Dialysis mode decision-making auxiliary system for patient with end-stage kidney disease
Through multi-dimensional data collection, geographic information combined with adaptive weight allocation and balanced optimization methods, the problem of insufficient data support for dialysis method decision-making in patients with end-stage renal disease is solved, and precise decision-making assistance and user experience improvement is achieved.
Patent Information
- Application Number
- CN202510238654.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-03
- Publication Date
- 2025-07-25
AI Technical Summary
The lack of intuitive and accurate data support in the prior art makes it difficult to achieve accurate decision-making assistance for dialysis methods for patients with end-stage renal disease.
Physiological and non-physiological indicators are obtained through the multi-dimensional data acquisition module, and the accessibility index is generated by the integrated geographic information. The feature coupling module is used for adaptive weight allocation, and the fitness analysis unit performs balanced optimization and finally interactive display is performed by the visual terminal.
It provides visual data support to help patients make dialysis methods that are more in line with their own reality, and improves the accuracy and user experience of decisions.
Smart Images
Figure CN120376094A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of data processing, and in particular to a dialysis mode decision-making support system for patients with end-stage renal disease. Background Art
[0002] The choice of dialysis method not only involves the patient's physiological indicators, but is also affected by a variety of non-physiological factors, such as the geographical location of the dialysis center, transportation convenience, etc. At present, when patients choose dialysis methods, they mostly rely on the doctor's experience and the patient's own subjective wishes. They often lack comprehensive and accurate data support, which cannot help patients quickly understand and make independent decisions, and it is difficult to achieve accurate decision-making assistance. Summary of the invention
[0003] The present invention provides a dialysis mode decision support system for end-stage renal disease patients to solve the technical problems in the prior art of lack of intuitive and accurate data support and difficulty in achieving accurate decision support, and achieves the technical effect of providing visual data support and accurate decision support.
[0004] The dialysis method decision support system for end-stage renal disease patients provided by the present invention comprises:
[0005] The multi-dimensional data collection module is used to collect patient data and establish a patient data set, wherein the patient data set includes a physiological indicator set and a non-physiological indicator set.
[0006] An integrated geographic information module is used to generate a dialysis center accessibility index, which includes real-time transportation time cost and public transportation coverage density.
[0007] The feature coupling module is used to map the patient data set and the accessibility index into a unified vector space and then perform adaptive weight allocation.
[0008] The fitness analysis unit is used to configure a balance fitness function based on the adaptive weight distribution result, perform balance optimization using the balance fitness function and vector data mapped to the unified vector space, and establish a balance optimization result.
[0009] The visualization terminal is used to interactively display the balance optimization result.
[0010] In a feasible implementation, the multi-dimensional data acquisition module includes:
[0011] The physiological data acquisition terminal is used to collect the patient's physiological monitoring data, which includes residual renal function indicators, blood pressure fluctuation coefficients, and body surface area parameters, and synchronize the physiological monitoring data to the multi-dimensional data acquisition module via the Internet.
[0012] An interactive acquisition terminal is used to perform medical insurance data docking according to the patient's information, obtain the annual dialysis limit and reimbursement ratio, and synchronously obtain the patient's family economic status. After establishing non-physiological monitoring data, it is synchronized to the multi-dimensional data acquisition module.
[0013] In a feasible implementation, the feature coupling module includes:
[0014] The first coupling channel is used to perform patient status time series analysis on the physiological index set in the patient dataset and establish a time series feature set.
[0015] The second coupling channel is used to perform keyword joint extraction on the non-physiological index set in the patient dataset through a knowledge language model and establish a keyword embedding vector.
[0016] The normalization channel is used to receive the time series feature set, the keyword embedding vector, and the reachability index, perform normalization processing within the same scale range, and then perform feature splicing.
[0017] The unified space mapping channel is used to perform vector space mapping of the feature splicing result through a multi-layer perceptron.
[0018] In a feasible implementation, in the feature coupling module, performing adaptive weight allocation includes:
[0019] The similar label matching unit is used to obtain the patient label of the patient, perform similar label matching based on the patient label, and establish a similar label matching result.
