CKD nutrition management visual monitoring method and system based on big data

Abstract

The invention provides a CKD nutrition management visual monitoring method and system based on big data, and relates to the technical field of data processing, and the method comprises the steps: carrying out the data preprocessing according to real-time health data, and obtaining a standardized health data set; according to the standardized health data set, evaluating nutritional requirements of the patient, and combining disease progression, individual differences and diet preferences of the patient to generate a personalized nutrition management scheme, the personalized nutrition management scheme including recommended diet, a nutrition supplement scheme and an intake behavior adjustment scheme; according to the personalized nutrition management scheme, a visual display interface is generated, and the visual display interface displays the nutrition state, diet suggestions, potential health risks and predicted effects of the patient; adjusting a personalized nutrition management scheme in real time according to feedback data and health state changes provided by the patient in the use process, and presenting an updated scheme through a visual display interface; the autonomy and accuracy of nutrition management are improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a method and system for visual monitoring of CKD nutritional management based on big data. Background Art

[0002] Currently, nutritional management for chronic kidney disease (CKD) primarily relies on manual assessment and adjustment by clinicians using methods such as blood tests and urine analysis, combined with patients' dietary records and medical history information. However, traditional nutritional management methods mostly rely on manual calculations and empirical judgment, lacking efficient, real-time dynamic monitoring methods. Existing technical solutions typically include regular reports based on laboratory data and subjective analysis by doctors, but in long-term monitoring, the real-time nature of data and personalized adaptation have significant limitations. With the development of big data technology, some systems have attempted to provide patients with more precise nutritional plans through data fusion and analysis, but most of these systems still cannot achieve comprehensive data visualization and struggle to reflect changes in patients' health in real time.

[0003] In actual applications, although existing CKD nutrition management systems assist decision-making through automated data collection and analysis, they often have major technical flaws in specific scenarios. For example, existing systems often have difficulty in real-time processing and dynamic updating of big data, resulting in delayed system feedback. In the process of nutritional adjustment for patients, especially in terms of diet control and sodium restriction, individual differences among patients are not fully considered, and the system is unable to capture and feedback in real time the subtle fluctuations in the patient's diet and health status. This prevents patients from receiving timely and accurate nutritional adjustments during treatment, thereby affecting their recovery effect, especially in patients whose condition changes more rapidly. Summary of the Invention

[0004] The purpose of the present invention is to provide a CKD nutritional management visualization monitoring method and system based on big data, aiming to solve the problems mentioned in the background technology.

[0005] In order to solve the above technical problems, the technical solutions of the present invention are as follows:

[0006] In a first aspect, a visual monitoring method for CKD nutritional management based on big data is provided, the method comprising:

[0007] Based on real-time health data, data preprocessing is performed to obtain standardized health data sets;

[0008] Assess the patient's nutritional needs based on standardized health data sets and generate a personalized nutritional management plan based on the patient's disease progression, individual differences, and dietary preferences. The personalized nutritional management plan includes a recommended diet, nutritional supplement plan, and dietary behavior adjustment plan.

[0009] Generate a visual display interface based on the personalized nutrition management plan, which displays the patient's nutritional status, dietary recommendations, potential health risks, and expected effects;

[0010] Based on the feedback data and health status changes provided by patients during use, the personalized nutrition management plan is adjusted in real time, and the updated plan is presented through a visual display interface.

[0011] Preferably, the patient's nutritional needs are assessed based on standardized health data sets, and a personalized nutritional management plan is generated based on the patient's disease progression, individual differences, and dietary preferences, including:

[0012] Extracting key nutrition-related indicators from a standardized health data set to construct a health status parameter set, the health status parameter set including renal function indicators, nutrient intake ratios, and trace element levels;

[0013] According to the health status parameter set, the patient's nutritional metabolic load is analyzed, the nutritional intake restriction conditions and nutritional supplement priority are determined, and the nutritional regulation factor set is obtained;

[0014] Based on the nutritional regulatory factor set and combined with the patient's previous disease development trajectory and physical characteristics, a nutritional recommendation matrix is ​​generated. The nutritional recommendation matrix is ​​based on the nutrient category, intake level and intake frequency.

[0015] Based on the nutrition recommendation matrix and combined with the patient's preference characteristics reflected in the historical diet record, a nutrition recommendation optimization model is formed to output a personalized nutrition management plan, which includes a recommended diet, a nutritional supplement plan, and an intake behavior adjustment plan.

[0016] Preferably, the personalized nutrition management plan is adjusted in real time based on the feedback data and health status changes provided by the patient during use, including:

[0017] Collect behavioral feedback data, periodic health indicator change data, and subjective feedback information from patients during the implementation of the nutrition plan to construct a feedback data set. The feedback data set includes intake deviation records, physiological response records, indicator fluctuation series, and patient satisfaction records for the personalized nutrition management plan.

[0018] Based on the feedback data set, the adaptability of the existing personalized nutrition management plan is evaluated and adjustment trigger factors are extracted. The adjustment trigger factors include the nutritional intake deviation rate, the abnormal frequency of indicators and the feedback matching score;

[0019] According to the adjustment trigger factors, local parameters are modified based on the original nutrition recommendation matrix, the nutrition recommendation matrix and nutrition recommendation optimization model are updated, and the adjusted personalized nutrition management plan is output;

[0020] The adjusted personalized nutrition management plan will be synchronously updated to the visual display interface.

[0021] Preferably, a nutrition recommendation matrix is ​​generated based on the nutritional regulatory factor set, combined with the patient's previous disease development trajectory and physical characteristics, including:

[0022] Constructing a nutritional factor structure diagram based on the intake restriction parameters and the supplementation priority parameters in the nutritional regulation factor set, wherein the nutritional factor structure diagram maps the nutritional intake target to the control boundary;

[0023] Extract the disease evolution sequence based on the patient's previous disease development trajectory, which includes the change trends and prognosis characteristics of the main indicators at each stage;

[0024] Extracting a static individual parameter set based on the patient's physical characteristics, wherein the static individual parameter set includes weight, age, metabolic type, and exercise habit factors;

[0025] A multidimensional correlation analysis is performed on the nutritional factor structure diagram, disease evolution sequence, and static individual parameter set to construct nutritional correlation mapping data. Based on the nutritional impact weight distribution rule, a nutritional recommendation matrix is ​​generated. The nutritional recommendation matrix is ​​based on the dimensions of nutrient category, intake level, and intake frequency.

[0026] Preferably, based on the nutrition recommendation matrix and combined with the patient's preference characteristics reflected in historical dietary records, a nutrition recommendation optimization model is formed to output a personalized nutrition management plan, including:

[0027] Based on the patient's historical dietary records, extract food frequency statistics and intake habit distribution to construct a dietary behavior feature map;

[0028] Based on the nutrition recommendation matrix and the dietary behavior characteristic map, similarity calculation is performed to generate a set of preference matching factors;

[0029] Based on the preference matching factor set, a nutrition recommendation optimization model is constructed. The nutrition recommendation optimization model determines the implementation order and weight of each recommendation item based on the weighted fusion between recommendation priority and acceptance;

[0030] According to the nutrition recommendation optimization model, various food items in the nutrition recommendation matrix are adjusted to output a personalized nutrition management plan.

[0031] Preferably, based on the feedback data set, the adaptability of the existing personalized nutrition management program is evaluated and the adjustment trigger factors are extracted, including:

[0032] The nutritional intake deviation rate was calculated based on the intake deviation records in the feedback data set;

[0033] Based on the physiological response records and indicator floating sequences in the feedback data set, the fluctuation trend of key indicators in a continuous time period is extracted, and the frequency value of abnormal fluctuation events is calculated to generate the indicator abnormal frequency;

[0034] Based on the subjective feedback information in the feedback dataset and the patient's satisfaction record with the personalized nutrition management plan, the feedback matching score is calculated;

[0035] The nutritional intake deviation rate, indicator abnormality frequency and feedback matching score were used as evaluation dimensions to construct an adaptability evaluation vector.

[0036] According to the adaptability evaluation vector and the corresponding preset adjustment threshold group, the trigger state of each dimension indicator is determined, and the indicators that have been determined to be in the trigger state are extracted to form an adjustment trigger factor set.

[0037] Preferably, according to the adjustment trigger factor, local parameters are modified on the basis of the original nutrition recommendation matrix, the nutrition recommendation matrix and the nutrition recommendation optimization model are updated, and the adjusted personalized nutrition management plan is output, including:

[0038] According to the intake deviation items in the adjustment trigger factor set, the nutritional indicators that deviate from the intake target in the nutritional recommendation matrix are identified, their recommended frequencies are adjusted, and the deviation correction sub-matrix is ​​generated;

[0039] According to the abnormal frequency of indicators in the adjustment trigger factor set, the intake tolerance range and intake time limit of the relevant nutrients in the nutrition recommendation matrix are adjusted to generate a restriction parameter set;

[0040] According to the feedback matching scores in the adjustment trigger factor set, the recommendation acceptance factor in the nutrition recommendation optimization model is adjusted, and the implementation order and weight of each recommendation item are reconstructed to obtain an updated nutrition recommendation optimization model;

[0041] The bias correction sub-matrix, the restriction parameter set and the updated nutrition recommendation optimization model are integrated to generate an adjusted nutrition recommendation matrix;

[0042] Based on the adjusted nutrition recommendation matrix and the updated nutrition recommendation optimization model, the adjusted personalized nutrition management plan is output.

