A method and system for preparing a low endotoxin hemodialysis concentrate

Abstract

The application relates to the technical field of hemodialysis, in particular to a preparation method and system of a low-endotoxin hemodialysis concentrate, which comprises the following steps: obtaining a preparation data set, extracting a temperature sequence, a pH value sequence and an endotoxin content index value, calculating a membrane flux distribution sequence and an adsorbent activity distribution sequence, performing primary dialysis path optimization to generate a first complete optimization network; identifying a target path node pair according to a target endotoxin content value, performing secondary path expansion to generate a second complete optimization network; and finally identifying a temperature and pH value control sequence corresponding to the target path node pair to complete the preparation. The application significantly reduces the test workload through optimization of the network, improves the endotoxin removal efficiency and the control precision of the process parameters, and solves the problems of low endotoxin removal efficiency and poor product batch stability in the prior art.

Description

Technical Field

[0001] This invention belongs to the field of hemodialysis technology, specifically a method and system for preparing low endotoxin hemodialysis concentrate. Background Technology

[0002] With the continuous development of hemodialysis technology, the importance of low-endotoxin hemodialysis concentrate in improving patient treatment outcomes and quality of life is becoming increasingly prominent. However, existing methods and systems for preparing hemodialysis concentrate still have shortcomings in reducing endotoxin content, ensuring product quality stability, and simplifying process flows, hindering their further promotion and application. Among existing technologies, patent CN103432638B proposes a production device that uses a dissolving tank, adsorbent, and filter connected in series, combined with a pressure pump and built-in adsorbent design, significantly reducing endotoxin content to the international standard of ≤0.25 EU / ml, while simultaneously improving product clarity to the standard of ≤0.5 colorimetric solution. However, this technical solution is highly dependent on the type and amount of adsorbent used, and adsorbent performance degradation may lead to a decrease in the consistency of treatment effects. Furthermore, this solution does not fully consider the precise control of key parameters such as temperature and pH during the preparation process, which may result in large batch-to-batch quality fluctuations, making it difficult to meet the high standards of clinical requirements for ultrapure dialysis solutions. Furthermore, patent CN118948886B proposes a method for preparing hemodialysis concentrate using citric acid as the acid in solution A. This method improves the stability and quality safety of the hemodialysis concentrate through secondary feeding, adjusting the order of raw material addition, and strict control of temperature and pH. However, this method requires a high degree of automation in the dosage and timing of citric acid addition, increasing process complexity and operational difficulty. Simultaneously, this method primarily focuses on optimizing solution A, while offering limited measures for endotoxin removal in solution B. This may result in difficulty in further reducing the endotoxin level of the final mixture, failing to fully meet the requirements for preparing low-endotoxin hemodialysis concentrate.

[0003] The aforementioned problems indicate that existing methods and systems for preparing hemodialysis concentrate still have certain shortcomings in terms of endotoxin removal efficiency, precise control of process parameters, batch stability of products, and overall system integration. Therefore, developing a low-endotoxin hemodialysis concentrate preparation method and system that can optimize the endotoxin removal process, achieve real-time monitoring and automatic adjustment of key parameters, and improve system integration and intelligence is of significant practical importance and application value. This invention aims to solve the above problems through technological innovation, significantly reduce endotoxin content, ensure product quality stability and consistency, and meet the modern clinical demand for efficient and safe hemodialysis concentrate. Summary of the Invention

[0004] This invention provides a method and system for preparing low-endotoxin hemodialysis concentrate, primarily aimed at solving the problems of low endotoxin removal efficiency, inaccurate process parameter control, and poor batch stability in existing hemodialysis concentrate preparation processes. To achieve the above objective, this invention provides a method for preparing low-endotoxin hemodialysis concentrate, comprising: acquiring a hemodialysis concentrate preparation dataset; sequentially extracting preparation data from the preparation dataset, extracting temperature, pH, and endotoxin content index values ​​from the preparation data, calculating membrane flux distribution and adsorbent activity distribution sequences based on the temperature and pH sequences respectively; performing a dialysis path optimization on the preparation data based on the membrane flux and adsorbent activity distribution sequences to obtain an optimized path, constructing path nodes of the optimized path based on the endotoxin content index values ​​to obtain an iterative optimization network; determining whether the preparation data in the preparation dataset has been completely extracted; if the preparation data... If the preparation data is not completely extracted from the centralized preparation dataset, the process returns to the steps described above for sequentially extracting preparation data from the preparation dataset. If the preparation data in the preparation dataset is completely extracted, the iterative optimization network is used as the first fully optimized network. The target endotoxin content value is obtained, and target path node pairs are identified in the first fully optimized network based on the target endotoxin content value. A second path expansion is performed on the first fully optimized network based on the target path node pairs to obtain the second fully optimized network. Target path nodes are identified on the second fully optimized network based on the target endotoxin content value, and the temperature and pH control sequences corresponding to the target path nodes are identified. Low endotoxin hemodialysis concentrate is prepared based on the temperature and pH control sequences.