[0020] The feature initial weight establishment unit is used to establish the feature initial weight by using the similar label matching result.
[0021] The feature dot product attention score calculation unit is used to calculate the feature dot product attention score of the unified vector space mapping result.
[0022] The adaptive weight allocation unit is used to scale the result of the feature dot product attention score calculation, perform normalization processing, and perform proportional weighting according to the normalization processing result and the patient's selection preference to complete the adaptive weight allocation.
[0023] In a feasible implementation, the fitness analysis unit includes:
[0024] The balanced fitness function construction unit is used to construct a balanced fitness function. The balanced fitness function is constructed based on the adaptive weight allocation result, and the evaluation features of the balanced fitness function include economic burden feature, effect feature, and convenience feature.
[0025] An optimization unit, configured to configure an initial solution set according to vector data mapped to a unified vector space, evaluate the fitness of the initial solution set by using the balanced fitness function to establish a fitness evaluation result, iteratively update the initial solution set by using the fitness evaluation result, and establish a balanced optimization result according to the iterative update result.
[0026] In a feasible implementation manner, in the optimization unit, the iterative update of the initial solution set by using the fitness evaluation result includes:
[0027] A target solution determination and follower solution marking unit, configured to determine a target solution in the initial solution set according to the fitness evaluation result, and mark non-target solutions as follower solutions.
[0028] An optimization bias coefficient generation unit, configured to perform imbalance calculation on all follower solutions to generate an optimization bias coefficient.
[0029] A single-iteration optimization unit, configured to perform single-iteration optimization of the initial solution set according to the optimization bias coefficient and the target solution.
[0030] An iterative update completion unit, configured to complete iterative update according to multi-round iterative optimization results.
[0031] In a feasible implementation manner, the visualization terminal further includes:
[0032] A user interaction module, configured to read reading preference data of a patient, establish a reading scoring sample set, perform display matching analysis on the reading scoring sample set, and establish a patient display preference.
[0033] A display module, configured to perform interactive display after sorting the balanced optimization result through the patient display preference.
[0034] In a feasible implementation manner, the end-stage renal disease patient dialysis modality decision-making assistance system further includes:
[0035] A self-optimization feedback module, configured to receive feedback data of a patient and associated users, perform co-directional optimization analysis based on the feedback data to establish feedback optimization data, and update the system through the feedback optimization data.
[0036] The dialysis method decision - making assistance system for end - stage renal disease patients disclosed by the present invention performs data collection of patients through a multi - dimensional data collection module, establishes a patient data set, and the patient data set includes a physiological index set and a non - physiological index set; an integrated geographic information module generates a dialysis center accessibility index, and the accessibility index includes real - time traffic time cost and public transportation coverage density; after the feature coupling module maps the patient data set and the accessibility index to a unified vector space, it performs adaptive weight allocation; the fitness analysis unit configures a balanced fitness function based on the adaptive weight allocation result, uses the balanced fitness function and the vector data mapped to the unified vector space for balanced optimization, and establishes a balanced optimization result; the visualization terminal interactively displays the balanced optimization result, solves the technical problems of lacking intuitive and accurate data support and being difficult to achieve precise decision - making assistance, and realizes the technical effect of providing visual data support and precise decision - making assistance. BRIEF DESCRIPTION OF THE DRAWINGS
[0037] Figure 1 It is a schematic structural diagram of the dialysis method decision - making assistance system for end - stage renal disease patients of the present invention;
[0038] Figure 2 It is a schematic flow diagram of the feature coupling module in the dialysis method decision - making assistance system for end - stage renal disease patients of the present invention performing adaptive weight allocation.
[0039] Description of reference numerals: multi - dimensional data collection module 11, integrated geographic information module 12, feature coupling module 13, similar label matching unit 131, feature initial weight establishment unit 132, feature dot - product attention score calculation unit 133, adaptive weight allocation unit 134, fitness analysis unit 14, visualization terminal 15. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0040] The following will combine the description of the drawings in the specification and specific embodiments to elaborate on the above - mentioned technical solutions in detail to better understand the above - mentioned technical solutions. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments for explaining the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention. In addition, it should be noted that for the sake of description, only parts related to the present invention are shown in the drawings rather than all.
