at the base of the wound, thus causing ischemic injury.

[0003] To avoid excessive suction at the end of the drainage process, existing drainage feedback control systems attempt to monitor pipeline pressure and use the absolute threshold or first derivative (rate of change) of the pressure to determine whether the medication has been drained, thereby controlling the start and stop of the motor. However, in reality, drainage from diabetic foot wounds typically exhibits a complex multiphase flow state, containing a large amount of free gas and solid necrotic tissue debris. When solid debris moves and tumbles inside the tubing, it can easily cause momentary physical blockage, generating a severe and brief pressure spike. Conventional derivative or threshold control algorithms struggle to isolate this physical noise, easily misinterpreting the momentary resistance spike caused by solid debris as a resistance surge indicating that the free fluid has been completely drained, thus issuing an incorrect shutdown command. This leads to premature and abnormal termination of drainage and incomplete wound cleaning. Furthermore, existing equipment lacks a mechanism for retrospective analysis to historical pressure-free safe states, directly relying on the absolute pressure value at or near the shutdown moment as the benchmark for subsequent control. In reality, at the moment the shutdown threshold is triggered, the sponge dressing has usually already undergone significant shrinkage and deformation. The pressure data at that moment essentially already includes the error of excessive compression of granulation tissue. If this is directly used as the upper limit of the vacuum pump drive for the next cycle, the newly formed granulation tissue will inevitably suffer repeated mechanical compression. Summary of the Invention

[0004] To address the technical problem of existing drainage feedback control systems failing to accurately distinguish between transient blockage and actual fluid emptying in multiphase flow environments, and also failing to eliminate deformation errors caused by dressing compression during shutdown, leading to inaccurate drug delivery and suction negative pressure, this invention aims to provide a continuous drug delivery negative pressure drainage control system for diabetic foot wounds. The specific technical solution adopted is as follows: This invention provides a continuous drug delivery negative pressure drainage control system for diabetic foot wounds, the system comprising: The data acquisition module is used to acquire the weight data of the drained fluid and the pressure data of the wound at each moment during the continuous negative pressure drainage process of diabetic foot wound; the drainage cycle of the drainage process is divided into time windows according to the preset duration, and based on the weight data and pressure data in each time window, the drainage volume, average pressure consumption and unit drainage pressure consumption value of each time window are acquired; The connection weight acquisition module is used to extract a continuous preset number of unit drainage pressure consumption values, including the current time window, to form an analysis sequence. The visibility graph algorithm is used to map each value in the analysis sequence to a node according to the time sequence and generate an undirected adjacency matrix based on geometric anti-occlusion constraints. Combined with the weight transfer algorithm, the latest connection weight of the node corresponding to the time window is obtained. The current cycle analysis module is used to define the current time window where the latest connection weights for a specified number of consecutive time windows are all greater than a preset judgment threshold as the shutdown time window; it traces back from the shutdown time window to preset historical time windows, uses the reciprocal of the latest connection weights of each preset historical time window as a penalty factor, and performs a weighted average of the pressure consumption average of each preset historical time window to obtain the maximum negative pressure limit value of the current cycle; and it obtains the total drainage volume of the current cycle based on all drainage volumes from the start of the current cycle to the shutdown time window. The drainage control module is used to control the drainage for the next cycle based on the maximum negative pressure limit and total drainage volume of the current cycle.

[0005] Furthermore, the methods for obtaining the drainage volume, average pressure consumption, and unit drainage pressure consumption are as follows: For any time window, the weight data within that time window is differentiated and filtered to obtain the instantaneous flow rate. The instantaneous flow rate is then integrated over time and divided by a preset density to obtain the drainage volume for that time window. If the drainage volume is less than 0, it is forcibly assigned the value of 0. Integrate the pressure data within the time window over time to obtain the cumulative pressure value for that time window. Divide the accumulated pressure value by the duration of the time window to obtain the average pressure consumption of the time window; The ratio of the accumulated pressure value to the sum of the drainage volume and a preset minimum positive number is used as the unit drainage pressure consumption value for that time window.

[0006] Furthermore, the method for obtaining the analysis sequence is as follows: If the number of time windows up to the current execution reaches a preset number within the current cycle, then extract a preset number of consecutive units of drainage pressure consumption values, including the current time window and its preceding time windows, to form an analysis sequence. If the number of time windows up to the current execution point has not reached the preset number, the control system maintains the low-pressure collection state. When the drainage volume corresponding to the consecutive reference number of time windows is less than or equal to the preset minimum positive number, the system forcibly exits the low-pressure collection state and establishes the current time window as the shutdown time window. The system extracts the unit drainage pressure consumption value of all time windows accumulated in the current cycle to form an analysis sequence.

[0007] Furthermore, the method for obtaining the geometric anti-occlusion constraint conditions is as follows: A two-dimensional coordinate system is constructed using the time series index as the horizontal axis and the unit drainage pressure consumption value as the vertical axis, and the node corresponding to each element in the analysis sequence in the two-dimensional coordinate system is determined. For the first and second nodes that are not temporally adjacent, and any intermediate node located between the time axes of the first and second nodes, calculate the longitudinal difference between the second node and the first node as the total analysis difference; calculate the position ratio of the intermediate node in the total time span between the first and second nodes as the analysis ratio of the intermediate node; wherein, the first node is located before the second node in time sequence. Multiply the total analytical difference by the analytical ratio, and add it to the unit drainage pressure consumption value corresponding to the first node to obtain the theoretical longitudinal height of the intermediate node; If the unit drainage pressure consumption value of all intermediate nodes located between the time axes of the first and second nodes is less than their theoretical longitudinal height, then the first and second nodes satisfy the geometric anti-shading constraint conditions.