[0043] In a second aspect, a CKD nutritional management visualization monitoring system based on big data is provided, the system comprising:

[0044] A health data acquisition module is used to obtain real-time health data of patients, including blood test data, urine test data, diet records and lifestyle information. The data is collected in real time through smart devices, sensors and electronic health record systems and transmitted to the data platform;

[0045] A data preprocessing module is used to perform data preprocessing based on real-time health data to obtain a standardized health data set, which includes various health indicators, long-term health trends and current status information of the patient;

[0046] A nutrition management plan generation module is used to assess the patient's nutritional needs based on a standardized health data set and generate a personalized nutrition management plan based on the patient's disease progression, individual differences, and dietary preferences. The personalized nutrition management plan includes a recommended diet, nutritional supplement plan, and dietary behavior adjustment plan;

[0047] A visualization display module is used to generate a visualization display interface based on the personalized nutrition management plan, which displays the patient's nutritional status, dietary recommendations, potential health risks, and expected effects;

[0048] The feedback adjustment module is used to adjust the personalized nutrition management plan in real time based on the feedback data and health status changes provided by the patient during use, and present the updated plan through a visual display interface.

[0049] The above solution of the present invention includes at least the following beneficial effects:

[0050] This invention addresses existing nutritional management solutions' shortcomings, including poor real-time performance, lack of individual adaptability, and lack of data visualization support, by providing a big data-based visual monitoring method for CKD nutritional management. This method first uses multi-source intelligent terminal devices to collect patients' blood test data, urine test data, dietary records, and lifestyle information in real time and transmits them to a unified data platform. This effectively addresses the traditional system's reliance on manual data entry and long data update cycles, enabling real-time acquisition and dynamic integration of CKD-related health data.

[0051] Based on data acquisition, the present invention further designs a standardized data preprocessing process to normalize data from different sources into a uniformly structured health data set, which contains quantifiable health indicators, trend information, and status determination results, making subsequent nutrition assessment and plan generation more efficient and accurate. At the same time, the system can dynamically generate personalized nutrition management plans based on multiple factors such as the patient's disease stage, nutritional status, and individual preferences. The plan not only includes dietary recommendations, but also covers nutritional supplement suggestions and intake behavior adjustments. It significantly improves the individual adaptability of nutritional intervention and makes up for the defect that the existing technology fails to fully respond to individual differences.

[0052] Notably, this method presents personalized nutritional management plans through a structured, visual interface. This graphically displays the patient's nutritional structure, key health indicator trends, and potential risk assessment results, making it easier for patients and healthcare professionals to understand and make nutritional adjustments in a timely manner. This approach effectively enhances the interactivity and transparency of CKD nutritional management, overcoming the issues of delayed data feedback and difficulty interpreting information in existing systems.

[0053] Furthermore, this invention supports real-time adjustments to nutritional plans based on patients' dynamic feedback data and changes in health status, thereby establishing a complete closed-loop nutritional management mechanism. This feedback-driven iterative optimization mechanism not only increases the system's sensitivity and responsiveness to health fluctuations, but also improves the accuracy and safety of nutritional control in rapidly evolving disease scenarios, meeting the long-term, dynamic nutritional intervention needs of CKD patients.

[0054] Therefore, compared with the existing technology, the present invention has significant technological progress in realizing personalized, real-time and visual nutritional management. It can be widely used in various application scenarios such as clinical auxiliary treatment, home health management and intelligent monitoring of chronic diseases, and has outstanding practical value and promotion potential. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 This is a flowchart of a method for visually monitoring CKD nutritional management based on big data provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0056] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0057] like Figure 1As shown, an embodiment of the present invention proposes a CKD nutritional management visualization monitoring method based on big data, the method comprising:

[0058] S100: Acquire the patient's real-time health data, including blood test data, urine test data, dietary records, and lifestyle information. The data is collected in real time through smart devices, sensors, and electronic health record systems and transmitted to the data platform;

[0059] S200, performing data preprocessing based on the real-time health data to obtain a standardized health data set, wherein the standardized health data set includes various health indicators, long-term health trends, and current status information of the patient;

[0060] S300, assessing the patient's nutritional needs based on a standardized health data set, and generating a personalized nutritional management plan based on the patient's disease progression, individual differences, and dietary preferences. The personalized nutritional management plan includes a recommended diet, a nutritional supplement plan, and an intake behavior adjustment plan;

[0061] S400: Generate a visual display interface based on the personalized nutrition management plan, wherein the visual display interface displays the patient's nutritional status, dietary recommendations, potential health risks, and expected effects;

[0062] S500: Based on the feedback data and health status changes provided by patients during use, the personalized nutrition management plan is adjusted in real time, and the updated plan is presented through a visual display interface.

[0063] In an embodiment of the present invention, this method constructs a dynamic information set covering physiological status and behavioral habits by collecting real-time health data of patients from multiple dimensions. The patient's blood test data and urine test data, as key clinical indicators reflecting their renal function and metabolic status, can be periodically collected through hospital detection systems, portable collection devices or other means; living habit information can be recorded and obtained by wearable devices, including daily exercise duration, work and rest time, water and salt intake, etc.; dietary records can be actively entered by patients or automatically extracted using image recognition technology to obtain structured data containing intake categories, intake frequency and time tags. These heterogeneous data are uniformly encoded and transmitted to the data platform through the sensor network or electronic health record system to form a time-synchronized and clearly sourced original health data set.

[0064] To improve data quality and facilitate subsequent modeling, this method performs standardized preprocessing on the collected data. This processing can include operations such as data format unification, missing value filling, outlier detection and removal, and dimensional conversion, outputting a standardized health dataset with a standardized structure. While retaining the original timestamps and source labels, this dataset maps multiple dimensional information such as serum creatinine, urine protein, total energy intake, and nutrient intake ratios into a set of health indicators. It can also construct a time series structure to display stage-by-stage health change trends, such as the upward trend of creatinine, the fluctuation range of urine protein, and the historical deviation of potassium and phosphorus intake, which is conducive to assessing the patient's stage-by-stage nutritional status.

[0065] Based on a standardized health dataset, this method combines current health indicators, historical behavioral characteristics, and disease progression to generate a personalized nutrition management plan that includes dietary recommendations, nutritional supplementation suggestions, and dietary behavior optimization strategies through parameter reasoning and rule combination. Dietary recommendations can be combined with a list of appropriate foods based on the patient's nutritional deficiencies and existing contraindications. Nutritional supplementation recommendations can be strengthened by considering the missing trace elements. Dietary recommendations are structured to optimize the timing and frequency of intake based on the patient's past intake frequency and load response, to avoid continuous high-load intake.

[0066] To enhance the understanding and effectiveness of the plan, this method presents personalized nutrition management plans in a graphical interface, including nutritional structure diagrams, dietary recommendations, and key indicator trend charts. This can improve patients' understanding and compliance. Furthermore, the system supports the collection of patient feedback and health trends during implementation, and automatically activates the update process when deviations are detected, enabling dynamic adjustments to the nutrition management plan and improving the responsiveness and adaptability of the overall intervention.

[0067] This approach, through the construction of standardized data collection pathways, data processing procedures, and personalized output mechanisms, provides continuous support for nutritional management in CKD patients at home. Compared to traditional approaches that rely on static questionnaires and periodic assessments, it offers greater real-time, refined, and adaptive capabilities, significantly improving the effectiveness and intelligence of individualized interventions.

[0068] In a preferred embodiment of the present invention, based on standardized health data sets, the patient's nutritional needs are assessed, and a personalized nutritional management plan is generated based on the patient's disease progression, individual differences, and dietary preferences, including:

[0069] Extracting key nutrition-related indicators from a standardized health data set to construct a health status parameter set, the health status parameter set including renal function indicators, nutrient intake ratios, and trace element levels;

[0070] According to the health status parameter set, the patient's nutritional metabolic load is analyzed, the nutritional intake restriction conditions and nutritional supplement priority are determined, and the nutritional regulation factor set is obtained;

[0071] Based on the nutritional regulatory factor set and combined with the patient's previous disease development trajectory and physical characteristics, a nutritional recommendation matrix is ​​generated. The nutritional recommendation matrix is ​​based on the nutrient category, intake level and intake frequency.

[0072] Based on the nutrition recommendation matrix and combined with the patient's preference characteristics reflected in the historical diet record, a nutrition recommendation optimization model is formed to output a personalized nutrition management plan, which includes a recommended diet, a nutritional supplement plan, and an intake behavior adjustment plan.