[0005] Further, the step of calculating the membrane flux distribution sequence and the adsorbent activity distribution sequence based on the temperature sequence and the pH sequence respectively includes: sequentially extracting temperature values ​​from the temperature sequence, and experimentally determining the distribution of membrane flux based on the temperature values ​​to obtain the membrane flux distribution sequence; sequentially extracting pH values ​​from the pH sequence, and experimentally determining the change in adsorbent activity based on the pH values ​​to obtain the adsorbent activity distribution sequence.

[0006] Furthermore, before calculating the membrane flux distribution sequence and adsorbent activity distribution sequence based on the temperature sequence and pH sequence respectively, the method further includes: obtaining a membrane flux threshold and a temperature threshold; calibrating the membrane flux ratio coefficient experimentally based on the membrane flux threshold and temperature threshold; obtaining an adsorbent activity threshold and a pH threshold; and calibrating the adsorbent activity ratio coefficient experimentally based on the adsorbent activity threshold and pH threshold.

[0007] Further, the step of optimizing the dialysis path based on the membrane flux distribution sequence and the adsorbent activity distribution sequence to obtain an optimized path includes: obtaining an initial dialysis path; sequentially extracting membrane flux values ​​and adsorbent activity values ​​from the membrane flux distribution sequence and the adsorbent activity distribution sequence, respectively; determining segmented control points on the initial dialysis path based on the membrane flux values; identifying the flow direction of the initial dialysis path and calculating the fluid distribution angle based on the flow direction and the adsorbent activity value; randomly selecting a radial offset angle within a preset radial control range; constructing an iterative dialysis path at the segmented control points based on the fluid distribution angle and the radial offset angle; determining whether the membrane flux values ​​and adsorbent activity values ​​have been completely extracted; if the membrane flux values ​​and adsorbent activity values ​​have not been completely extracted, updating the initial dialysis path using the iterative dialysis path and returning to the steps of sequentially extracting membrane flux values ​​and adsorbent activity values ​​from the membrane flux distribution sequence and the adsorbent activity distribution sequence, respectively; if the membrane flux values ​​and adsorbent activity values ​​have been completely extracted, the optimized path is obtained.

[0008] Further, the step of identifying target path node pairs in the first fully optimized network based on the target endotoxin content value includes: sequentially identifying path-end control points in the first fully optimized network, wherein the path-end control points refer to segment control points directly connected to the path nodes; identifying the path node set on the path-end control points, and identifying the isotopic end concentration range of the path node set; determining whether the target endotoxin content value belongs to the isotopic end concentration range; if the target endotoxin content value does not belong to the isotopic end concentration range, then returning to the step of sequentially identifying path-end control points in the first fully optimized network; if the target endotoxin content value belongs to the isotopic end concentration range, then using the path node set as the first node set to obtain multiple sets of first node sets; in the first fully optimized network... In the optimization network, adjacent control node pairs with the same angle are identified sequentially. An adjacent control node pair with the same angle refers to two path nodes with the same fluid distribution angle and adjacent control points at the end of their respective paths. The adjacent concentration ranges of the adjacent control node pairs with the same angle are identified. It is determined whether the target endotoxin content value belongs to the adjacent concentration range with the same angle. If the target endotoxin content value does not belong to the adjacent concentration range with the same angle, the process returns to the steps described above for sequentially identifying adjacent control node pairs with the same angle in the first fully optimized network. If the target endotoxin content value belongs to the adjacent concentration range with the same angle, the adjacent control node pairs with the same angle are used as second node pairs, resulting in multiple sets of second node pairs. Based on the multiple sets of first node pairs and the multiple sets of second node pairs, the target path node pairs are identified in the first fully optimized network using the target endotoxin content value.

[0009] Further, the step of identifying target path node pairs in the first fully optimized network based on multiple sets of first nodes and multiple sets of second nodes includes: sequentially extracting first node sets from the multiple sets of first nodes; extracting first node pairs of the target endotoxin content value from the first node sets; identifying first adjacent concentration value pairs of the first node pairs to obtain multiple sets of first adjacent concentration value pairs; sequentially extracting second node pairs from the multiple sets of second node pairs; identifying second adjacent concentration value pairs of the second node pairs to obtain multiple sets of second adjacent concentration value pairs; evaluating the first adjacent concentration difference set and the second adjacent concentration difference set in the multiple sets of first adjacent concentration value pairs and the multiple sets of second adjacent concentration value pairs respectively using a preset distance relationship; identifying the minimum concentration difference value in the first adjacent concentration difference set and the second adjacent concentration difference set; identifying the target adjacent concentration value pair corresponding to the minimum concentration difference value; and identifying the target path node pair corresponding to the target adjacent concentration value pair.