[0008] Furthermore, the method for obtaining the undirected adjacency matrix is ​​as follows: For any two nodes, if the two nodes are not temporally adjacent, and the two nodes satisfy the geometric anti-occlusion constraint, then a connecting edge is established between the two nodes; if the two nodes do not satisfy the geometric anti-occlusion constraint, then a connecting edge is not established between the two nodes. When the two nodes are temporally adjacent, a connecting edge is directly established between the two nodes. Using nodes as rows and columns of a matrix, assign the value 1 to the corresponding element of two nodes with connected edges and the value 0 to the corresponding element of two nodes without connected edges, thus generating an undirected adjacency matrix.

[0009] Furthermore, the method for obtaining the latest connection weight is as follows: The number of neighboring nodes and the total number of connected edges between neighboring nodes are determined based on the undirected adjacency matrix. Each node is assigned an initial connection weight. For any node, according to the preset damping coefficient, the connection weight of each of the node's neighboring nodes in the previous round is divided by the total number of its connected edges to obtain the allocation weight. The sum of the allocation weights received by the node from all its neighboring nodes is added to the basic compensation value determined based on the preset damping coefficient and the total number of nodes, and the connection weight of the node is iteratively updated. After multiple iterations until convergence, the converged connection weights corresponding to the node are taken as the latest connection weights for that node.

[0010] Furthermore, the method for obtaining the total drainage volume is as follows: The total drainage volume for the current cycle is obtained by arithmetically summing the drainage volumes corresponding to all time windows from the first time window to the shutdown time window.

[0011] Furthermore, the method for controlling the drainage in the next cycle is as follows: The total drainage volume of the current cycle is compared with the preset minimum flushing volume. The maximum value of the two is divided by the preset single-cycle displacement constant of the drug delivery pump to obtain the target total number of drive cycles for the next drug delivery stage. The drug delivery pump motor is then controlled to run the target total number of drive cycles. During the negative pressure drainage phase of the next cycle, pressure data is monitored in real time. When the pressure data reaches the maximum value between the highest negative pressure limit of the current cycle and the preset minimum effective suction negative pressure constant, the current drive duty cycle of the vacuum pump is locked.

[0012] Furthermore, the method for obtaining the preset judgment threshold is as follows: The ratio of a preset constant to a preset quantity is used as the preset judgment threshold.

[0013] Furthermore, the method for obtaining the drainage cycle is as follows: The continuous cycle of drug administration, resting, and drainage is assigned an incremental integer number to define the drainage cycle.

[0014] The present invention has the following beneficial effects: This invention first divides each drainage cycle of the drainage process into time windows of preset duration, discretizing the continuous analog signal into time-series slices for easy algorithm analysis. To extract the core basic features that stably reflect the wound drainage status, the invention then obtains the drainage volume, average pressure consumption, and unit drainage pressure consumption value for each time window based on weight and pressure data within each time window, facilitating the subsequent construction of a one-dimensional consumption sequence and a two-dimensional topology network. To accurately filter out high-frequency solid-state blockage noise, the invention extracts a preset number of unit drainage pressure consumption values, including the current time window, to form an analysis sequence, ensuring sufficient historical reference windows. Furthermore, a visibility graph algorithm is used to analyze each... Numerical values ​​are mapped to nodes in a time sequence and an undirected adjacency matrix is ​​generated based on geometric anti-occlusion constraints. By transforming one-dimensional time-series signals into two-dimensional topological features, the geometric occlusion relationships during the resistance increase process are accurately reflected. This helps to remove interference from the absolute amplitude of simple one-dimensional numerical values ​​and isolate solid impurities in multiphase flow environments. To quantify the pivotal role of nodes in the network, a weight transfer algorithm is used to obtain the latest connection weights of nodes corresponding to time windows, accurately reflecting the true resistance level of the current drainage and facilitating subsequent accurate identification of the free liquid emptying status. To filter out single-time topological weight spikes caused by solid debris inside the pipeline, the latest connection weights corresponding to a consecutive specified number of time windows are averaged. Within the current time window of the preset judgment threshold, the shutdown time window is used to ensure that the free fluid has been stably depleted and the sponge dressing has been firmly adhered to the granulation tissue. This avoids misjudging instantaneous spikes caused by solid debris as the depletion of free fluid, which could lead to premature shutdown. From the shutdown time window, preset historical time windows are traced back to extract clean historical data that has not undergone severe mechanical compression or deformation, providing samples for accurately deducing the safe pressure limit boundary that does not damage the granulation tissue. Furthermore, the reciprocal of the latest connection weight of each preset historical time window is used as a penalty factor to perform a weighted average of the pressure consumption mean of each preset historical time window, obtaining the highest negative pressure limit value for the current cycle. This proactively reduces the proportion of high-pressure data near the shutdown moment and excludes data from the system. The deformation interference caused by the compression of the cotton dressing allows for the accurate determination of the true pressure reference standard during the smooth drainage phase of the liquid phase. Based on the total drainage volume from the start of the current cycle to the shutdown time window, the total drainage volume of the current cycle is obtained, eliminating the false space caused by free air bubbles and dressing deformation, and objectively reflecting the effective physical volume of the wound cavity that can actually hold liquid in the current cycle. Furthermore, based on the highest negative pressure limit and total drainage volume of the current cycle, the drainage of the next cycle is controlled, achieving adaptive dynamic matching between the infusion volume of the medicine and the suction force of the vacuum pump. This completes the closed-loop control of equipment operation based on actual drainage to determine the medication and based on safety to determine the suction, which not only prevents excess medicine from overflowing but also avoids mechanical damage to granulation tissue. Attached Figure Description