[0041] Embodiment Figure 1 It is a schematic structural diagram of the dialysis method decision - making assistance system for end - stage renal disease patients of the present invention, wherein the dialysis method decision - making assistance system for end - stage renal disease patients includes:
[0042] The multi-dimensional data acquisition module 11 is used to perform data acquisition of patients and establish a patient data set, and the patient data set includes a physiological index set and a non-physiological index set.
[0043] Specifically, first, the multi-dimensional data acquisition module 11 acquires the physiological and non-physiological indexes of the patient to obtain the patient data set, thereby providing comprehensive basic input information for subsequent data processing; among them, the physiological index set includes data related to the patient's physical health, such as renal function indexes, blood pressure, etc.; the non-physiological index set involves other information related to the patient's dialysis decision-making, such as economic status, medical insurance situation, patient location, etc.
[0044] In some embodiments, the multi-dimensional data acquisition module 11 includes:
[0045] A physiological data acquisition terminal for acquiring the patient's physiological monitoring data, and the physiological monitoring data includes residual renal function indexes, blood pressure fluctuation coefficients, body surface area parameters, and synchronizes the physiological monitoring data to the multi-dimensional data acquisition module 11 through the Internet; an interactive acquisition terminal for docking medical insurance data according to the patient's information, obtaining the annual dialysis limit and reimbursement ratio, and synchronously obtaining the patient's family economic status, and after establishing non-physiological monitoring data, synchronizing it to the multi-dimensional data acquisition module 11.
[0046] Specifically, the physiological data acquisition terminal obtains physiological monitoring data by connecting to a medical monitoring device or an electronic health record system, and the physiological monitoring data includes residual renal function indexes (reflecting the remaining function of the kidneys), blood pressure fluctuation coefficients (reflecting blood pressure stability), and body surface area parameters (used to calculate drug doses or dialysis doses).
[0047] Specifically, the interactive acquisition terminal docks with the medical insurance-related database according to the patient's identity information to obtain non-physiological monitoring data, including the annual dialysis limit and reimbursement ratio, and the patient's economic situation, and this acquisition process meets the management and use requirements and privacy requirements of medical insurance data.
[0048] By simultaneously acquiring the patient's physiological and non-physiological data, the subsequent decision-making assistance and the generated auxiliary information can more accurately reflect the actual situation of the patient, provide a more comprehensive and scientific basis for the user to select a dialysis method, and help the patient make a more practical choice that suits him / herself.
[0049] The integrated geographic information module 12 is used to generate a dialysis center accessibility index, and the accessibility index includes real-time traffic time cost and public transportation coverage density.
[0050] Specifically, the dialysis center accessibility index is used to measure the real-time path cost from the patient's location to the target dialysis center, including the traffic time cost (real-time traffic time cost) and the traffic convenience cost (public transportation coverage density).
[0051] Specifically, the real-time traffic time cost can be characterized as the shortest time required to reach the target dialysis center from the patient's current location; the public transportation coverage density refers to the number of lines and the distribution density of stations of public transportation (such as buses and subways) around the dialysis center, which reflects the convenience of the patient using public transportation to go to the dialysis center and can indirectly reflect the economic cost of the patient using public transportation to go to the dialysis center. In other words, the larger the number of lines and the distribution density of stations of public transportation around the dialysis center, the lower the probability of giving up public transportation and choosing other transportation methods with higher costs, and the lower the expected cost can be considered.