[0010] Further, the step of performing secondary path expansion on the first fully optimized network based on the target path node pair to obtain the second fully optimized network includes: obtaining the node relationship attributes and overlapping control paths of the target path node pair; if the node relationship attributes are preset co-located end nodes, then performing dialysis experiments based on the overlapping control paths and preset pH gradients to obtain a set of test concentration values; sequentially extracting test concentration values ​​from the set of test concentration values, identifying the pH values ​​of the test concentration values, and calculating the test distribution angle of the pH values; identifying the target end control points of the target path node pair, and constructing end dialysis paths at the target end control points based on the test distribution angle; constructing test nodes based on the test concentration values, and connecting the test nodes... A second fully optimized network is obtained by linking nodes to the end-dialysis path. If the node relationship attribute is a preset angularly adjacent node, a dialysis experiment is performed according to the overlapping control path and the preset temperature gradient to obtain a set of experimental concentration values. Experimental concentration values ​​are extracted sequentially from the set of experimental concentration values, the temperature of the experimental concentration values ​​is identified, and the experimental membrane flux value at the temperature is calculated. The upper-level control point and the target fluid distribution angle of the target path node pair are identified, and the target end-control point is calculated according to the upper-level control point and the experimental membrane flux value. An end-dialysis path is constructed according to the target end-control point and the target fluid distribution angle. Experimental nodes are constructed according to the experimental concentration values, and the experimental nodes are linked to the end-dialysis path to obtain the second fully optimized network.

[0011] Further, the step of identifying target path nodes on the second fully optimized network based on the target endotoxin content value includes: sequentially extracting test nodes in the second fully optimized network and obtaining the test concentration value corresponding to the test node; calculating the concentration difference between the test concentration value and the target endotoxin content value to obtain a concentration difference set; extracting the minimum concentration difference from the concentration difference set and identifying the target path node corresponding to the minimum concentration difference.

[0012] Further, the identification of the temperature and pH regulation sequence corresponding to the target path node includes: extracting the target membrane flux distribution sequence and the target adsorbent activity distribution sequence of the target path node in the second fully optimized network; identifying the target temperature sequence and the target pH sequence corresponding to the target membrane flux distribution sequence and the target adsorbent activity distribution sequence; and constructing the temperature and pH regulation sequence based on the target temperature sequence and the target pH sequence.

[0013] To achieve the above objectives, the present invention also provides a system for preparing low-endotoxin hemodialysis concentrate, comprising: a first fully optimized network generation module for acquiring a preparation dataset of hemodialysis concentrate; sequentially extracting preparation data from the preparation dataset, extracting temperature sequence, pH value sequence, and endotoxin content index value from the preparation data, calculating membrane flux distribution sequence and adsorbent activity distribution sequence based on the temperature sequence and pH value sequence, respectively; performing a dialysis path optimization on the preparation data based on the membrane flux distribution sequence and adsorbent activity distribution sequence to obtain an optimized path, constructing path nodes of the optimized path based on the endotoxin content index value to obtain an iterative optimization network; determining whether the preparation data in the preparation dataset has been completely extracted; and if the preparation data in the preparation dataset has not been completely extracted, returning to the previous state. The above steps involve sequentially extracting preparation data from the preparation dataset; if all preparation data has been extracted from the preparation dataset, the iterative optimization network is used as the first fully optimized network; the target path node pair identification module is used to obtain the target endotoxin content value and identify the target path node pair in the first fully optimized network based on the target endotoxin content value; the second fully optimized network generation module is used to perform secondary path expansion on the first fully optimized network based on the target path node pair to obtain the second fully optimized network; the low endotoxin hemodialysis concentrate preparation module is used to identify the target path node on the second fully optimized network based on the target endotoxin content value, identify the temperature and pH control sequence corresponding to the target path node, and prepare the low endotoxin hemodialysis concentrate according to the temperature and pH control sequence.

[0014] To address the aforementioned problems, the present invention also provides an electronic device comprising: a memory storing at least one instruction; and a processor executing the instruction stored in the memory to implement the aforementioned method for preparing low endotoxin hemodialysis concentrate.