[0015] Figure 1This is a structural block diagram of a continuous drug delivery negative pressure drainage control system for diabetic foot wounds, provided in one embodiment of the present invention. Detailed Implementation

[0016] The following description, in conjunction with the accompanying drawings, details the specific scheme of a continuous drug delivery negative pressure drainage control system for diabetic foot wounds provided by the present invention.

[0017] Example: This invention proposes a continuous drug delivery negative pressure drainage control system for diabetic foot wounds. Please refer to [link / reference]. Figure 1 The diagram shows a structural block diagram of a continuous drug delivery negative pressure drainage control system for diabetic foot wounds according to an embodiment of the present invention. The system includes: a data acquisition module 10, a connection weight acquisition module 20, a current cycle analysis module 30, and a drainage control module 40.

[0018] The data acquisition module 10 is used to acquire the weight data of the drained fluid and the pressure data of the wound at each moment during the continuous negative pressure drainage process of the diabetic foot wound; the drainage cycle of the drainage process is divided into time windows according to a preset duration, and based on the weight data and pressure data in each time window, the drainage volume, average pressure consumption and unit drainage pressure consumption value of each time window are acquired.

[0019] Specifically, to objectively monitor the actual fluid mass discharged from the diabetic foot wound and eliminate errors caused by free gas, this embodiment uses a weighing sensor installed at the bottom of the drainage waste collection tank to acquire the weight data of the discharged fluid at each moment during the continuous drug administration negative pressure drainage of the diabetic foot wound, reflecting the true physical extraction volume. Pressure sensors placed inside the wound dressing cavity or near the wound end of the drainage tube acquire the pressure data of the wound at each moment during the continuous drug administration negative pressure drainage of the diabetic foot wound, reflecting the actual negative pressure suction intensity experienced by the wound. To ensure a strict temporal correspondence between the drainage volume and pressure output, this embodiment sets the weight and pressure data to be collected synchronously at a frequency of 10Hz, ensuring accurate capture of instantaneous blockage characteristics and flow rate changes within the tube. The implementer can set the collection frequency based on the performance of the underlying hardware interface and the computing power of the microprocessor; no limitation is imposed here.

[0020] Considering that perfusion and aspiration alternate during drainage, to achieve cross-cycle parameter transfer and dynamic evolution management, the continuously executed drug administration, settling, and drainage cycles are assigned incremental integer numbers to define the drainage cycle. That is, a complete "drug administration-settling-drainage" process is defined as an independently numbered control closed-loop cycle, which facilitates subsequent closed-loop control of the hardware actions in the next cycle based on historical cycle data. To discretize the continuous analog signal into time-series slices that are easy for algorithm analysis, each drainage cycle is divided into time windows of preset duration, facilitating the extraction of topological resistance characteristics per unit time. In this embodiment, the preset duration is set to 1 second to ensure a balance between the monitoring accuracy of resistance changes and the computing power of the embedded system. Implementers can set the preset duration according to the flow rate change rate and hardware computing limits; no limitation is imposed here.

[0021] Considering that directly using instantaneous raw data is easily affected by high-frequency mechanical noise and cannot characterize the cumulative effect, in order to extract the core basic feature tuples that can stably reflect the wound drainage status, we then obtain the drainage volume, average pressure consumption, and unit drainage pressure consumption value for each time window based on the weight and pressure data within each time window, providing a data foundation for subsequent analysis. The method for obtaining the drainage volume, average pressure consumption, and unit drainage pressure consumption is as follows: For any time window, the weight data within that time window is differentiated and filtered to obtain the instantaneous flow rate, resulting in a smooth instantaneous mass change rate. To effectively eliminate the volume expansion error caused by free bubbles in the multiphase flow, the instantaneous flow rate is integrated over time and divided by a preset density to obtain the drainage volume of that time window, objectively reflecting the true volume of the liquid and solid substances actually discharged from the wound. In this embodiment, the preset density is set to 1.05 g / cm³ based on the average density of typical diabetic foot exudate obtained through clinical statistics. The preset density can be set according to the actual situation and is not limited here. It should be noted that, considering that the derivative signal of the weighing sensor will generate instantaneous apparent negative fluctuations when the drainage waste collection tank is subjected to accidental shaking or bumps, in order to filter the apparent weight, reduce errors, and prevent abnormal calculation of subsequent physical resistance, if the drainage volume is less than 0, the drainage volume is forcibly assigned to 0. To quantify the total mechanical load continuously applied to the wound surface by the system for fluid aspiration within a given time window, the pressure data within this time window is integrated over time to obtain the cumulative pressure value, reflecting the total energy consumption of granulation tissue under negative pressure traction. To provide a standardized pressure reference baseline, the cumulative pressure value is divided by the duration of the time window to obtain the average pressure consumption within that window, laying the foundation for historical pressure retrospection and weighted penalty. To accurately assess the physical resistance encountered when extracting a unit volume of fluid within this time window, the cumulative pressure value is compared with the drainage volume and pre-extraction volume. The ratio of the sum of the smallest positive numbers is used as the unit drainage pressure consumption value for this time window. That is, the sum of the drainage volume and the preset smallest positive number is used as the denominator. This avoids the drainage volume approaching 0 when the free fluid is exhausted at the end of the drainage period, which would cause microprocessor numerical underflow errors and computational deadlocks when direct division is performed. In this embodiment, the preset smallest positive number is set to 0.05 mL to ensure that the system is given a weak but non-zero virtual volume base during the fluid drainage stage to maintain the stable operation of the algorithm. The implementer can set the preset smallest positive number according to the conventional inner diameter of the medical drainage tube and the lower limit volume of a single drop of exudate. There is no limitation here.