[0073] In this embodiment of the present invention, when generating a personalized nutrition management plan, this method first extracts key nutrition-related indicators from a standardized health dataset to construct a set of health parameters encompassing nutritional status. This parameter set may include renal function indicators (such as eGFR and creatinine concentration), nutrient intake ratios (such as protein / energy and sodium / potassium ratios), and trace element levels (such as calcium, iron, and zinc concentrations). These indicators comprehensively reflect the patient's current metabolic state and nutritional burden.

[0074] On this basis, we can use the construction of a nutritional metabolic load assessment model or set rule logic to analyze the intake limits and supplementation priorities of various nutrients. For example, when it is identified that the patient has a high blood phosphorus level and a low calcium state, a lower upper limit for phosphorus intake can be set, while the recommended priority for calcium supplementation can be increased. Through this type of analysis process, a set of nutritional regulation factors is formed, each of which can characterize the control intensity and recommendation priority of a certain type of nutrient, providing a structural basis for subsequent matrix generation.

[0075] By combining a patient's disease trajectory (e.g., CKD stage, previous treatment history, and disease stability) with static individual characteristics (e.g., age, weight, metabolic type, and exercise frequency), a nutritional recommendation matrix can be generated through association analysis or conditional matching. This matrix is ​​constructed using nutrient categories as the primary axis, with matching intervals for each nutrient category in terms of intake and frequency, thereby mapping the relationship between food items and nutritional needs in a structured manner.

[0076] Furthermore, based on the patient's preferences as shown in their historical dietary records (e.g., frequency of specific food intake, food rejections), the fit between recommended items and patient acceptance is assessed, generating a nutritional recommendation optimization model that incorporates acceptance factors. By incorporating a coupling weight between recommendation acceptance and priority, this model optimizes or quantitatively modifies the ranking of recommended items, ultimately producing a personalized nutritional management plan that better matches the patient's behavioral habits.

[0077] Among them, key nutrition-related indicators are extracted based on the standardized health data set to construct a health status parameter set, which includes renal function indicators, nutrient intake ratios, and trace element levels. In the specific implementation, the items in the standardized health data set are first classified and processed, including laboratory indicators (such as creatinine, urea nitrogen, clearance rate, etc. in blood tests), daily intake information recorded in the diet log (such as the total intake of protein, sodium, phosphorus, potassium, etc.), and lifestyle habit indicators (such as weight changes, exercise frequency, etc.). From these multidimensional data, several factors that have the most significant impact on the nutritional status of CKD patients are screened out.

[0078] The renal function indicators usually include eGFR (estimated glomerular filtration rate), serum creatinine (SCr), urine protein level (UPCR), urea nitrogen (BUN) and other indicators, which are used to reflect the patient's renal metabolic waste clearance ability and are an important basis for judging whether the nutritional load can be metabolized normally.

[0079] Nutrient intake ratio refers to the distribution of nutrients in the nutritional structure with protein, carbohydrates, and fat as the main energy sources. It can also include factors such as protein quality distribution (the proportion of high-quality protein) and total fluid intake. This parameter is obtained by comparing actual intake data with the recommended intake standard.

[0080] Trace element levels include indicators such as blood sodium (Na+), blood potassium (K+), blood phosphorus (P), blood calcium (Ca2+), iron (Fe), and zinc (Zn). The concentration levels of these trace nutrients directly affect the internal environment stability and metabolic adaptability of CKD patients.

[0081] After integration, the above three types of indicators constitute a health status parameter set, which serves as the basic input for the next step of calculating nutritional metabolic load and assessing risk level.

[0082] In the actual implementation process, this method not only retains the medical rationality of the recommended content, but also fully considers the feasibility of the patient's intake behavior, achieving a dynamic balance between the scientific nature of the recommendation and behavioral adaptability, thereby improving the response rate and compliance of the CKD population to nutritional intervention in the long-term management.

[0083] Among them, according to the health status parameter set, the patient's nutritional metabolic load is analyzed, the nutritional intake restriction conditions and nutritional supplement priority are determined, and the nutritional regulation factor set is obtained. It is usually achieved in the following way:

[0084] First, using renal function as the primary control factor, we construct nutritional metabolic capacity intervals. Taking eGFR as an example, patients can be divided into different CKD stages, such as G1 (>90 mL / min / 1.73 m²) to G5 (<15 mL / min / 1.73 m²). Each stage has a different tolerance for metabolic load, with progressively more severe restrictions on protein, sodium, phosphorus, and other nutrients.

[0085] On this basis, the nutrient intake ratio is used as the load source and cross-matched with renal function parameters. If protein intake is high and clearance capacity is low, an intake limiting factor is automatically generated. Similarly, when trace element levels such as blood phosphorus exceed the recommended limit and intake does not fall below the recommended lower limit, an upper intake factor for that element is generated.

[0086] The priority of supplementation is determined by the degree of deficiency and weighted importance. For example, if a patient's iron stores are low, their iron intake is insufficient for a long time, and they also have symptoms of anemia, iron will be listed as a high-priority supplement factor.

[0087] Through the above method, a set of nutritional regulation factors including the upper and lower limits of nutritional intake, intake frequency restrictions, element priority supplementation sequence, etc. is generated, providing a quantitative basis for the construction of the subsequent recommendation matrix.

[0088] In a preferred embodiment of the present invention, a visual display interface is generated according to the personalized nutrition management plan, including:

[0089] Constructing a multidimensional visualization data set based on the personalized nutrition management plan, the multidimensional visualization data set includes a nutrition intake structure diagram, a key indicator change trend diagram, and a health status risk map;

[0090] Designing interactive data display components based on multidimensional visualization data sets, including a diet recommendation module, a nutrition assessment module, and a health warning module;

[0091] The interactive data display component is integrated into the visual display interface, which supports simultaneous viewing by doctors and patients.

[0092] In an embodiment of the present invention, in order to enhance the comprehensibility and operability of the personalized nutrition management plan, this method constructs a multidimensional visualization data set based on the plan generation and integrates it into an interactive display interface to support patients and doctors to obtain and interactively execute nutrition management content in a two-way manner.

[0093] The multidimensional visualization data set includes a nutritional intake structure diagram, a key indicator change trend diagram, and a health status risk map. The nutritional intake structure diagram is mainly used to visualize the nutritional components in the recommended recipes. It uses bar charts, radar charts, etc. to show the deviation between the recommended amount and the actual intake of components such as protein, carbohydrates, fat, sodium, potassium, etc., to assist patients in understanding the need for structural adjustment. The key indicator change trend diagram is constructed based on the detection cycle data, such as the dynamic change curve of creatinine levels, the fluctuation range of urine protein, etc., to facilitate the observation of health improvement trends before and after intervention. The health status risk map is based on a preset model to match the current health indicators with the risk classification standards. Risk radar charts, heat maps, etc. can be used to highlight abnormal items, which helps doctors quickly identify potential risks.

[0094] Based on data visualization, this method designs three types of interactive display modules: the dietary advice module is used to display daily recommended recipes, recommended intake frequency and intake time, and supports users to replace or modify recipe items with their preferences; the nutritional assessment module is used to dynamically calculate the compliance level of each nutrient component and display the deviation between the intake structure and the target structure through a graphical interface; the health warning module judges potential risks based on key indicators and preset thresholds. When an indicator exceeds the control range or the change trend is abnormal, it will actively trigger a prompt and provide targeted nutritional adjustment suggestions.

[0095] By integrating these components into a visual display interface, doctors can remotely access patients' nutritional status and performance feedback, while patients can independently review daily recommendations and implementation status, forming a digital management model for doctor-patient collaboration. This approach significantly enhances the transparency and implementation efficiency of nutritional recommendations, providing a powerful tool to support daily implementation, doctor follow-up, and risk monitoring in CKD patients during chronic disease management.

[0096] In a preferred embodiment of the present invention, the personalized nutrition management plan is adjusted in real time based on the feedback data and health status changes provided by the patient during use, including:

[0097] Collect behavioral feedback data, periodic health indicator change data, and subjective feedback information from patients during the implementation of the nutrition plan to construct a feedback data set. The feedback data set includes intake deviation records, physiological response records, indicator fluctuation series, and patient satisfaction records for the personalized nutrition management plan.

[0098] Based on the feedback data set, the adaptability of the existing personalized nutrition management plan is evaluated and adjustment trigger factors are extracted. The adjustment trigger factors include the nutritional intake deviation rate, the abnormal frequency of indicators and the feedback matching score;

[0099] According to the adjustment trigger factors, local parameters are modified based on the original nutrition recommendation matrix, the nutrition recommendation matrix and nutrition recommendation optimization model are updated, and the adjusted personalized nutrition management plan is output;

[0100] The adjusted personalized nutrition management plan will be synchronously updated to the visual display interface.