[0015] To address the aforementioned problems, the present invention also provides a computer-readable storage medium storing at least one instruction, which is executed by a processor in an electronic device to implement the method for preparing low endotoxin hemodialysis concentrate described above.

[0016] This invention constructs a first fully optimized network using existing preparation datasets, significantly reducing the workload of experiments. However, since the endotoxin content index value in the first fully optimized network may differ from the target endotoxin content value, further optimization of the preparation process parameters for hemodialysis concentrate is necessary. Firstly, target path node pairs can be directly identified in the first fully optimized network based on the target endotoxin content value. Then, a second path expansion is performed on the first fully optimized network based on the target path node pairs to obtain a second fully optimized network. This narrows the experimental scope to the process parameter range of the target path node pairs. Finally, target path nodes are identified on the second fully optimized network based on the target endotoxin content value, and the corresponding temperature and pH control sequences are identified. This allows for the preparation of low-endotoxin hemodialysis concentrate based on the temperature and pH control sequences. Therefore, this invention solves the problems of low endotoxin removal efficiency, inaccurate process parameter control, and poor batch stability in existing hemodialysis concentrate preparation processes. Attached Figure Description

[0017] Figure 1 This is a schematic flowchart of a method for preparing low endotoxin hemodialysis concentrate according to an embodiment of the present invention.

[0018] Figure 2 This is a functional block diagram of a low endotoxin hemodialysis concentrate preparation system provided in an embodiment of the present invention;

[0019] Figure 3 This is a schematic diagram of an electronic device for implementing the method for preparing low endotoxin hemodialysis concentrate according to an embodiment of the present invention. Detailed Implementation

[0020] This invention provides a method and system for preparing low-endotoxin hemodialysis concentrate, primarily aimed at solving the problems of low endotoxin removal efficiency, inaccurate process parameter control, and poor batch stability in existing hemodialysis concentrate preparation processes. To achieve the above objectives, this invention combines the attached... Figure 1 To be continued Figure 3The specific implementation method is described in detail. In practical applications, the method and system operate on an electronic device, wherein the memory stores at least one instruction, and the processor executes the instruction to implement the method for preparing low endotoxin hemodialysis concentrate.

[0021] First, according to Figure 1 The flowchart shown illustrates that the first step in the preparation method is to obtain a preparation dataset of the hemodialysis concentrate. This dataset originates from historical experimental records or real-time monitoring data during the production process, including temperature sequences, pH sequences, and endotoxin content values. These data, collected by sensors or laboratory analytical instruments, form a complete preparation dataset, which serves as the basis for subsequent network optimization. After obtaining the preparation dataset, the data is extracted and processed sequentially. For example, in a certain experiment, suppose the preparation dataset contains 100 sets of data, each recording specific values ​​for temperature, pH, and endotoxin content. To further optimize process parameters, the membrane flux distribution sequence and adsorbent activity distribution sequence need to be calculated separately. The specific calculation method is as follows: temperature values ​​are extracted sequentially from the temperature sequence, and the membrane flux distribution under different temperature conditions is experimentally determined to obtain the membrane flux distribution sequence; simultaneously, pH values ​​are extracted sequentially from the pH sequence, and the adsorbent activity changes under different pH conditions are experimentally determined to obtain the adsorbent activity distribution sequence. Furthermore, the membrane flux ratio coefficient and adsorbent activity ratio coefficient need to be calibrated in advance for more accurate subsequent calculations. Assuming a membrane flux threshold of 50 units and a temperature threshold of 40℃, the membrane flux proportionality coefficient was experimentally calibrated to be 0.8. Simultaneously, assuming an adsorbent activity threshold of 60 units and a pH threshold of 7.2, the adsorbent activity proportionality coefficient is 0.9. These proportionality coefficients will be used for subsequent path optimization and node identification.

[0022] Next, the preparation data were optimized using the membrane flux distribution sequence and the adsorbent activity distribution sequence to obtain the optimized path. For example... Figure 1As shown, the generation of the optimized path depends on the setting of the initial dialysis path. The initial dialysis path can be a preset fluid flow path, such as a straight path from the inlet to the outlet. During the optimization process, the membrane flux value and adsorbent activity value are extracted sequentially from the membrane flux distribution sequence and the adsorbent activity distribution sequence, respectively. Assuming that the currently extracted membrane flux value is 45 units and the adsorbent activity value is 55 units, segmented control points are determined on the initial dialysis path based on the membrane flux value. The selection of segmented control points is based on the calculation of the fluid distribution angle, that is, by identifying the flow direction of the initial dialysis path and combining it with the adsorbent activity value to calculate the fluid distribution angle. The formula for calculating the fluid distribution angle is θ = arctan(ka*V), where ka is the adsorbent activity proportional coefficient and V is the adsorbent activity value. In this example, the fluid distribution angle θ = arctan(0.9*55) ≈ 1.52 radians. In addition, a radial offset angle φ needs to be randomly selected within a preset radial control range to increase the diversity of the path. Finally, an iterative dialysis path is constructed at the segmented control points based on the fluid distribution angle θ and the radial offset angle φ. If the membrane flux and adsorbent activity values ​​are not fully extracted, the initial dialysis path is updated using the iterative dialysis path, and the above steps are repeated; if the extraction is complete, an optimized path is obtained.