[0022] The connection weight acquisition module 20 is used to extract a continuous preset number of unit drainage pressure consumption values, including the current time window, to form an analysis sequence. The visibility graph algorithm is used to map each value in the analysis sequence to a node according to the time sequence and generate an undirected adjacency matrix based on geometric anti-occlusion constraints. Combined with the weight transfer algorithm, the latest connection weight of the node corresponding to the time window is obtained.

[0023] Specifically, considering that the instantaneous blockage of solid necrotic debris in multiphase flow pipelines can cause extremely high pressure spikes, this embodiment extracts a series of continuously preset unit drainage pressure consumption values, including the current time window, to form an analysis sequence in order to accurately filter out high-frequency solid blockage noise, ensuring sufficient historical reference windows. To remove interference from absolute amplitude in the one-dimensional time-series signal, a visibility graph algorithm is used to map each value in the analysis sequence to nodes according to time sequence and generate an undirected adjacency matrix based on geometric anti-occlusion constraints. This transforms the one-dimensional time-series signal into two-dimensional topological features, accurately reflecting the geometric occlusion relationship during the resistance increase process, which is beneficial for isolating solid impurity interference in multiphase flow environments. To quantify the pivotal position of nodes in the network, a weight transfer algorithm is combined to obtain the latest connection weights of nodes corresponding to the time window, accurately reflecting the true resistance level of the current drainage, which is beneficial for subsequent accurate identification of the free liquid drainage status. The visibility graph algorithm and weight transfer algorithm are well-known and will not be elaborated further.

[0024] Preferably, in one feasible embodiment, the method for obtaining the analysis sequence is as follows: Within the current cycle, if the number of time windows up to the current execution reaches a preset number, it indicates that the accumulated historical data within the current cycle has filled the preset observation field. To perform panoramic occlusion judgment, a preset number of consecutive unit drainage pressure consumption values, including the current time window and its preceding time windows, are extracted sequentially to form an analysis sequence. If the number of time windows up to the current execution does not reach the preset number, it indicates that the system is in a cold start collection phase where the required topological feature sequence has not yet been filled. To prevent blind aspiration, the control system maintains a low-pressure collection state to implement hardware fallback protection. Considering that very small ulcer wounds may experience premature fluid desiccation, further... When the drainage volume corresponding to the consecutive reference quantity time windows is less than or equal to a preset minimum positive number (e.g., 0.05 mL), it indicates that the limited free fluid in the wound has been completely drained, thus forcibly exiting the low-pressure collection state and establishing the current time window as the shutdown time window, avoiding a deadlock state of continuous dry drainage due to incomplete sequence. To ensure that subsequent safety baseline calculations have valid input parameters in the case of forced shutdown due to insufficient preset quantity, the control system further extracts the unit drainage pressure consumption values ​​corresponding to all time windows accumulated in the current cycle to form an analysis sequence. The actual quantity is used to reduce the dimension and replace the preset quantity to execute the visibility graph algorithm and weight transfer algorithm, obtaining the latest connection weights corresponding to each accumulated time window for subsequent calculation of the maximum negative pressure limit value. In this embodiment, the preset quantity is set to 40 to ensure coverage of a complete long-cycle obstruction fluctuation. The implementer can set the preset quantity according to the specific clinical drainage characteristics, which is not limited here. The reference quantity is set to 5 to ensure effective confirmation of fluid depletion. The implementer can set the reference quantity according to the sensor sensitivity, which is not limited here.