[0052] Exemplarily, to generate the dialysis center accessibility index, first, obtain the real-time location of the patient, and extract multiple dialysis centers within a preset radius of the real-time location, or the first n dialysis centers closest to the real-time location of the patient, and based on the geographic information system, calculate the real-time traffic time costs to the above-mentioned multiple dialysis centers respectively in combination with real-time traffic data (such as road condition information); then, traverse the obtained multiple dialysis centers, analyze the public transportation lines and station distributions around the dialysis centers, and calculate the corresponding public transportation coverage density; finally, based on the multiple dialysis centers as the association basis, associate and store the above-mentioned real-time traffic time costs and public transportation coverage density, and after dimensionless processing, output them as the accessibility index.
[0053] Through the above process, the accessibility index of the dialysis center generated considering the real-time traffic time cost and the public transportation coverage density can more accurately reflect the actual difficulty level of the patient going to the dialysis center, thereby providing more practically meaningful and reference-valued decision-making assistance information for the patient.
[0054] The feature coupling module 13 is used to perform adaptive weight allocation after mapping the patient data set and the accessibility index to a unified vector space.
[0055] Specifically, through the feature coupling module 13, the accessibility index and the patient data set are converted into vector forms in the same mathematical space for subsequent processing and analysis to obtain uniformly quantified adaptive weights; among them, adaptive weight allocation is a process of dynamically adjusting the weights of each feature according to the importance and relevance of the data, so as to more accurately reflect its contribution to the auxiliary information.
[0056] In some embodiments, the feature coupling module 13 includes:
[0057] The first coupling channel is used to perform patient state time series analysis on the set of physiological indicators in the patient dataset and establish a time series feature set; the second coupling channel is used to perform keyword joint extraction on the set of non-physiological indicators in the patient dataset through a knowledge language model and establish keyword embedding vectors; the normalization channel is used to receive the time series feature set, the keyword embedding vectors, and the reachability index, perform normalization processing within the same scale range, and then perform feature splicing; the unified space mapping channel is used to perform vector space mapping on the feature splicing result through a multi-layer perceptron.
[0058] Specifically, the time series feature set is a set of features obtained by performing time series analysis on the physiological data (such as heart rate, blood pressure, etc.) of the patient, and is used to reflect the trend of the patient's health status changing over time; the keyword embedding vector is a vector representation of the keywords extracted by performing text processing on the non-physiological data of the patient through a knowledge language model. In other words, through the keyword embedding vector, the text data is converted into a numerical form that can be input into a machine learning model.
[0059] Specifically, first, a sequence model (such as LSTM or GRU) is used to perform time series analysis on the physiological indicator data of the patient to generate a time series feature set; then, a knowledge language model (such as BERT, GPT, etc.) is used to perform natural language processing on the non-physiological data of the patient to extract keywords and generate embedding vectors for subsequent understanding and processing; next, data from different sources such as the physiological time series feature set, the non-physiological keyword embedding vectors, and the reachability index are normalized. Exemplarily, z-score normalization or Min-Max normalization is used to scale all data to a unified scale range, and the normalized features are spliced to form a comprehensive feature vector that integrates physiological, non-physiological, and other relevant information; finally, through the hierarchical feature processing ability of a multi-layer perceptron (MLP), spatial mapping is performed on the spliced features to map multiple spliced feature vectors to a unified vector space, thereby enhancing the expression ability of multiple features and avoiding weight allocation caused by different data types.
[0060] In some implementation manners, as Figure 2 shown, in the feature coupling module 13, performing adaptive weight allocation includes:
[0061] A similar label matching unit 131 is used to obtain the patient label of a patient, perform similar label matching based on the patient label, and establish a similar label matching result; a feature initial weight establishing unit 132 is used to establish a feature initial weight by using the similar label matching result; a feature dot product attention score calculating unit 133 is used to calculate the feature dot product attention score of the unified vector space mapping result; an adaptive weight allocation unit 134 is used to scale the result of the feature dot product attention score calculation, perform normalization processing, and complete adaptive weight allocation after proportional weighting according to the normalization processing result and the patient selection preference.
[0062] Specifically, the patient label is the identification information or classification label of the patient, which describes the patient's health status, disease type, treatment plan or other relevant features. Through this patient label, an object similar to the patient's condition can be accurately matched, and the corresponding weight is used as the feature initial weight. In other words, the feature initial weight reflects the importance of different features in the patient group similar to the patient.