[0023] After optimizing the dialysis pathway once, the path nodes of the optimized pathway are constructed based on the endotoxin content index value, resulting in an iterative optimization network. It is then determined whether all preparation data in the preparation dataset has been extracted. If not, the process returns to the previous steps; if extraction is complete, the iterative optimization network is used as the first fully optimized network. The construction process of the first fully optimized network is as follows: Figure 2As shown, this is accomplished by the first fully optimized network generation module. In actual operation, assuming the target endotoxin content is 10 units, it is necessary to identify target path node pairs in the first fully optimized network based on this target value. The identification process of target path node pairs includes the following steps: First, identify the path end control points sequentially in the first fully optimized network. Path end control points refer to segment control points directly connected to path nodes. Assume the path node set at the current path end control point is {A,B,C}, and its corresponding end concentration range is [8,12] units. If the target endotoxin content value belongs to this range, then the path node set is used as the first node set, resulting in multiple sets of first node sets; otherwise, return to the above steps to continue identifying other path end control points. Next, identify the same-angle adjacent control node pairs sequentially in the first fully optimized network. Same-angle adjacent control node pairs refer to two path nodes with the same fluid distribution angle and adjacent path end control points. Assume the current same-angle adjacent control node pair is (A,B), and its same-angle adjacent concentration range is [9,11] units. If the target endotoxin content value falls within this range, then adjacent control node pairs at the same angle are used as second node pairs, resulting in multiple sets of second node pairs; otherwise, the above steps are repeated to continue identifying other adjacent control node pairs at the same angle. Finally, based on multiple sets of first node sets and multiple sets of second node pairs, target path node pairs are identified in the first fully optimized network using the target endotoxin content value. Specifically, first node sets are extracted sequentially from multiple sets of first node sets, first node pairs of target endotoxin content values ​​are extracted from the first node sets, and first adjacent concentration value pairs of the first node pairs are identified, resulting in multiple sets of first adjacent concentration value pairs; simultaneously, second node pairs are extracted sequentially from multiple sets of second node pairs, and second adjacent concentration value pairs of the second node pairs are identified, resulting in multiple sets of second adjacent concentration value pairs. The first adjacent concentration difference set and the second adjacent concentration difference set in the multiple sets of first adjacent concentration value pairs and the multiple sets of second adjacent concentration value pairs are evaluated using preset distance relationships, and the minimum concentration difference value is identified within them. Assuming the minimum concentration difference value is 0.5 units, the target adjacent concentration value pair corresponding to this minimum concentration difference value is identified, and the target path node pair corresponding to the target adjacent concentration value pair is further identified.

[0024] After identifying the target path node pairs, a second path expansion is performed on the first fully optimized network based on the target path node pairs to obtain the second fully optimized network. For example... Figure 2As shown, this process is completed by the second fully optimized network generation module. Specifically, it first obtains the node relationship attributes and overlapping control paths of the target path node pairs. If the node relationship attributes are preset co-located end nodes, then a dialysis experiment is performed according to the overlapping control paths and preset pH gradients to obtain a set of experimental concentration values. Assuming the current pH gradient is 0.1 units, the set of experimental concentration values ​​is {9.8, 10.0, 10.2} units. The experimental concentration values ​​are extracted sequentially from the set of experimental concentration values, the pH values ​​of the experimental concentration values ​​are identified, and the experimental distribution angle of the pH values ​​is calculated. The formula for calculating the experimental distribution angle is θ' = arctan(kp*C), where kp is the pH proportionality coefficient and C is the experimental concentration value. Assuming kp is 0.7, then for an experimental concentration value of 10.0 units, the experimental distribution angle θ' = arctan(0.7*10.0) ≈ 1.29 radians. The target end control points of the target path node pairs are identified, and an end dialysis path is constructed at the target end control points according to the experimental distribution angle. Experimental nodes are constructed based on the experimental concentration values, and these nodes are linked to the end-dialysis path to obtain the second fully optimized network. If the node relationship attribute is a preset angularly adjacent node, dialysis experiments are conducted based on the overlapping control path and the preset temperature gradient to obtain the experimental concentration value set. Assuming the current temperature gradient is 2℃, the experimental concentration value set is {9.5, 10.0, 10.5} units. Experimental concentration values ​​are extracted sequentially from the set, the temperature of each concentration value is identified, and the experimental membrane flux value for that temperature is calculated. The formula for calculating the experimental membrane flux value is F = kf * T, where kf is the membrane flux proportionality coefficient and T is the temperature value. Assuming kf is 0.8, for an experimental concentration value of 10.0 units, the experimental membrane flux value F = 0.8 * 40 = 32 units. The upper-level control point and the target fluid distribution angle of the target path node pair are identified, and the target end-control point is calculated based on the upper-level control point and the experimental membrane flux value. The terminal dialysis path is constructed based on the target terminal control point and the target fluid distribution angle, and the experimental nodes are linked to the terminal dialysis path to obtain the second fully optimized network.