[0025] Preferably, in one feasible embodiment of this method, the method for obtaining the geometric anti-occlusion constraint is as follows: a two-dimensional coordinate system is constructed using the time series index as the abscissa and the unit drainage pressure consumption value as the ordinate. The node corresponding to each element in the analysis sequence in the two-dimensional coordinate system is determined, which facilitates the subsequent intuitive geometric occlusion determination and topology network construction. For the first node and the second node that are not temporally adjacent, and any intermediate node located between the time axes of the first node and the second node, the longitudinal difference between the second node and the first node (i.e., the unit drainage pressure consumption value corresponding to the second node minus the unit drainage pressure consumption value corresponding to the first node) is calculated as the total analysis difference. The position ratio of the intermediate node in the total time span between the first node and the second node is calculated as the analysis ratio of the intermediate node (i.e., the abscissa corresponding to the intermediate node is subtracted from the abscissa corresponding to the first node to obtain the analysis numerator, the abscissa corresponding to the second node is subtracted from the abscissa corresponding to the first node to obtain the analysis denominator, and the ratio of the analysis numerator to the analysis denominator is used as the analysis ratio of the intermediate node). Among them, the first node is located before the second node in time series. To accurately map the line-of-sight connectivity between the first and second nodes, the total analytical difference is multiplied by the analytical ratio and then added to the unit drainage pressure consumption value corresponding to the first node to obtain the theoretical vertical height of the intermediate node. This theoretical height represents the theoretical projection height of the line connecting the top of the first node and the top of the second node at the intermediate node in a geometric sense. If the unit drainage pressure consumption value corresponding to all intermediate nodes located between the first and second nodes is less than their theoretical vertical height, it indicates that no node between the first and second nodes produces a sudden spike sufficient to block the first node from looking at the second node. In this case, the first and second nodes satisfy the geometric anti-occlusion constraint condition.

[0026] Preferably, in one feasible method of this embodiment, the method for obtaining the undirected adjacency matrix is ​​as follows: For any two nodes, when the two nodes are not temporally adjacent, if the two nodes satisfy the geometric anti-occlusion constraint, it means that the temporal evolution between the two nodes is not affected by high-frequency blocking interference, and then a connecting edge is established between the two nodes; if the two nodes do not satisfy the geometric anti-occlusion constraint, it means that there is obvious resistance peak occlusion between the two nodes, and then no connecting edge is established between the two nodes; when the two nodes are temporally adjacent, it means that there is no interference from any intermediate nodes between the two nodes, and then a connecting edge is directly established between the two nodes; in order to construct standardized algorithm input parameters, the nodes are used as the rows and columns of the matrix, and the elements at the corresponding positions of the two nodes with established connecting edges are assigned the value 1, and the elements at the corresponding positions of the two nodes without established connecting edges are assigned the value 0, generating an undirected adjacency matrix with a size determined by a preset number.

[0027] Preferably, in one feasible method of this embodiment, the latest connection weight is obtained as follows: First, based on the undirected adjacency matrix, determine the neighboring nodes of each node and the total number of connected edges between the neighboring nodes (i.e., the degree of the neighboring nodes); considering the resistance transmission characteristics between different nodes, in order to fairly assign the initial state, assign each node an initial connection weight (i.e., the ratio of a constant 1 to the total number of nodes, i.e., a preset number, to ensure that the sum of the initial connection weights of the entire network is 1). For any node, according to the preset damping coefficient, divide the connection weight of each neighboring node in the previous round by the total number of connected edges to obtain the allocation weight, which accurately reflects the probability contribution value of each neighboring node to the current state of the node. In order to prevent the weight loss of isolated nodes in the undirected graph, add the sum of the allocation weights received by the node from all its neighboring nodes to the basic compensation value determined based on the preset damping coefficient and the total number of nodes, and iteratively update the connection weight of the node; wherein, the calculation formula for updating the connection weight is: In the formula, is the connection weight corresponding to the nth iteration of this node; M is the total number of nodes, i.e., the preset number; d is the preset damping coefficient; Y is the total number of adjacent nodes of this node; This represents the connection weight of the y-th neighboring node in the (n-1)th iteration. This represents the total number of connected edges between the y-th neighboring node of this node; Basic compensation value; To assign weights, this embodiment sets a preset damping coefficient d of 0.85 to ensure the algorithm achieves the best balance between reflecting local connectivity characteristics and the probability of global random walks. Implementers can set the preset damping coefficient d according to the typical blockage frequency of the drainage pipeline; no limitation is imposed here. It should be noted that in the first iteration (i.e....) When the formula is used, the connection weights of each adjacent node in the previous round are all taken as the set initial connection weights; After multiple iterations until convergence (i.e., reaching a preset number of iterations, such as 20, or the absolute value of the difference between the node weights in two consecutive rounds is less than a preset minimum threshold, such as 0.0001), the convergence connection weight corresponding to the node is taken as the latest connection weight of the node. The larger the latest connection weight, the more it indicates that the drainage resistance within the time window corresponding to the node shows a continuous and unobstructed high-level increase, objectively characterizing the real physical process of the sponge dressing gradually compressing the granulation tissue.

[0028] The current cycle analysis module 30 is used to take the current time window where the latest connection weights of a specified number of consecutive time windows are all greater than a preset judgment threshold as the shutdown time window; to trace back from the shutdown time window to the preset historical time windows, and to use the reciprocal of the latest connection weights of each preset historical time window as a penalty factor to perform a weighted average of the pressure consumption average of each preset historical time window to obtain the highest negative pressure limit value of the current cycle; and to obtain the total drainage volume of the current cycle based on all drainage volumes from the start of the current cycle to the shutdown time window.