[0101] In an embodiment of the present invention, after executing the personalized nutrition management plan, this method further introduces a real-time feedback mechanism to achieve dynamically adjusted target adaptation. In the process of implementing the recommended diet, nutritional supplements, and intake behavior adjustments, patients continuously collect nutrition-related behavioral feedback data and periodic health indicator change data, and collect subjective feedback information, including satisfaction scores or descriptions of feelings. A feedback dataset is formed by the fusion of multiple types of feedback information. The structure of this dataset clearly records the actual implementation of intake behavior, the correlation between intake and response, and the fluctuation status of indicators over time, providing a basis for subsequent analysis.

[0102] In the feedback data set, intake deviation records can reflect the degree of deviation between patients and the nutritional recommendation matrix in actual implementation, such as failure to consume recommended foods, failure to meet recommended frequency, or failure to consume on time, revealing individual implementation differences; physiological response records identify the potential causal relationship between certain intake behaviors and adverse physiological reactions through temporal comparison of behaviors and health status; indicator floating sequences are based on the indicator change trend within a continuous time period, which can reveal the risk of sudden changes in physiological status caused by intake or other factors; subjective feedback information is used to capture patients' acceptance of the current nutritional management plan, reflecting the degree of fit between the difficulty of behavioral execution and subjective preferences.

[0103] This method leverages this feedback data to comprehensively assess the adaptability of the current nutritional management plan. Quantitative processing is performed across multiple dimensions, including deviation rate, abnormality frequency, and fit score. Feedback factors that have reached dynamic adjustment thresholds are extracted as adjustment triggers, indicating that the original recommended plan is inconsistent with the current status in some dimensions and requires revision. Incorporating these triggers, local parameter adjustments are made based on the original recommendations, ensuring continued consistency between the management strategy and the patient's status.

[0104] By incorporating execution feedback into a dynamic closed-loop control system, the complete process of personalized nutrition management from initial recommendation to continuous adjustment is achieved, enhancing the program's responsiveness to individual differences, behavioral deviations, and status changes, ensuring that the program always matches the patient's current health needs and behavioral abilities, and significantly improving the long-term effectiveness and sustainability of the intervention.

[0105] In a preferred embodiment of the present invention, a nutrition recommendation matrix is ​​generated based on the nutritional regulatory factor set, combined with the patient's previous disease development trajectory and physical characteristics, including:

[0106] Constructing a nutritional factor structure diagram based on the intake restriction parameters and the supplementation priority parameters in the nutritional regulation factor set, wherein the nutritional factor structure diagram maps the nutritional intake target to the control boundary;

[0107] Extract the disease evolution sequence based on the patient's previous disease development trajectory, which includes the change trends and prognosis characteristics of the main indicators at each stage;

[0108] Extracting a static individual parameter set based on the patient's physical characteristics, wherein the static individual parameter set includes weight, age, metabolic type, and exercise habit factors;

[0109] A multidimensional correlation analysis is performed on the nutritional factor structure diagram, disease evolution sequence, and static individual parameter set to construct nutritional correlation mapping data. Based on the nutritional impact weight distribution rule, a nutritional recommendation matrix is ​​generated. The nutritional recommendation matrix is ​​based on the dimensions of nutrient category, intake level, and intake frequency.

[0110] In an embodiment of the present invention, in the process of generating a structured nutrition recommendation matrix, this method uses a set of nutrition regulatory factors as the core input, combines the patient's disease progression trajectory and physical characteristics, and constructs a parameterized recommendation model driven by multidimensional constraints and recommendation priorities. The restriction parameters in the nutrition regulatory factor set can be determined based on the critical values ​​of nutritional metabolic load or health indicators. For example, the upper limit of protein intake may be based on urea nitrogen levels, sodium intake restrictions and edema risk, and the supplementation priority can be determined based on the trace element gap and the duration of low indicator values.

[0111] Based on this, this method extracts information about a patient's disease evolution, including historical diagnostic stages, outcomes, and treatment response characteristics, forming a disease evolution sequence that reveals the inherent rhythm and key inflection points of individual disease progression. Furthermore, physical characteristics such as age, weight, metabolic capacity, and exercise habits form a static set of individual parameters that reflect a patient's tolerance to nutritional intake regimens and baseline consumption levels.

[0112] By cross-correlating nutritional factor structure diagrams, disease evolution sequences, and static parameter sets, we construct nutritional association mapping data, model the relationship between nutrient categories and patient status, and assign priorities and parameter boundaries for intake and frequency to various nutritional intake items based on nutritional impact weighting rules. The final output is a three-dimensional nutritional recommendation matrix, covering complete recommended information from category selection to intake frequency and dosage, providing a highly actionable data foundation for subsequent matching and optimization.

[0113] This method improves the interpretability and adjustment flexibility of the nutrition recommendation plan by clarifying the parameter source and structure mapping logic, avoiding the uncertainty and incompatibility of traditional recipe recommendations that rely solely on nutritional reference intakes, ensuring that nutritional recommendations are more in line with the individual needs and medical boundary conditions of CKD patients, and enhancing the scientific nature and adaptability of the plan.

[0114] Among them, according to the intake restriction parameters and supplement priority parameters in the nutritional regulation factor set, a nutritional factor structure diagram is constructed, and the nutritional factor structure diagram maps the nutritional intake target and the control boundary. Specifically:

[0115] The nutritional factor structure diagram is a graphical modeling form used to represent individual nutritional needs and intake constraints. Its core purpose is to establish a correspondence between nutritional intake recommendation goals and medical control boundaries, so as to serve as the basis for boundary determination and screening of nutritional recommendations.

[0116] During the specific implementation process, the system first receives the data fields passed in by the nutritional regulation factor set, including: the recommended intake target values ​​of various nutrients (such as the recommended daily intake of protein, sodium, phosphorus, and potassium), the upper and lower limits of intake (determined by the patient's condition and reference guidelines), and the supplementation priority (indicating the severity of the current intake deviation of the nutrient and the clinical urgency of supplementation).

[0117] After structured processing, the above data is constructed into a two-dimensional graph, with the horizontal axis representing the nutrient type and the vertical axis representing the intake range. Each nutrient is represented in the graph as an intake range block, including the recommended target value centerline and its upper and lower control boundaries. If the priority parameter value of a nutrient component exceeds the set threshold, the system will highlight the target value block for that component in the structure graph using a marker or weight.

[0118] This structural diagram serves as the basis for subsequent matrix generation and deviation identification, achieving boundary control and target reinforcement of nutritional recommendations. It can dynamically respond to changes in the condition or feedback results for redrawing or recalculation.

[0119] Among them, based on the patient's previous disease development trajectory, the disease evolution sequence is extracted. The disease evolution sequence includes the change trends and prognosis characteristics of the main indicators in each stage. Specifically:

[0120] The disease evolution sequence is used to express the trend information of the patient's disease indicators changing over time during the previous diagnosis and treatment process. It is the core basis for evaluating long-term impact and stage adaptability in nutritional regulation decisions.

[0121] In practical applications, the system needs to extract the values ​​of key CKD-related indicators (such as eGFR, SCr, UPCR, and BUN) at different time points from patients' historical health records, periodic examination reports, and hospitalization records. Through time series construction and normalization, a set of indicator status points is formed, arranged in increasing order over time.

[0122] The system can analyze the disease change trends at different stages based on the slope and volatility of indicator changes, such as the accelerated decline period, stable period, and deterioration fluctuation period; and further mark its prognosis characteristics by statistically analyzing the medical intervention results or natural relief after this stage, such as stabilization after intervention, transition to a high-risk period, and continued deterioration of indicators.

[0123] The final disease evolution sequence can be represented as a time-indexed state sequence vector. Each node contains attributes such as the indicator numerical status, change trend type, outcome label and occurrence time, which are used to subsequently construct a mapping model with the relationship with nutritional factors to determine the direction (promotion / inhibition) and sensitivity of a certain type of nutritional factor on the evolution of the disease at the corresponding stage.

[0124] Among them, according to the patient's physical characteristics, a static individual parameter set is extracted, and the static individual parameter set includes weight, age, metabolic type and exercise habit factors. Specifically:

[0125] The static individual parameter set reflects the important basic physiological characteristics of the patient under long-term stable conditions, does not change with short-term treatment, and is the basic variable group for the recommendation system to perform personalized intake adjustment.

[0126] During the system implementation process, this parameter set is obtained through initial diagnosis file registration, questionnaire assessment, smart device detection or long-term observation, and includes the following categories:

[0127] Weight: directly affects total energy requirements and nutrient dosage conversion. It can be used together with height to calculate BMI and determine whether a person is malnourished or overweight.

[0128] Age: reflects metabolic efficiency and intake tolerance. Elderly patients usually have reduced protein utilization, so nutritional supplementation strategies need to be adjusted accordingly.

[0129] Metabolic type: Based on previous medical records and metabolic characteristics assessment (such as basal metabolic rate (BMR), glucose and lipid metabolism response, etc.), it is marked as a high metabolizer, a medium metabolizer, or a low metabolizer to determine nutrient absorption efficiency and clearance capacity;

[0130] Exercise habit factor: It comes from the activity monitoring devices worn by patients (such as bracelets, mobile phone sensors) or daily behavior check-ins. It is divided into static type (little exercise), moderately active, high-intensity active, etc., and has a significant impact on carbohydrate and protein requirements.