[0025] After constructing the second fully optimized network, target path nodes are identified on the second fully optimized network based on the target endotoxin content value. This process is as follows: Figure 2 As shown, this is accomplished by the low endotoxin hemodialysis concentrate preparation module. In actual operation, the experimental nodes in the second fully optimized network are extracted sequentially, and the experimental concentration values ​​corresponding to the experimental nodes are obtained. The concentration difference between the experimental concentration value and the target endotoxin content value is calculated to obtain a concentration difference set. Assuming the current experimental concentration value is 10.0 units and the target endotoxin content value is 10.0 units, the concentration difference is 0 units. The minimum concentration difference is extracted from the concentration difference set, and the target path node corresponding to the minimum concentration difference is identified. Assuming the minimum concentration difference is 0 units, the corresponding target path node is the final target path node.

[0026] Finally, the temperature and pH control sequences corresponding to the target path nodes were identified, and low-endotoxin hemodialysis concentrate was prepared based on these sequences. Specifically, the target membrane flux distribution sequence and target adsorbent activity distribution sequence of the target path nodes were extracted in the second fully optimized network, and their corresponding target temperature and pH sequences were identified. Assuming the target temperature sequence is {38, 40, 42}℃ and the target pH sequence is {7.0, 7.2, 7.4}, a temperature and pH control sequence was constructed based on these sequences. In the actual preparation process, the temperature and pH were gradually adjusted according to the control sequences to ensure that the endotoxin content reached the target value. For example, the temperature was first adjusted to 38℃ and the pH to 7.0 for preliminary dialysis; then the temperature was adjusted to 40℃ and the pH to 7.2 for intermediate-stage dialysis; finally, the temperature was adjusted to 42℃ and the pH to 7.4 for final dialysis. Through these steps, the endotoxin removal efficiency can be significantly improved, while ensuring precise control of process parameters and batch stability of the product.

[0027] In summary, this invention achieves efficient preparation of low-endotoxin hemodialysis concentrate by constructing a first fully optimized network and a second fully optimized network, combined with target endotoxin content values ​​for path optimization and node identification. This method not only reduces experimental workload but also significantly improves the accuracy and stability of the preparation process, providing reliable technical support for the production of hemodialysis concentrate.

Claims

1. A method for preparing a low-endotoxin hemodialysis concentrate, characterized in that, The method includes: Obtain a dataset for the preparation of hemodialysis concentrate; Preparation data are extracted sequentially from the preparation dataset. Temperature sequence, pH value sequence and endotoxin content index value are extracted from the preparation data. Membrane flux distribution sequence and adsorbent activity distribution sequence are calculated based on the temperature sequence and pH value sequence, respectively. Based on the membrane flux distribution sequence and the adsorbent activity distribution sequence, the preparation data is optimized once to obtain an optimized path. The path nodes of the optimized path are constructed based on the endotoxin content index value to obtain an iterative optimization network. Determine whether all the data to be prepared in the prepared dataset has been extracted; If the preparation data in the preparation dataset is not completely extracted, the process returns to the steps described above for sequentially extracting the preparation data from the preparation dataset; if the preparation data in the preparation dataset is completely extracted, the iterative optimization network is used as the first fully optimized network. Obtain the target endotoxin content value, and identify the target path node pair in the first fully optimized network based on the target endotoxin content value; Based on the target path nodes, a second fully optimized network is obtained by performing secondary path expansion on the first fully optimized network. Based on the target endotoxin content value, target path nodes are identified on the second fully optimized network, and temperature and pH control sequences corresponding to the target path nodes are identified. Low endotoxin hemodialysis concentrate is prepared according to the temperature and pH control sequences.