[0029] Specifically, it is known that fluid pipelines mixed with air bubbles and necrotic debris will generate instantaneous resistance spikes, which overlap with the sustained high resistance after the liquid has been completely drained in one dimension. To filter out the interference of single-time topology weight spikes caused by solid debris inside the pipeline, the current time window where the latest connection weights corresponding to a specified number of consecutive time windows are all greater than a preset judgment threshold is used as the shutdown time window. This ensures that the free liquid has stabilized and dried up and that the sponge dressing has effectively adhered to the granulation tissue, avoiding misjudging instantaneous spikes caused by solid debris as the emptying of free liquid and causing premature shutdown. In this embodiment, the specified number is set to 3 to ensure verification. The continuous, monotonous increase in resistance can be controlled by the operator by setting a specified number based on the normal flow rate and tubing length; no limit is imposed here. A preset judgment threshold is set as the ratio of a preset constant to a preset number (i.e., the total length of the analysis sequence), establishing a relative limit standard adapted to the scale of dynamic sequences. The preset constant is set to 3.5, meaning the threshold is defined as 3.5 times the equal weight of nodes within the sequence. The operator can set the preset constant based on the actual porosity of the dressing in clinical practice; no limit is imposed here. Simultaneously with establishing the shutdown time window, the control system immediately sends a power-off command to the vacuum pump power supply circuit to cut off excessive negative pressure suction as soon as possible.

[0030] To accurately deduce the safe pressure limit boundary without damaging granulation tissue, a preset historical time window is traced back from the shutdown time window to extract clean historical data that has not undergone severe mechanical compression deformation. It should be noted that to ensure the validity of the algorithm's calculation sequence and that the data is within the effective observation field, the number of backtracking time windows must be strictly less than the preset number. Here, the historical time window specifically refers to all actually available time windows from the first time window of the current cycle to the shutdown time window. If the actual number of available time windows is less than the preset backtracking number, it indicates that the wound is extremely small or the drainage rate is extremely fast, causing the shutdown to occur prematurely. To avoid negative or zero indexes in the underlying array that could trigger out-of-bounds errors and deadlocks, the data will be directly retrieved from the current time window. All actual available time windows between the first time window of the previous cycle and the shutdown time window are used as preset historical time windows. If the number of actual available time windows is greater than or equal to the preset backtracking number, it means that sufficient previous historical data has been accumulated in the current cycle. Then, a preset number of consecutive backtracking historical time windows are extracted from the shutdown time window as preset historical time windows. This ensures that a sufficient number of historical samples are provided for smooth weighting while eliminating invalid data from the distant past. In this embodiment, the preset backtracking number is set to 15 to ensure that sufficient historical samples are provided to calculate a stable safety pressure baseline. The implementer can set the preset backtracking number according to the conventional displacement of the vacuum pump and the average depth of the wound. This is not limited here. Considering that the pressure values ​​at the moment of shutdown and in the immediate preceding period already include extreme errors caused by the compression of the dressing tissue, in order to proactively reduce the proportion of high-pressure data near the shutdown moment and eliminate deformation interference caused by the compression of the sponge dressing, the reciprocal of the latest connection weight of each preset historical time window is used as a penalty factor to perform a weighted average of the pressure consumption average of each preset historical time window, thereby obtaining the highest negative pressure limit value of the current cycle and accurately tracing back to the true pressure reference standard of the smooth liquid phase evacuation stage; the formula for calculating the highest negative pressure limit value is as follows: In the formula, I represents the maximum negative pressure limit for the current cycle; I represents the total number of preset historical time windows. The latest connection weight for the i-th preset historical time window is greater than 0; The penalty factor for the i-th preset historical time window; This represents the average pressure consumption for the i-th preset historical time window. It should be noted that the closer to the shutdown time, the more severe the sponge compression, the greater the node connection weight, and the smaller the penalty factor. The high average pressure consumption is significantly weakened in the weighted average, thus accurately extracting the safe benchmark under unpressurized conditions. To achieve adaptive matching between the infusion volume of the medicine in the next cycle and the actual volume of the wound cavity, the total drainage volume of the current cycle is obtained based on all drainage volumes from the start of the current cycle to the stop time window. This is achieved by arithmetically summing the drainage volumes corresponding to all time windows from the first time window to the stop time window within the current cycle, eliminating false spaces caused by free air bubbles and dressing deformation, and objectively reflecting the effective physical volume of the wound cavity that can actually hold liquid in the current cycle.

[0031] The drainage control module 40 is used to control the drainage for the next cycle based on the maximum negative pressure limit and total drainage volume of the current cycle.

[0032] Specifically, as granulation tissue continues to grow, the actual volume of the cavity in the wound that can accommodate the medication is in a process of continuous physiological evolution and shrinkage. Furthermore, excessive suction negative pressure can cause traction and ischemic damage to newly formed capillaries. In order to achieve adaptive dynamic matching between the medication perfusion volume and the vacuum pump suction force, and based on the maximum negative pressure limit value and total drainage volume of the current cycle, the drainage of the next cycle is controlled. This completes the closed-loop control of equipment operation based on actual drainage to determine medication and based on safety retrospective to determine suction, which not only prevents excess medication from overflowing, but also avoids mechanical damage to the granulation tissue.