[0063] Specifically, first, obtain the patient label of the patient, and perform similarity measurement on the patient based on the patient label (such as cosine similarity, Euclidean distance, etc.), match the patients with higher label similarity, so as to find the patient group with the same or similar health status or disease type, and obtain the corresponding feature weight of this group as the feature initial weight. Among them, the above feature initial weight is the central value (such as the average value) of the feature weights corresponding to the similar matching results; then, calculate the dot product of the unified vector space mapping result to obtain the feature dot product attention score, which is used to reflect the similarity or correlation between features; further, scale the calculation of the feature dot product attention score to ensure the consistency of the numerical range, so that different features have the same influence during the calculation, and are more suitable for subsequent weighted calculation; finally, weight the normalized features according to the patient's personal selection preference. For example, if the patient prefers certain features (such as the closest distance or the lowest cost) during the health management process, then increase the weights of these features correspondingly to complete the adaptive weight allocation.
[0064] Through the adaptive weight allocation, the importance of each feature is automatically adjusted, so as to ensure that when facing different patients, their health status and needs can be accurately reflected, thereby improving the reference value of the final output auxiliary information.
[0065] A fitness analysis unit 14 is used to configure a balanced fitness function based on the adaptive weight allocation result, perform balanced optimization by using the balanced fitness function and the vector data mapped to the unified vector space, and establish a balanced optimization result.
[0066] Specifically, the balanced fitness function is initialized and configured based on the adaptive weight allocation result to adapt to patients in different states and with personalized needs. Among them, by way of example, the balanced fitness function is a mathematical function that combines multiple performance indicators and is used to quantitatively evaluate the advantages and disadvantages of different dialysis methods (different dialysis centers). Among them, each performance indicator corresponds to an adaptive weight, thereby adaptively ensuring the balance between various objectives (such as the balance between accuracy, efficiency, stability, etc.). Among them, different dialysis schemes are represented by multiple vector data mapped to a unified vector space.
[0067] Specifically, the optimization process is evaluated and guided by the balanced fitness function, including mapping each candidate solution (dialysis method) into vector data in a unified vector space, using the balanced fitness function as the cost function, and combining optimization algorithms (such as genetic algorithms, particle swarm optimization, or gradient descent, etc.) to perform iterative balanced optimization until the optimal solution is found.
[0068] In some embodiments, the fitness analysis unit 14 includes:
[0069] A balanced fitness function construction unit for constructing a balanced fitness function, which is constructed based on the adaptive weight allocation result, and the evaluation features of the balanced fitness function include economic burden features, effect features, and convenience features; an optimization unit for configuring an initial solution set according to the vector data mapped to the unified vector space, using the balanced fitness function to evaluate the fitness of the initial solution set, establishing a fitness evaluation result, using the fitness evaluation result to iteratively update the initial solution set, and establishing a balanced optimization result according to the iterative update result.
[0070] Specifically, the balanced fitness function balances and optimizes among multiple objectives by measuring the contributions of different solutions (i.e., feature combinations or decision configurations) to multiple objectives (such as economic burden, effect, convenience, etc.). Among them, the economic burden feature is used to measure the economic cost during the implementation of the scheme, including medical expenses, time spent, path costs, etc.; the effect feature is used to measure the dialysis effectiveness and effect corresponding to the scheme, that is, to focus on the actual efficacy of the scheme; the convenience feature is used to measure the convenience degree during the implementation of the scheme, such as travel convenience, time consumption, user experience, etc.
[0071] Specifically, in the optimization unit, first, according to the data mapped to the unified vector space, an initial solution set is configured. Multiple solutions in this initial solution set correspond to various dialysis methods, providing a starting point for optimization. Then, each solution in the initial solution set is respectively evaluated for fitness using the balanced fitness function. Among them, the fitness score of each solution reflects its comprehensive performance in terms of economic burden, effect, and convenience. Furthermore, according to the fitness evaluation results, each solution in the solution set is iteratively updated to approach the optimal solution, and finally, a balanced optimization result is obtained. This result represents the best balance among multiple objectives, that is, the optimal solution found between economic burden, effect, and convenience.