2. The method for preparing low endotoxin hemodialysis concentrate as described in claim 1, characterized in that, The calculation of the membrane flux distribution sequence and the adsorbent activity distribution sequence based on the temperature sequence and pH sequence, respectively, includes: Temperature values ​​are extracted sequentially from the temperature sequence, and the distribution of membrane flux is experimentally determined based on the temperature values ​​to obtain the membrane flux distribution sequence; pH values ​​are extracted sequentially from the pH value sequence, and the changes in adsorbent activity are experimentally determined based on the pH values ​​to obtain the adsorbent activity distribution sequence.

3. The method for preparing low endotoxin hemodialysis concentrate as described in claim 2, characterized in that, Before calculating the membrane flux distribution sequence and adsorbent activity distribution sequence based on the temperature sequence and pH sequence respectively, the method further includes: Obtain the membrane flux threshold and temperature threshold; calibrate the membrane flux ratio coefficient experimentally based on the membrane flux threshold and temperature threshold; Obtain the adsorbent activity threshold and pH threshold; calibrate the adsorbent activity ratio coefficient through experiments based on the adsorbent activity threshold and pH threshold.

4. The method for preparing low endotoxin hemodialysis concentrate as described in claim 1, characterized in that, The preparation data is optimized using a single dialysis path based on the membrane flux distribution sequence and the adsorbent activity distribution sequence to obtain an optimized path, including: Obtain the initial dialysis path; Membrane flux values ​​and adsorbent activity values ​​were extracted sequentially from the membrane flux distribution sequence and the adsorbent activity distribution sequence, respectively. Based on the membrane flux value, segmented control points are determined on the initial dialysis path; Identify the flow direction of the initial dialysis path, and calculate the fluid distribution angle based on the flow direction and the adsorbent activity value; Randomly select a radial offset angle within the preset radial adjustment range; Based on the fluid distribution angle and radial offset angle, an iterative dialysis path is constructed at the segmented control points; Determine whether the membrane flux and adsorbent activity values ​​have been completely extracted; If the membrane flux value and adsorbent activity value are not completely extracted, the initial dialysis path is updated using the iterative dialysis path, and the steps of extracting the membrane flux value and adsorbent activity value sequentially from the membrane flux distribution sequence and adsorbent activity distribution sequence are returned; if the membrane flux value and adsorbent activity value are completely extracted, the optimized path is obtained.

5. The method for preparing low endotoxin hemodialysis concentrate as described in claim 1, characterized in that, The step of identifying target path node pairs in the first fully optimized network based on the target endotoxin content value includes: In the first fully optimized network, path-end control points are identified sequentially, wherein the path-end control points refer to segment control points that are directly connected to the path nodes. Identify the set of path nodes at the end control point of the path, and identify the isotopic end concentration range of the path node set; Determine whether the target endotoxin content value belongs to the isotope terminal concentration range; If the target endotoxin content value does not belong to the isotopic terminal concentration range, then return to the steps of sequentially identifying path terminal control points in the first fully optimized network; if the target endotoxin content value belongs to the isotopic terminal concentration range, then use the path node set as the first node set to obtain multiple sets of first node sets. In the first fully optimized network, pairs of adjacent control nodes with the same angle are identified sequentially. The pairs of adjacent control nodes with the same angle refer to two path nodes with the same fluid distribution angle and adjacent control points at the end of their respective paths. Identify the concentration ranges of adjacent control nodes at the same angle; Determine whether the target endotoxin content value belongs to the adjacent concentration range at the same angle; If the target endotoxin content value does not belong to the same-angle adjacent concentration range, then return to the above steps of sequentially identifying the same-angle adjacent control node pairs in the first fully optimized network; if the target endotoxin content value belongs to the same-angle adjacent concentration range, then take the same-angle adjacent control node pairs as the second node pairs, and obtain multiple sets of second node pairs. Based on multiple sets of first nodes and multiple sets of second nodes, the target path node pairs are identified in the first fully optimized network using the target endotoxin content value.

6. The method for preparing low endotoxin hemodialysis concentrate as described in claim 5, characterized in that, The step of identifying target path node pairs in the first fully optimized network based on multiple sets of first nodes and multiple sets of second nodes includes: First node sets are extracted sequentially from the multiple sets of first node sets. First node pairs of the target endotoxin content value are extracted from the first node sets. First adjacent concentration value pairs of the first node pairs are identified to obtain multiple sets of first adjacent concentration value pairs. Extract the second node pairs sequentially from the multiple sets of second node pairs, identify the second adjacent concentration value pairs of the second node pairs, and obtain multiple sets of second adjacent concentration value pairs; The first adjacent concentration difference set and the second adjacent concentration difference set in the multiple sets of first adjacent concentration value pairs and multiple sets of second adjacent concentration value pairs are evaluated respectively using preset distance relationships; Identify the minimum concentration difference value in the first set of adjacent concentration differences and the second set of adjacent concentration differences; Identify the target adjacent concentration value pair corresponding to the minimum concentration difference, and identify the target path node pair corresponding to the target adjacent concentration value pair.