[0131] The above physical characteristics are input into the nutrition recommendation algorithm in the form of standardized vectors, which are used to adjust the core processes such as intake target weight, intake frequency setting, and recommendation ranking optimization to ensure that the plan is well adapted to individual characteristics.

[0132] Among them, a multidimensional correlation analysis is performed on the nutritional factor structure diagram, the disease evolution sequence, and the static individual parameter set to construct a nutritional correlation mapping table, and based on the nutritional impact weight distribution rule, a nutritional recommendation matrix is ​​generated. The nutritional recommendation matrix uses nutrient category, intake level and intake frequency as dimensions. This step is the core processing module for constructing a personalized nutritional recommendation model. Its technical essence lies in: by introducing the patient's long-term disease development trajectory (dynamic factors) and static individual characteristic parameters (such as age, metabolic type) into the nutritional regulation modeling, a quantitative recommendation model based on correlation weights is generated.

[0133] A nutritional factor structure diagram is a spatial distribution diagram of nutritional factors constructed based on a set of nutritional regulatory factors, with nutritional intake targets (e.g., daily protein target) and intake boundaries (upper and lower limits) as coordinate axes. This structure diagram provides a boundary framework for nutritional control.

[0134] The disease evolution sequence is derived from a patient's historical test reports, disease staging records, and intervention response history, forming a sequential chain of dynamic indicator changes along a timeline. By calculating the correlation between changes in key nutritional intake and fluctuations in the disease (such as the rate of GFR decline and the frequency of proteinuria exacerbations), the degree of positive and negative impact of nutritional factors on disease progression can be identified.

[0135] Static individual parameter sets provide the basis for personalized patient metabolism. For example, those who are older or have lower metabolic rates have worse tolerance for high energy intake and therefore need to be weighted down when recommending intake.

[0136] The above three types of information are integrated through multidimensional mapping technology (such as principal component analysis, association rule mining, fuzzy logic model, etc.) to construct a nutritional association mapping table. Each nutritional factor in the table corresponds to a recommended value range and implementation weight that is adapted to the patient's condition.

[0137] Finally, based on the nutritional impact weight distribution rule, a nutritional recommendation matrix is ​​generated. The three dimensions of the matrix are:

[0138] Nutritional content categories (such as protein, sodium, phosphorus, potassium, etc.);

[0139] Intake level (divided into interval levels, such as 0–10g, 10–20g);

[0140] Frequency of intake (e.g., once daily, twice weekly, restricted intake, etc.).

[0141] This matrix not only expresses the recommended content, but also clarifies the execution frequency and intake priority, providing a structured basis for subsequent model optimization and dynamic correction.

[0142] In a preferred embodiment of the present invention, a nutrition recommendation optimization model is formed based on the nutrition recommendation matrix and combined with the patient's preference characteristics reflected in the historical diet record to output a personalized nutrition management plan, including:

[0143] Based on the patient's historical dietary records, extract food frequency statistics and intake habit distribution to construct a dietary behavior feature map;

[0144] Based on the nutrition recommendation matrix and the dietary behavior characteristic map, similarity calculation is performed to generate a set of preference matching factors;

[0145] According to the preference matching factor set, a nutrition recommendation optimization model is constructed. The nutrition recommendation optimization model determines the implementation order and weight of each recommendation item based on the weighted fusion between recommendation priority and acceptance; wherein, ,

[0146] For the The weight coefficient of each recommendation item is used to determine the priority recommendation level of food in the personalized nutrition plan. 、 、 is the weighting coefficient, satisfying , For the The gap in nutrients corresponding to each recommended item is the difference between the recommended intake and the current intake. is the maximum gap value among all nutrients, used for normalization. For the The risk value of exceeding the standard for each recommended item is: For the The tolerable upper limit of intake of the recommended ingredients, For the patient's history of The frequency of intake of recommended items, is the total intake frequency, that is, the sum of the historical frequencies of all recommended items, For the The number of times the recommendation item was rejected, left over, replaced, or other records were not accepted. For the The impact value of the recommended items on key health indicators is expressed as The ratio of the nutritional factors contained in the recommended intake of each recommended item to the target values ​​of key health indicators, is the target value of the key health indicator at the current stage of the disease, is the value of the recommended intake time range, such as 8:00 to 12:00 is 4; each recommendation item corresponds to a preference matching factor; they are a one-to-one mapping relationship, and the preference matching factor reflects the "acceptance" of the recommendation item for the individual patient;

[0147] According to the nutrition recommendation optimization model, various food items in the nutrition recommendation matrix are adjusted to output a personalized nutrition management plan.

[0148] In an embodiment of the present invention, to improve the compatibility between the nutritional recommendation matrix and the patient's dietary habits, this method further constructs a nutritional recommendation optimization model that dynamically adjusts the priority and implementation weight of recommended items based on the behavioral data in the patient's historical dietary records. The food frequency statistics and intake habit distribution contained in historical dietary records can be used to derive dietary preference characteristics, forming a dietary behavior characteristic map that is stable and quantifiable in terms of food categories, intake time periods, and intake continuity.

[0149] By comparing each nutrient configuration in the recommendation matrix with the dietary behavior profile, similarity metrics are calculated to form a set of preference matching factors. This set of factors reflects the degree of consistency between the recommended items and the patient's past dietary preferences, providing an acceptance basis for subsequent model construction.

[0150] On this basis, the nutrition recommendation optimization model is constructed by weighted fusion between recommendation priority and acceptance. Priority can be derived from the supplementation or restriction level of nutrition regulatory factors, and acceptance is weighted according to the historical preference distribution. Through multi-dimensional weight fusion, the execution ranking and individualized weight of each recommendation item are output, and the recommendation matrix is ​​fine-tuned accordingly. Food items can be promoted to primary recommendations or downgraded to alternative items according to the size of the weight, thereby improving the feasibility of the recommendation content and patient satisfaction.

[0151] Ultimately, the new plan generated based on the optimization model not only retains the nutritional structure required for medical rationality, but also takes into account the patient's behavioral constraints, achieving a smooth transition from scientific recommendation to behavioral adaptation, significantly enhancing the actual feasibility of the recommended plan, and improving the sustained and stable effect of nutritional intervention in real situations.

[0152] In practice, the above calculation formula prioritizes the recommended items in a patient's personalized nutrition plan. This paper constructs a comprehensive evaluation function based on multi-source data to determine the weight coefficient for each recommended item based on the patient's current condition. This coefficient is generated by comprehensively considering the patient's nutritional intake gap, individual acceptance, disease stage adaptability, and intake timing flexibility.

[0153] Specifically, for the first For each recommended item, we first extract the nutritional gap corresponding to the item based on the standardized health data set, that is, the difference between the recommended intake and the current actual intake, which is recorded as To facilitate comparison between different nutrients, the maximum gap value among all ingredients is recorded as , and calculate the normalized gap weight accordingly. At the same time, obtain the risk value of the component from the nutritional regulation factor set and the upper limit of tolerance for intake , by suppressing and regulating the risk of exceeding the standard, nutritional priority parameters are formed.

[0154] Then, the frequency of the patient's intake of the recommended item in the recent period was extracted from the historical diet records. Total intake frequency during this period Calculate the acceptance score together. In order to eliminate food items that have a history of rejection, surplus, or substitution, count the number of negative interactions that appear in the dietary behavior characteristic map. , reducing its acceptance score in an exponential decay form. Furthermore, in order to reflect the adaptability of the recommendation item at the current stage of the disease, the impact value of the health indicator associated with the nutritional factor is extracted from the patient's disease evolution sequence. and the target value of the indicator at the current stage Compare the differences. The smaller the difference, the more the nutrient composition meets the current treatment needs.

[0155] In addition, the intake period restriction of the nutritional recommendation item will affect the flexibility of its actual implementation. Therefore, by analyzing the recommended time range specified in the restriction parameter set , calculate its duration , which is normalized to a score relative to the full 24 hours of the day and used to adjust the final weight.

[0156] The above four factors constitute the structural basis of weight evaluation. The final weight of the recommendation item is constructed by multiplying the weighted fusion of the three factors by the time period score. Different patients can adjust the fusion weight adaptively according to their own nutritional management needs. 、 、 , to highlight the intervention focus at different stages. The weight is ultimately used to optimize the ranked recommendations and improve the feasibility and compliance of nutritional interventions for patients.

[0157] Based on the patient's historical dietary records, food frequency statistics and intake distribution are extracted to construct a dietary behavior profile. The technical essence of this step is to statistically analyze the structured or semi-structured data in the patient's historical dietary records to construct a quantitative model that reflects their individual dietary behavior preferences. This model, as a dietary behavior profile, is used in subsequent matching calculations for personalized recommendations.