[0033] The method for controlling drainage in the next cycle is as follows: The total drainage volume of the current cycle is compared with the preset minimum guaranteed flushing volume. The maximum value is divided by the preset single-cycle displacement constant of the drug delivery pump to obtain the target total number of drive cycles for the next cycle's drug delivery phase. The drug delivery pump motor is then controlled to achieve this target total number of drive cycles, ensuring that the total amount of medication to be infused into the wound is exactly equal to the actual effective physical space on the wound surface in the previous cycle. Simultaneously, a forced bottom-line protection mechanism prevents a new cycle from being triggered by occasional severe blockage of the tubing in the previous cycle (causing the total drainage volume to approach zero). Drug administration stalls and lockouts; This embodiment sets a preset minimum guaranteed irrigation volume of 5 mL to ensure that the wound surface is maintained at a minimum level of physical irrigation and moist environment. The implementer can set the preset minimum guaranteed irrigation volume according to the clinical wound area assessment, which is not limited here; The preset single-revolution displacement constant of the drug administration pump is set to 0.2 mL / revolution to ensure that the required drug volume is accurately quantified and converted into the number of mechanical revolutions that the motor can directly execute. The implementer can calibrate the preset single-revolution displacement constant of the drug administration pump according to the inner diameter of the pump tube and the rotor extrusion specification of the selected medical peristaltic pump, which is not limited here; To prevent a situation where the extremely smooth fluid flow in the previous cycle leads to an excessively low safety limit, and in the new cycle, encountering viscous necrotic material results in insufficient pumping force to overcome resting resistance, thus causing a permanent liquid lock-in, pressure data is monitored in real time during the negative pressure drainage phase of the next cycle. When the pressure data reaches the maximum value between the current cycle's highest negative pressure limit and the preset minimum effective suction negative pressure constant, the current drive duty cycle of the vacuum pump is locked. By introducing a lower limit cutoff constraint, a safety net is introduced to ensure that the vacuum pump has the basic liquid drainage capacity to overcome the initial resting resistance in the new cycle, while maintaining uniform suction within this safe range. In this embodiment, the preset minimum effective suction negative pressure constant is set to 5 kPa to ensure basic liquid drainage power. The implementer can set the preset minimum effective suction negative pressure constant according to the expected viscosity limit of the drainage material; this is not limited here.

[0034] It should be noted that for the first drainage cycle, since the system's internal time-series cache queue lacks historical reference data from the previous cycle, directly retrieving variables would result in the extraction of null values, causing control stagnation. Therefore, in the first drainage cycle, the control system directly retrieves pre-programmed startup baseline parameters (such as the default exploratory negative pressure threshold and baseline drug delivery volume) to complete the initial cycle's drug delivery and drainage operation. From the second drainage cycle onwards, the system automatically disables the startup baseline parameters, fully taking over and strictly relying on the aforementioned cross-cycle parameter flow and feedback mechanism for dynamic closed-loop control.

[0035] In summary, this embodiment acquires weight and pressure data during the drainage process, divides time windows, and calculates drainage volume, average pressure consumption, and unit drainage pressure consumption. A preset number of unit drainage pressure consumption values ​​are extracted to form a sequence, which is then converted into an undirected adjacency matrix using a visibility graph algorithm. A weighted transfer algorithm is used to obtain the latest connection weights of nodes to reflect the actual resistance. Time windows where the weight is consistently greater than a preset threshold are used as shutdown time windows to prevent misjudgments of instantaneous blockage. Historical time windows are backtracked, and a weighted average is calculated using the reciprocal of the weight as a penalty factor to deduce the safe maximum negative pressure limit without pressure error, and the total drainage volume is accumulated. Finally, the next cycle is controlled based on this. This invention achieves dynamic adaptive matching of drug dosage and aspiration negative pressure, preventing drug spillage and protecting granulation tissue.

Claims

1. A continuous drug delivery negative pressure drainage control system for diabetic foot wounds, characterized in that, The system includes: The data acquisition module is used to acquire the weight data of the drained fluid and the pressure data of the wound at each moment during the continuous negative pressure drainage process of diabetic foot wound; the drainage cycle of the drainage process is divided into time windows according to the preset duration, and based on the weight data and pressure data in each time window, the drainage volume, average pressure consumption and unit drainage pressure consumption value of each time window are acquired; The connection weight acquisition module is used to extract a continuous preset number of unit drainage pressure consumption values, including the current time window, to form an analysis sequence. The visibility graph algorithm is used to map each value in the analysis sequence to a node according to the time sequence and generate an undirected adjacency matrix based on geometric anti-occlusion constraints. Combined with the weight transfer algorithm, the latest connection weight of the node corresponding to the time window is obtained. The current cycle analysis module is used to define the current time window where the latest connection weights for a specified number of consecutive time windows are all greater than a preset judgment threshold as the shutdown time window; it traces back from the shutdown time window to preset historical time windows, uses the reciprocal of the latest connection weights of each preset historical time window as a penalty factor, and performs a weighted average of the pressure consumption average of each preset historical time window to obtain the maximum negative pressure limit value of the current cycle; and it obtains the total drainage volume of the current cycle based on all drainage volumes from the start of the current cycle to the shutdown time window. The drainage control module is used to control the drainage for the next cycle based on the maximum negative pressure limit and total drainage volume of the current cycle.