[0072] In some implementation manners, in the optimization unit, using the fitness evaluation results to perform iterative update of the initial solution set includes:
[0073] A target solution determination and follower solution marking unit is used to determine the target solution in the initial solution set according to the fitness evaluation results and mark non-target solutions as follower solutions; an optimization bias coefficient generation unit is used to perform imbalance calculation on all follower solutions to generate an optimization bias coefficient; a one-time iterative optimization unit is used to perform one-time iterative optimization of the initial solution set according to the optimization bias coefficient and the target solution; an iterative update completion unit is used to complete iterative update according to the results of multiple rounds of iterative optimization.
[0074] Specifically, according to the fitness evaluation results, the initial solution corresponding to the optimal fitness is selected as the target solution, and other solutions are marked as follower solutions. Among them, compared with the target solution, follower solutions are those that perform poorly in the current iteration and do not meet the main objectives.
[0075] Specifically, according to the performance differences of each follower solution in the solution set based on multiple indicators (such as economic burden, effect, and convenience, etc.) in the balanced fitness function and the target solution, the imbalance situation of each solution in these indicators is judged, so as to identify which indicators perform poorly in the current solution and which aspects need to be adjusted and optimized, that is, to find out the direction of imbalance. Then, an optimization bias coefficient is generated according to the degree of imbalance, indicating in which directions the optimization process needs to further strengthen the adjustment. Through imbalance calculation and the optimization bias coefficient, it helps the optimization process to more precisely adjust the direction of the follower solutions, and then more efficiently find the balanced solution that meets multiple objectives.
[0076] Furthermore, various indicators of the follower solutions are adjusted according to the optimization bias coefficient, gradually guiding them to near the target solution to perform one-time iterative optimization of the initial solution set. Then, the above-mentioned fitness evaluation and the marking process of the target solution and follower solutions are repeated, and multiple rounds of iterative optimization are performed to gradually update the initial solution set until the solutions in the solution set meet the preset optimization objectives.
[0077] Through the above process, the convergence speed, accuracy, and intelligence of the optimization are improved. For example, for a solution in the initial solution set with a heavy economic burden but good dialysis effect, the deficiency in its economic burden can be discovered through imbalance calculation and optimized and adjusted in subsequent iterations. Eventually, an economical and effective dialysis solution can be found, thereby better meeting the personalized needs of patients and improving the practicality of the generated auxiliary information and the user experience.
[0078] The visualization terminal 15 is used to interactively display the balance optimization result.
[0079] Exemplarily, after the optimization process is completed, all solutions in the balance optimization result are sorted according to the fitness, and the top k optimal solutions are selected, where k is an adjustable parameter used to control the number of displayed results, ensuring that the visualization terminal 15 only displays the most relevant and valuable solutions, avoiding information overload, and enabling users to focus on the most representative results.
[0080] Furthermore, the selected top k optimized solutions are sorted and displayed in a certain order. Exemplarily, the display can be in the form of charts, tables, graphs, etc. for users to understand and compare. In addition, users can view the detailed content of each solution through interactive operations, such as the specific values of indicators such as economic burden, effect, and convenience.
[0081] By serially displaying the optimization result, the visualization terminal 15 provides an intuitive interface that enables users to quickly understand and compare the optimization result, improving the readability and logic of the finally provided auxiliary information, thereby further enhancing the user experience.
[0082] In some embodiments, the visualization terminal 15 further includes:
[0083] A user interaction module, which is used to read the reading preference data of the patient, establish a reading scoring sample set, perform display matching analysis on the reading scoring sample set, and establish the patient's display preference; a display module, which is used to interactively display the balance optimization result after sorting it according to the patient's display preference.
[0084] Specifically, the reading preference data is data generated when the patient interacts with the visualization terminal 15, including the display format, font size, content display method, display hierarchy of information, etc. that the patient likes. Through the reading preference data, the interface can be personalized adjusted to enhance the user experience of the patient.