[0158] Specifically, dietary records may include daily diet logs filled out by patients, food types and intake times automatically identified by smart recording devices, or synchronized data from third-party health applications. After data analysis, a food frequency statistics table is first constructed to count the historical frequency of food corresponding to each type of nutrient. For example, it is recorded that in the past 30 days, the patient consumed eggs 15 times, soy products 12 times, lean meat 10 times, etc. The frequency of intake of each type of food reflects the patient's basic dietary tendencies.

[0159] Second, we analyze the distribution of intake times, intake ranges, and the number of food types in a single meal to identify the distribution of dietary habits. For example, some patients tend to consume more carbohydrates at breakfast and more protein at dinner; some patients consume only three to four meals per day. This information is used to construct dynamic weights to constrain recommended frequency and time periods.

[0160] Finally, the above frequency statistics and habit distribution are mapped into a unified multidimensional vector space to form a dietary behavior feature map for subsequent similarity calculation module to call.

[0161] Among them, based on the nutrition recommendation matrix and the dietary behavior characteristic map, similarity calculation is performed to generate a preference matching factor set. The purpose of this step is to quantitatively compare the nutrition recommendation goals with the patient's preferences, so as to adjust the weight of the recommendation content and ensure the feasibility and acceptability of the recommendation.

[0162] In practice, each recommended food category in the nutrition recommendation matrix has attributes such as recommended frequency, recommended intake, and recommended time period, which together form a recommendation vector. Each item in the dietary behavior feature map corresponds to a feature vector consisting of actual intake frequency, habitual intake, and typical intake time.

[0163] Each recommendation item is matched against its corresponding dietary behavior feature vector using a vector similarity algorithm (such as cosine similarity, Euclidean distance, or weighted Jaccard coefficient). The resulting preference matching factors form a set of preference matching factors, each reflecting the patient's acceptance of a particular recommendation item. This factor will be incorporated as a key parameter in the subsequent construction of the recommendation weighting model to achieve personalized optimization of recommended content.

[0164] In a preferred embodiment of the present invention, an interactive data presentation component is designed based on a multidimensional visualization dataset, including:

[0165] Based on the dietary recommendations in the personalized nutrition management plan, a dietary recommendation module is constructed. The dietary recommendation module is used to display daily recommended recipes, food intake frequency and intake time recommendations, and supports dish substitutions or combination adjustments based on user feedback;

[0166] Based on the nutritional indicators and intake data extracted from the multidimensional visualization data set, a nutritional assessment module is constructed. The nutritional assessment module is used to calculate the deviation between the current nutritional intake structure and the target nutritional structure, and graphically display the compliance status of each nutritional component;

[0167] Based on the key indicator change trend chart and health status risk map, a health warning module is constructed. The health warning module automatically determines potential health risks based on preset risk thresholds, outputs risk level prompts, and provides corresponding nutritional intervention suggestions;

[0168] The dietary advice module, nutritional assessment module and health warning module are integrated into the interactive display component, providing multi-dimensional indicator switching, historical record comparison and user-customized view functions, supporting two-way interaction between doctors and patients.

[0169] In an embodiment of the present invention, in order to achieve intuitive expression and user interaction of personalized nutrition management plans, this method constructs an interactive data display component based on a multidimensional visualization data set, aiming to improve the comprehension, executability and communication efficiency between doctors and patients of nutritional recommendations. After the generation of the personalized nutrition management plan is completed, the data dimensions related to the nutrition structure, intervention goals and health warnings in the plan are immediately extracted to form a visualization data set with a nutrition intake structure diagram, a key indicator change trend diagram and a health status risk map as the core. The above data set covers three dimensions: static structure, dynamic change and potential risks, and can comprehensively reflect the nutritional status from the cross-sectional and longitudinal time dimensions.

[0170] On this basis, three types of interactive data display modules with clear functions are designed. The dietary recommendation module is used to present daily recommended recipes, specific intake frequency and recommended intake time period, allowing users to independently choose alternative foods or adjust the combination according to their own tastes, available ingredients or subjective feedback during the implementation of the plan, thereby improving the flexibility of the recommended content. The nutritional assessment module is used to calculate the deviation between the nutritional intake structure and the target structure based on the latest collected intake behavior and test data, and graphically display the degree of compliance or excess of key nutrients such as protein, sodium, and potassium to help patients understand the deviation direction and adjustment space of nutritional intake. The health warning module provides risk classification prompts for situations where indicators continue to deviate from recommended values ​​or show abnormal fluctuations based on the key indicator change trend chart and health status risk map, and provides optional nutritional adjustment suggestions or review suggestions to assist doctors in formulating response measures.

[0171] The three modules are integrated into a unified user interface. Through data switching, indicator comparison, and historical review capabilities, they support simultaneous patient self-review and remote physician intervention. This approach breaks down the information barriers between recommended solutions and implementation feedback, enhancing management transparency, interactivity, and responsiveness, providing an efficient human-computer collaborative interface for chronic nutritional management.

[0172] In a preferred embodiment of the present invention, the adaptability of the existing personalized nutrition management program is evaluated based on the feedback data set, and the adjustment trigger factors are extracted, including:

[0173] The nutritional intake deviation rate was calculated based on the intake deviation records in the feedback data set;

[0174] Based on the physiological response records and indicator floating sequences in the feedback data set, the fluctuation trend of key indicators in a continuous time period is extracted, and the frequency value of abnormal fluctuation events is calculated to generate the indicator abnormal frequency;

[0175] Based on the subjective feedback information in the feedback dataset and the patient's satisfaction record with the personalized nutrition management plan, the feedback matching score is calculated;

[0176] The nutritional intake deviation rate, indicator abnormality frequency and feedback matching score were used as evaluation dimensions to construct an adaptability evaluation vector.

[0177] According to the adaptability evaluation vector and the corresponding preset adjustment threshold group, the trigger state of each dimension indicator is determined, and the indicators that have been determined to be in the trigger state are extracted to form an adjustment trigger factor set.

[0178] In an embodiment of the present invention, in the process of implementing personalized nutritional management, this method introduces a scheme adaptability evaluation mechanism driven by feedback data in order to achieve dynamic adaptation and continuous optimization. First, the feedback data of the patient in the process of implementing the management plan is collected to construct a feedback data set, which includes intake deviation records, physiological response records, indicator floating sequences and subjective feedback information. Among them, the intake deviation record reflects the patient's implementation deviation in actual eating behavior compared with the recommended plan, such as insufficient intake, time delay or food replacement; the physiological response record is used to capture the temporal correlation between intake behavior and potential adverse reactions, and assist in identifying highly sensitive intake items or intake intolerance periods; the indicator floating sequence reveals the fluctuation trend of key indicators such as creatinine, urine protein, blood sodium, etc. in continuous time periods, and determines whether there are abnormal fluctuation events; the subjective feedback information includes the patient's qualitative or quantitative score of the feasibility, comfort or satisfaction of the plan.

[0179] The above-mentioned feedback information is extracted into quantitative evaluation dimensions such as nutritional intake deviation rate, indicator abnormality frequency and feedback matching score, forming a multi-component adaptability evaluation vector. In order to achieve intelligent judgment and automatic triggering update, an adjustment threshold group corresponding to each dimension is preset, which represents the maximum deviation or minimum matching value allowed in that dimension. During the evaluation process, the adaptability status of the current plan in each dimension is judged by comparing the adaptability evaluation vector with the preset threshold group dimension by dimension. When a dimension indicator exceeds the threshold, it is judged to be in a triggered state, that is, the factor corresponding to the indicator is extracted and included in the adjustment trigger factor set.

[0180] Through this mechanism, quantitative analysis and difference identification of the adaptability of nutritional plans can be achieved, and the sources of dimensions that need to be adjusted can be automatically summarized, providing a clear target direction for subsequent recommendation matrix and optimization model corrections, thereby forming a closed-loop management system of feedback-adjustment-optimization, and improving the individualized response capability and continuous optimization efficiency of CKD nutritional management.

[0181] To further enhance the dynamic adaptability of the personalized nutrition plan, the present invention introduces a feedback matching scoring mechanism into the above formula, serving as an important basis for adjusting the personalized management plan. This scoring integrates three types of information provided by the patient during the plan implementation process: intake behavior, physiological responses, and subjective feedback, to assess the degree of match between the current nutrition plan and the patient's actual condition.

[0182] During the execution process, firstly, the actual intake value of each recommended content of the patient is obtained according to the intake deviation record in the feedback dataset. and the recommended intake Compare and calculate the relative deviation ratio to reflect the intake compliance. The smaller the deviation, the higher the degree of patient implementation of the recommendation. Secondly, extract the actual indicator value representing the physiological response from the periodic health monitoring data The target value of the indicator within the normal range As a reference, calculate the physiological deviation rate. This item is used to determine whether the implementation of the program has a positive impact on the patient's physiological indicators. Third, collect the satisfaction evaluation submitted by the patient through the terminal feedback, including questionnaire scores, semantic tags, etc., and construct the subjective evaluation value after normalization. The above three factors are multiplied by the coefficients 、 、 After weighting, the basic score value is formed. In order to further enhance the coupling relationship of the health status dimension, the total evaluation value of the current patient's health status is introduced , and with a preset maximum rating value After normalization, the score is multiplied by the score to obtain the final feedback matching score.