2. The continuous drug delivery negative pressure drainage control system for diabetic foot wounds as described in claim 1, characterized in that, The methods for obtaining the drainage volume, average pressure consumption, and unit drainage pressure consumption are as follows: For any time window, the weight data within that time window is differentiated and filtered to obtain the instantaneous flow rate. The instantaneous flow rate is then integrated over time and divided by a preset density to obtain the drainage volume for that time window. If the drainage volume is less than 0, it is forcibly assigned the value of 0. Integrate the pressure data within the time window over time to obtain the cumulative pressure value for that time window. Divide the accumulated pressure value by the duration of the time window to obtain the average pressure consumption of the time window; The ratio of the accumulated pressure value to the sum of the drainage volume and a preset minimum positive number is used as the unit drainage pressure consumption value for that time window.

3. The continuous drug delivery negative pressure drainage control system for diabetic foot wounds as described in claim 2, characterized in that, The method for obtaining the analysis sequence is as follows: If the number of time windows up to the current execution reaches a preset number within the current cycle, then extract a preset number of consecutive units of drainage pressure consumption values, including the current time window and its preceding time windows, to form an analysis sequence. If the number of time windows up to the current execution point has not reached the preset number, the control system maintains the low-pressure collection state. When the drainage volume corresponding to the consecutive reference number of time windows is less than or equal to the preset minimum positive number, the system forcibly exits the low-pressure collection state and establishes the current time window as the shutdown time window. The system extracts the unit drainage pressure consumption value of all time windows accumulated in the current cycle to form an analysis sequence.

4. The continuous drug delivery negative pressure drainage control system for diabetic foot wounds as described in claim 1, characterized in that, The method for obtaining the geometric anti-occlusion constraints is as follows: A two-dimensional coordinate system is constructed using the time series index as the horizontal axis and the unit drainage pressure consumption value as the vertical axis, and the node corresponding to each element in the analysis sequence in the two-dimensional coordinate system is determined. For the first and second nodes that are not temporally adjacent, and any intermediate node located between the time axes of the first and second nodes, calculate the longitudinal difference between the second node and the first node as the total analysis difference. Calculate the proportion of the intermediate node's position within the total time span of the first and second nodes, and use this proportion as the analysis ratio for the intermediate node; where the first node precedes the second node in the time sequence. Multiply the total analytical difference by the analytical ratio, and add it to the unit drainage pressure consumption value corresponding to the first node to obtain the theoretical longitudinal height of the intermediate node. If the unit drainage pressure consumption value of all intermediate nodes located between the time axes of the first and second nodes is less than their theoretical longitudinal height, then the first and second nodes satisfy the geometric anti-shading constraint conditions.

5. A continuous drug delivery negative pressure drainage control system for diabetic foot wounds as described in claim 4, characterized in that, The method for obtaining the undirected adjacency matrix is ​​as follows: For any two nodes, if the two nodes are not temporally adjacent, and the two nodes satisfy the geometric anti-occlusion constraint, then a connecting edge is established between the two nodes; if the two nodes do not satisfy the geometric anti-occlusion constraint, then a connecting edge is not established between the two nodes. When the two nodes are temporally adjacent, a connecting edge is directly established between the two nodes. Using nodes as rows and columns of a matrix, assign the value 1 to the corresponding element of two nodes with connected edges and the value 0 to the corresponding element of two nodes without connected edges, thus generating an undirected adjacency matrix.

6. The continuous drug delivery negative pressure drainage control system for diabetic foot wounds as described in claim 5, characterized in that, The method for obtaining the latest connection weight is as follows: The number of neighboring nodes and the total number of connected edges between neighboring nodes are determined based on the undirected adjacency matrix. Each node is assigned an initial connection weight. For any node, according to the preset damping coefficient, the connection weight of each of the node's neighboring nodes in the previous round is divided by the total number of its connected edges to obtain the allocation weight. The sum of the allocation weights received by the node from all its neighboring nodes is added to the basic compensation value determined based on the preset damping coefficient and the total number of nodes, and the connection weight of the node is iteratively updated. After multiple iterations until convergence, the converged connection weights corresponding to the node are taken as the latest connection weights for that node.

7. A continuous drug delivery negative pressure drainage control system for diabetic foot wounds as described in claim 1, characterized in that, The method for obtaining the total drainage volume is as follows: The total drainage volume for the current cycle is obtained by arithmetically summing the drainage volumes corresponding to all time windows from the first time window to the shutdown time window.

8. The continuous drug delivery negative pressure drainage control system for diabetic foot wounds as described in claim 1, characterized in that, The method for controlling the drainage in the next cycle is as follows: The total drainage volume of the current cycle is compared with the preset minimum flushing volume. The maximum value of the two is divided by the preset single-cycle displacement constant of the drug delivery pump to obtain the target total number of drive cycles for the next drug delivery stage. The drug delivery pump motor is then controlled to run the target total number of drive cycles. During the negative pressure drainage phase of the next cycle, pressure data is monitored in real time. When the pressure data reaches the maximum value between the highest negative pressure limit of the current cycle and the preset minimum effective suction negative pressure constant, the current drive duty cycle of the vacuum pump is locked.

9. A continuous drug delivery negative pressure drainage control system for diabetic foot wounds as described in claim 1, characterized in that, The method for obtaining the preset judgment threshold is as follows: The ratio of a preset constant to a preset quantity is used as the preset judgment threshold.

10. A continuous drug delivery negative pressure drainage control system for diabetic foot wounds as described in claim 1, characterized in that, The method for obtaining the diversion cycle is as follows: The continuous cycle of drug administration, resting, and drainage is assigned an incremental integer number to define the drainage cycle.

Images