7. The method for preparing low endotoxin hemodialysis concentrate as described in claim 1, characterized in that, The step of performing secondary path expansion on the first fully optimized network based on the target path nodes to obtain a second fully optimized network includes: Obtain the node relationship attributes and overlapping control paths of the target path node pairs; If the node relationship attribute is a preset co-terminal node, then a dialysis test is performed according to the overlapping control path and the preset pH gradient to obtain a set of test concentration values. The test concentration values ​​are extracted sequentially from the set of test concentration values, the pH value of the test concentration values ​​is identified, and the test distribution angle of the pH value is calculated. Identify the target end control point of the target path node pair, and construct the end dialysis path at the target end control point according to the experimental distribution angle; Based on the experimental concentration values, experimental nodes are constructed, and the experimental nodes are linked to the end dialysis path to obtain the second fully optimized network; If the node relationship attribute is a preset adjacent node at the same angle, then a dialysis test is performed according to the overlapping control path and the preset temperature gradient to obtain a set of test concentration values. The test concentration values ​​are extracted sequentially from the set of test concentration values, the temperature of the test concentration values ​​is identified, and the test membrane flux value at the temperature is calculated. Identify the upper-level control point and target fluid distribution angle of the target path node pair, and calculate the target end control point based on the upper-level control point and the experimental membrane flux value; Construct the end-dialysis path based on the target end-control point and the target fluid distribution angle; Based on the experimental concentration values, experimental nodes are constructed, and the experimental nodes are linked to the end dialysis path to obtain the second fully optimized network.

8. The method for preparing low endotoxin hemodialysis concentrate as described in claim 1, characterized in that, The step of identifying target path nodes on the second fully optimized network based on the target endotoxin content value includes: In the second fully optimized network, test nodes are extracted sequentially, and the test concentration values ​​corresponding to the test nodes are obtained. Calculate the concentration difference between the experimental concentration value and the target endotoxin content value to obtain a set of concentration difference values; Extract the minimum concentration difference from the set of concentration differences, and identify the target path node corresponding to the minimum concentration difference.

9. The method for preparing low endotoxin hemodialysis concentrate as described in claim 1, characterized in that, The temperature and pH regulation sequence corresponding to the identified target path node includes: Extract the target membrane flux distribution sequence and the target adsorbent activity distribution sequence of the target path nodes in the second fully optimized network; Identify the target temperature sequence and target pH value sequence corresponding to the target membrane flux distribution sequence and the target adsorbent activity distribution sequence; A temperature and pH regulation sequence is constructed based on the target temperature sequence and the target pH sequence.

10. A system for preparing low endotoxin hemodialysis concentrate, characterized in that, The system includes: The first fully optimized network generation module (1) is used to obtain a preparation dataset of hemodialysis concentrate; sequentially extract preparation data from the preparation dataset, extract temperature sequence, pH value sequence and endotoxin content index value from the preparation data, calculate membrane flux distribution sequence and adsorbent activity distribution sequence based on the temperature sequence and pH value sequence respectively; optimize the dialysis path of the preparation data once based on the membrane flux distribution sequence and adsorbent activity distribution sequence to obtain an optimized path, construct the path nodes of the optimized path based on the endotoxin content index value to obtain an iterative optimization network; determine whether the preparation data in the preparation dataset has been completely extracted; if the preparation data in the preparation dataset has not been completely extracted, return to the above steps of sequentially extracting preparation data in the preparation dataset; if the preparation data in the preparation dataset has been completely extracted, use the iterative optimization network as the first fully optimized network. The target path node pair identification module (2) is used to obtain the target endotoxin content value and identify the target path node pair in the first fully optimized network based on the target endotoxin content value. The second fully optimized network generation module (3) is used to perform secondary path expansion on the first fully optimized network based on the target path nodes to obtain the second fully optimized network. The low endotoxin hemodialysis concentrate preparation module (4) is used to identify target path nodes on the second fully optimized network according to the target endotoxin content value, identify the temperature and pH value control sequence corresponding to the target path node, and prepare low endotoxin hemodialysis concentrate according to the temperature and pH value control sequence.

Images