[0085] Specifically, the reading score sample set is a sample set established by collecting and recording patients' feedback and reactions (such as click frequency, dwell time, scores, etc.) in different content display scenarios; by analyzing the collected patients' reading score samples, an exclusive display preference can be established for each patient, wherein the patient display preference includes the order of content arrangement, display format, font, color, graphic style, etc. Through the patient display preference, the patient can get a more intuitive and comfortable information presentation on the interface.
[0086] Furthermore, the dialysis method decision support system for end-stage renal disease patients also includes:
[0087] The self-optimization feedback module is used to receive feedback data from patients and associated users, and perform same-direction optimization analysis based on the feedback data, establish feedback optimization data, and update the system through the feedback optimization data.
[0088] Specifically, the feedback data from patients and associated users refers to the feedback from patients or other users related to the patients (such as doctors, family members, etc.) on the system usage experience, improvement suggestions, information satisfaction, etc., which include, for example, evaluation of the system interface, suggestions for dialysis plans, feedback on the accuracy of auxiliary information, etc.
[0089] Specifically, the same-direction optimization analysis is optimized based on the positive feedback from users (such as high satisfaction and satisfied needs) to improve the overall effect of the system and ensure that the direction of improvement meets the needs of patients and users. For example, if patients generally report that a certain function is very practical, then the function can be further strengthened; if a user points out that the interface design is inconvenient, the direction of interface optimization will be analyzed. Through the same-direction optimization analysis, it can be ensured that the optimization direction is consistent with the needs of users, avoid deviation from important needs in actual use, and continuously improve auxiliary performance and user satisfaction in long-term use.
[0090] In summary, the dialysis method decision support system for end-stage renal disease patients provided by the present invention has the following technical effects:
[0091] The multi-dimensional data collection module is used to collect patient data and establish a patient data set, which includes a physiological indicator set and a non-physiological indicator set; the integrated geographic information module generates a dialysis center accessibility index, which includes real-time transportation time cost and public transportation coverage density; the feature coupling module maps the patient data set and accessibility index to a unified vector space, and then performs adaptive weight allocation; the fitness analysis unit configures a balanced fitness function based on the adaptive weight allocation result, and uses the balanced fitness function and the vector data mapped to the unified vector space to perform balanced optimization and establish a balanced optimization result; the visualization terminal interactively displays the balanced optimization result, thereby achieving the technical effect of providing visual data support and accurate decision-making assistance.
[0092] It should be understood that the embodiments and the above descriptions disclosed in the present invention enable those skilled in the art to implement the present invention using the present invention. At the same time, the present invention is not limited to the above-mentioned part of the embodiments. It should be understood that those of ordinary skill in the art can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.
Claims
1. Decision assistance system for dialysis modality decision-making in end-stage renal disease patients, characterized in that, The decision-making assistance system for the dialysis method of end-stage renal disease patients includes: A multi-dimensional data collection module, which is used to collect the data of patients, establish a patient data set, and the patient data set includes a physiological index set and a non-physiological index set; An integrated geographic information module, which is used to generate a dialysis center accessibility index, and the accessibility index includes real-time traffic time cost and public transportation coverage density; A feature coupling module, which is used to perform adaptive weight allocation after mapping the patient data set and the accessibility index to a unified vector space; A fitness analysis unit, which is used to configure a balanced fitness function based on the adaptive weight allocation result, perform balanced optimization using the balanced fitness function and the vector data mapped to the unified vector space, and establish a balanced optimization result; A visualization terminal, which is used to interactively display the balanced optimization result.
2. The decision-making assistance system for dialysis methods for end-stage renal disease patients according to claim 1, wherein The multi-dimensional data collection module includes: A physiological data collection terminal, which is used to collect the physiological monitoring data of patients, and the physiological monitoring data includes residual renal function index, blood pressure fluctuation coefficient, body surface area parameter, and synchronize the physiological monitoring data to the multi-dimensional data collection module through the Internet; An interactive collection terminal, which is used to perform medical insurance data docking according to the information of patients, obtain the annual dialysis limit and reimbursement ratio, and synchronously obtain the family economic status of patients. After establishing non-physiological monitoring data, it is synchronized to the multi-dimensional data collection module.