[0183] The feedback matching score can not only be used as a factor in adaptive analysis to participate in trigger condition judgment, but can also be used in long-term tracking to build a closed-loop verification system for patient behavior and management plan effectiveness, providing a data basis for subsequent parameter updates of the nutrition recommendation matrix and recommendation optimization model.

[0184] Among them, based on the physiological response records and indicator floating sequences in the feedback data set, the fluctuation trends of key indicators in continuous time periods are extracted, and the frequency values ​​of abnormal fluctuation events are calculated to generate indicator abnormality frequencies. The function of this step is to quantify the frequency of abnormal indicator events within a certain period by performing trend identification and fluctuation detection on numerical records directly related to health status in the feedback data set, thereby providing a data basis for adaptive analysis.

[0185] Physiological response records primarily include data on adverse reactions experienced by patients after consuming certain nutrients over a certain period of time, such as elevated blood potassium, edema, and indigestion. Indicator floating series refers to time series data for indicators such as eGFR, blood phosphorus, blood pressure, and blood sugar.

[0186] During the analysis, a sliding time window algorithm is first used to extract the indicator change curve within a fixed time span, detecting any drastic changes that exceed the normal physiological fluctuation range. For example, a continuous increase of more than 30% in blood phosphorus over a 24-hour period is considered an abnormal fluctuation event.

[0187] The system then counts the number of abnormal fluctuations in various indicators over a user-defined observation period (e.g., the last 14 or 30 days) to form a frequency set. Each indicator is assigned a frequency value, with higher values ​​indicating more pronounced fluctuations due to dietary influences. This frequency value serves as an important reference for determining whether nutritional management plans need to be adjusted.

[0188] In a preferred embodiment of the present invention, according to the adjustment trigger factor, local parameters are modified on the basis of the original nutrition recommendation matrix, the nutrition recommendation matrix and the nutrition recommendation optimization model are updated, and the adjusted personalized nutrition management plan is output, including:

[0189] According to the intake deviation items in the adjustment trigger factor set, the nutritional indicators that deviate from the intake target in the nutritional recommendation matrix are identified, their recommended frequencies are adjusted, and the deviation correction sub-matrix is ​​generated;

[0190] According to the abnormal frequency of indicators in the adjustment trigger factor set, the intake tolerance range and intake time limit of the relevant nutrients in the nutrition recommendation matrix are adjusted to generate a restriction parameter set;

[0191] According to the feedback matching scores in the adjustment trigger factor set, the recommendation acceptance factor in the nutrition recommendation optimization model is adjusted, and the implementation order and weight of each recommendation item are reconstructed to obtain an updated nutrition recommendation optimization model;

[0192] The bias correction sub-matrix, the restriction parameter set and the updated nutrition recommendation optimization model are integrated to generate an adjusted nutrition recommendation matrix;

[0193] Based on the adjusted nutrition recommendation matrix and the updated nutrition recommendation optimization model, the adjusted personalized nutrition management plan is output.

[0194] In an embodiment of the present invention, after the aforementioned set of adjustment trigger factors is identified, in order to effectively modify the personalized nutrition management plan, this method further designs a plan adjustment process based on local parameter correction. First, based on the intake deviation item, the nutritional indicator dimensions in the nutrition recommendation matrix that deviate from the intake target are identified. Frequency reduction, intake level fine-tuning, or guided restriction operations are performed to construct a deviation correction submatrix. This submatrix focuses on the nutritional factors that need to be adjusted, controls the scope of modification, and avoids chain reactions caused by global interventions on other parameters.

[0195] Subsequently, based on the information on the frequency of abnormal indicators, we identify those nutritional factors that frequently induce fluctuations in health indicators, tighten their intake tolerance range in the recommendation matrix, and introduce intake time restriction rules, such as recommending time-sharing intake, prohibiting nighttime intake, etc., to construct a set of restriction parameters to reflect the behavioral boundary constraints of nutritional adjustments. At the same time, the recommendation acceptance factor in the nutrition recommendation optimization model is adjusted based on the feedback matching score. If the score shows that the recommendation item is difficult to implement or has low acceptance, the recommendation ranking is reconstructed, and the weight distribution is flexibly adjusted to re-establish the implementation order and priority of various food recommendation items, and finally form an updated nutrition recommendation optimization model.

[0196] These three adjustments impact the existing nutrition recommendation matrix from the perspectives of structural parameters, behavioral constraints, and recommendation strategies. By integrating the bias correction submatrix, the set of constraint parameters, and the updated optimization model, a new recommendation matrix with a complete structure, clear priorities, and behaviorally friendly optimization based on individual feedback is generated. Ultimately, based on this new recommendation matrix and the updated optimization model, a personalized nutrition management plan with enhanced adaptability is developed.

[0197] This solution not only achieves rapid response and personalized fine-tuning based on feedback data, but also maintains the recommended operational stability and high adaptability through parameterized structural changes and strategy reconstruction, allowing CKD patients to continue to receive optimal solution support that suits their current status during nutritional intervention.

[0198] Among them, according to the intake deviation items in the adjustment trigger factor set, the nutritional indicators that deviate from the intake target in the nutritional recommendation matrix are identified, their recommendation frequencies are adjusted, and the deviation correction sub-matrix is ​​generated. This step focuses on feeding back the difference between the actual intake situation and the recommended target to the nutritional recommendation to generate adjustment suggestions that are more in line with the patient's execution ability.

[0199] Intake deviation is derived from the comparison of actual intake values ​​with recommended values ​​in the patient feedback dataset, reflecting the degree of deviation of each nutrient during implementation. A pre-set deviation threshold (e.g., ±15%) is used to determine whether a particular nutrient is off target.

[0200] Once a deviation is identified, that metric is extracted from the original nutritional recommendation matrix and used as a correction target. Subsequently, the direction and severity of the deviation determine whether to increase or decrease the recommended frequency of that metric in the recommendation matrix. For example, if protein intake is chronically insufficient, the recommended frequency of protein recommendations may be increased from twice a week to once a day.

[0201] All adjustment results are summarized into the deviation correction sub-matrix, which only contains the modified content of the adjusted items and is used for subsequent fusion to generate the final recommendation matrix.

[0202] Among them, the deviation correction sub-matrix, the restriction parameter set and the updated nutrition recommendation optimization model are integrated to generate the adjusted nutrition recommendation matrix and output the adjusted personalized nutrition management plan based on the adjusted nutrition recommendation matrix and the updated nutrition recommendation optimization model. In this stage, all structured results from various feedback dimensions are unified and integrated to ensure the structural integrity and logical consistency of the recommendation matrix.

[0203] First, the deviation correction matrix serves as the basis for quantitative correction, primarily adjusting the recommended frequency and target intake value. The set of restriction parameters is derived from abnormal frequency analysis of indicators or physiological response information, typically manifesting as restrictive rules such as upper intake limits and intake time restrictions.

[0204] The nutrition recommendation optimization model models the acceptance of recommendations based on patient preferences. Its update process is affected by the feedback matching factor and may involve adjustments to recommendation weights or changes in recommendation order.

[0205] The fusion process is executed in a rule-driven manner. First, the adjustment order is confirmed: if a recommendation item has both intake deviation and physiological abnormalities, the adjustment of the restriction parameters is performed first, followed by frequency correction, and finally weight correction.

[0206] The integrated nutritional recommendation matrix is ​​the adjusted matrix, and all its dimensions (ingredient category, intake level, intake frequency, and acceptance weight) are personalized and remodeled through feedback-driven, reflecting real-time and dynamic regulation capabilities.

[0207] Finally, based on the adjusted matrix and optimization model, a complete personalized nutrition management plan is output, including recommended content, implementation frequency, intake strategy and suggestion feedback mechanism, and presented to patients and doctors in real time through a visual interface.

[0208] An embodiment of the present invention further provides a CKD nutritional management visualization monitoring system based on big data, the system comprising:

[0209] A health data acquisition module is used to obtain real-time health data of patients, including blood test data, urine test data, diet records and lifestyle information. The data is collected in real time through smart devices, sensors and electronic health record systems and transmitted to the data platform;

[0210] A data preprocessing module is used to perform data preprocessing based on real-time health data to obtain a standardized health data set, which includes various health indicators, long-term health trends and current status information of the patient;

[0211] A nutrition management plan generation module is used to assess the patient's nutritional needs based on a standardized health data set and generate a personalized nutrition management plan based on the patient's disease progression, individual differences, and dietary preferences. The personalized nutrition management plan includes a recommended diet, nutritional supplement plan, and dietary behavior adjustment plan;

[0212] A visualization display module is used to generate a visualization display interface based on the personalized nutrition management plan, which displays the patient's nutritional status, dietary recommendations, potential health risks, and expected effects;

[0213] The feedback adjustment module is used to adjust the personalized nutrition management plan in real time based on the feedback data and health status changes provided by the patient during use, and present the updated plan through a visual display interface.