3. The decision-making assistance system for dialysis methods for end-stage renal disease patients according to claim 1, wherein The feature coupling module includes: A first coupling channel, which is used to perform patient status time series analysis on the physiological index set in the patient data set and establish a time series feature set; A second coupling channel, which is used to perform keyword joint extraction on the non-physiological index set in the patient data set through a knowledge language model and establish a keyword embedding vector; A normalization channel, which is used to receive the time series feature set, the keyword embedding vector, and the accessibility index, perform normalization processing in the same scale range, and then perform feature splicing; A unified space mapping channel, which is used to perform vector space mapping on the feature splicing result through a multi-layer perceptron.
4. The decision-making assistance system for dialysis methods for end-stage renal disease patients according to claim 1, wherein, In the feature coupling module, performing adaptive weight allocation includes: A similar label matching unit, which is used to obtain the patient label of the patient, perform similar label matching based on the patient label, and establish a similar label matching result; A feature initial weight establishment unit, which is used to establish a feature initial weight using the similar label matching result; A feature dot product attention score calculation unit, which is used to calculate the feature dot product attention score of the unified vector space mapping result; An adaptive weight allocation unit, which is used to scale the result of the feature dot product attention score calculation, perform normalization processing, and perform proportional weighting according to the normalization processing result and the patient selection preference to complete the adaptive weight allocation.
5. The decision-making assistance system for dialysis methods for end-stage renal disease patients according to claim 1, wherein The fitness analysis unit includes: A balanced fitness function construction unit, which is used to construct a balanced fitness function. The balanced fitness function is constructed based on the adaptive weight allocation result, and the evaluation features of the balanced fitness function include economic burden feature, effect feature, and convenience feature; An optimization unit, configured to configure an initial solution set according to vector data mapped to a unified vector space, evaluate the fitness of the initial solution set by using the balance fitness function, establish a fitness evaluation result, perform iterative update of the initial solution set by using the fitness evaluation result, and establish a balance optimization result according to the iterative update result.
6. The decision-making assistance system for dialysis methods for end-stage renal disease patients according to claim 5, wherein In the optimization unit, performing iterative update of the initial solution set by using the fitness evaluation result includes: A target solution determination and follower solution marking unit, configured to determine a target solution in the initial solution set according to the fitness evaluation result, and mark non-target solutions as follower solutions; An optimization bias coefficient generation unit, configured to perform imbalance calculation on all follower solutions to generate an optimization bias coefficient; A primary iterative optimization unit, configured to perform primary iterative optimization of the initial solution set according to the optimization bias coefficient and the target solution; An iterative update completion unit, configured to complete iterative update according to multi-round iterative optimization results.
7. The decision-making assistance system for dialysis methods for end-stage renal disease patients according to claim 1, wherein The visualization terminal further includes: A user interaction module, configured to read reading preference data of a patient, establish a reading scoring sample set, perform display matching analysis of the reading scoring sample set, and establish a patient display preference; A display module, configured to perform interactive display after organizing the balance optimization result according to the patient display preference.
8. The decision-making assistance system for dialysis methods for end-stage renal disease patients according to claim 1, wherein The end-stage renal disease patient dialysis modality decision-making assistance system further includes: A self-optimization feedback module, configured to receive feedback data of a patient and associated users, perform co-directional optimization analysis based on the feedback data, establish feedback optimization data, and update the system by using the feedback optimization data.
9. An electronic device, characterized in that, The electronic device includes: A memory, configured to store executable instructions; A processor, configured to implement the end-stage renal disease patient dialysis modality decision-making assistance system according to any one of claims 1 to 8 when executing the executable instructions stored in the memory.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the end-stage renal disease patient dialysis modality decision-making assistance system according to any one of claims 1 to 8.