[0214] It should be noted that this system is a system corresponding to the above method, and all implementation methods in the above method embodiment are applicable to this embodiment and can achieve the same technical effects.

[0215] An embodiment of the present invention further provides a computing device comprising: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, performs the above-described method. All implementations in the above-described method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0216] The embodiment of the present invention further provides a computer-readable storage medium storing instructions, which, when executed on a computer, causes the computer to execute the above-described method. All implementations in the above-described method embodiment are applicable to this embodiment and can achieve the same technical effects.

[0217] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A visual monitoring method for CKD nutritional management based on big data, characterized in that: The method comprises: Based on real-time health data, data preprocessing is performed to obtain standardized health data sets; Assess the patient's nutritional needs based on standardized health data sets and generate a personalized nutritional management plan based on the patient's disease progression, individual differences, and dietary preferences. The personalized nutritional management plan includes a recommended diet, nutritional supplement plan, and dietary behavior adjustment plan. Generate a visual display interface based on the personalized nutrition management plan, which displays the patient's nutritional status, dietary recommendations, potential health risks, and expected effects; Based on the feedback data and health status changes provided by patients during use, the personalized nutrition management plan is adjusted in real time, and the updated plan is presented through a visual display interface.

2. The method for visual monitoring of CKD nutritional management based on big data according to claim 1, characterized in that: Based on standardized health data sets, we assess patients' nutritional needs and generate personalized nutritional management plans based on their disease progression, individual differences, and dietary preferences, including: Extracting key nutrition-related indicators from a standardized health data set to construct a health status parameter set, the health status parameter set including renal function indicators, nutrient intake ratios, and trace element levels; According to the health status parameter set, the patient's nutritional metabolic load is analyzed, the nutritional intake restriction conditions and nutritional supplement priority are determined, and the nutritional regulation factor set is obtained; Based on the nutritional regulatory factor set and combined with the patient's previous disease development trajectory and physical characteristics, a nutritional recommendation matrix is ​​generated. The nutritional recommendation matrix is ​​based on the nutrient category, intake level and intake frequency. Based on the nutrition recommendation matrix and combined with the patient's preference characteristics reflected in the historical diet record, a nutrition recommendation optimization model is formed to output a personalized nutrition management plan, which includes a recommended diet, a nutritional supplement plan, and an intake behavior adjustment plan.

3. The method for visual monitoring of CKD nutritional management based on big data according to claim 1, characterized in that: Based on the feedback data provided by patients during use and changes in their health status, the personalized nutrition management plan is adjusted in real time, including: Collect behavioral feedback data, periodic health indicator change data, and subjective feedback information from patients during the implementation of the nutrition plan to construct a feedback data set. The feedback data set includes intake deviation records, physiological response records, indicator fluctuation series, and patient satisfaction records for the personalized nutrition management plan. Based on the feedback data set, the adaptability of the existing personalized nutrition management plan is evaluated and adjustment trigger factors are extracted. The adjustment trigger factors include the nutritional intake deviation rate, the abnormal frequency of indicators and the feedback matching score; According to the adjustment trigger factors, local parameters are modified based on the original nutrition recommendation matrix, the nutrition recommendation matrix and nutrition recommendation optimization model are updated, and the adjusted personalized nutrition management plan is output; The adjusted personalized nutrition management plan will be synchronously updated to the visual display interface.

4. The method for visual monitoring of CKD nutritional management based on big data according to claim 2, characterized in that: Based on the nutritional regulatory factor set, combined with the patient's previous disease development trajectory and physical characteristics, a nutritional recommendation matrix is ​​generated, including: Constructing a nutritional factor structure diagram based on the intake restriction parameters and the supplementation priority parameters in the nutritional regulation factor set, wherein the nutritional factor structure diagram maps the nutritional intake target to the control boundary; Extract the disease evolution sequence based on the patient's previous disease development trajectory, which includes the change trends and prognosis characteristics of the main indicators at each stage; Extracting a static individual parameter set based on the patient's physical characteristics, wherein the static individual parameter set includes weight, age, metabolic type, and exercise habit factors; A multidimensional correlation analysis is performed on the nutritional factor structure diagram, disease evolution sequence, and static individual parameter set to construct nutritional correlation mapping data. Based on the nutritional impact weight distribution rule, a nutritional recommendation matrix is ​​generated. The nutritional recommendation matrix is ​​based on the dimensions of nutrient category, intake level, and intake frequency.

5. The method for visual monitoring of CKD nutritional management based on big data according to claim 4, characterized in that: Based on the nutrition recommendation matrix and the patient's preference characteristics reflected in their historical dietary records, a nutrition recommendation optimization model is formed to output a personalized nutrition management plan, including: Based on the patient's historical dietary records, extract food frequency statistics and intake habit distribution to construct a dietary behavior feature map; Based on the nutrition recommendation matrix and the dietary behavior characteristic map, similarity calculation is performed to generate a set of preference matching factors; Based on the preference matching factor set, a nutrition recommendation optimization model is constructed. The nutrition recommendation optimization model determines the implementation order and weight of each recommendation item based on the weighted fusion between recommendation priority and acceptance; According to the nutrition recommendation optimization model, various food items in the nutrition recommendation matrix are adjusted to output a personalized nutrition management plan.

6. The method for visual monitoring of CKD nutritional management based on big data according to claim 3, characterized in that: Based on the feedback data set, the adaptability of the existing personalized nutrition management plan is evaluated and the trigger factors for adjustment are extracted, including: The nutritional intake deviation rate was calculated based on the intake deviation records in the feedback data set; Based on the physiological response records and indicator floating sequences in the feedback data set, the fluctuation trend of key indicators in a continuous time period is extracted, and the frequency value of abnormal fluctuation events is calculated to generate the indicator abnormal frequency; Based on the subjective feedback information in the feedback dataset and the patient's satisfaction record with the personalized nutrition management plan, the feedback matching score is calculated; The nutritional intake deviation rate, indicator abnormality frequency and feedback matching score were used as evaluation dimensions to construct an adaptability evaluation vector. According to the adaptability evaluation vector and the corresponding preset adjustment threshold group, the trigger state of each dimension indicator is determined, and the indicators that have been determined to be in the trigger state are extracted to form an adjustment trigger factor set.

7. The method for visual monitoring of CKD nutritional management based on big data according to claim 6, characterized in that: Based on the adjustment trigger factors, local parameters are modified on the basis of the original nutrition recommendation matrix, the nutrition recommendation matrix and nutrition recommendation optimization model are updated, and the adjusted personalized nutrition management plan is output, including: According to the intake deviation items in the adjustment trigger factor set, the nutritional indicators that deviate from the intake target in the nutritional recommendation matrix are identified, their recommended frequencies are adjusted, and the deviation correction sub-matrix is ​​generated; According to the abnormal frequency of indicators in the adjustment trigger factor set, the intake tolerance range and intake time limit of the relevant nutrients in the nutrition recommendation matrix are adjusted to generate a restriction parameter set; According to the feedback matching scores in the adjustment trigger factor set, the recommendation acceptance factor in the nutrition recommendation optimization model is adjusted, and the implementation order and weight of each recommendation item are reconstructed to obtain an updated nutrition recommendation optimization model; The bias correction sub-matrix, the restriction parameter set and the updated nutrition recommendation optimization model are integrated to generate an adjusted nutrition recommendation matrix; Based on the adjusted nutrition recommendation matrix and the updated nutrition recommendation optimization model, the adjusted personalized nutrition management plan is output.

8. An autonomous mooring system for unmanned vessels, characterized in that: Applied to the method according to any one of claims 1 to 7, the system comprises: A health data acquisition module is used to obtain real-time health data of patients, including blood test data, urine test data, diet records and lifestyle information. The data is collected in real time through smart devices, sensors and electronic health record systems and transmitted to the data platform; A data preprocessing module is used to perform data preprocessing based on real-time health data to obtain a standardized health data set, which includes various health indicators, long-term health trends and current status information of the patient; A nutrition management plan generation module is used to assess the patient's nutritional needs based on a standardized health data set and generate a personalized nutrition management plan based on the patient's disease progression, individual differences, and dietary preferences. The personalized nutrition management plan includes a recommended diet, nutritional supplement plan, and dietary behavior adjustment plan; A visualization display module is used to generate a visualization display interface based on the personalized nutrition management plan, which displays the patient's nutritional status, dietary recommendations, potential health risks, and expected effects; The feedback adjustment module is used to adjust the personalized nutrition management plan in real time based on the feedback data and health status changes provided by the patient during use, and present the updated plan through a visual display interface.

9. A computing device, characterized in that include: one or more processors; A storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a program, which, when executed by a processor, implements the method according to any one of claims 1 to 7.

Images