A method and system for constructing a personalized nutritional therapy knowledge base for critically ill patients

Abstract

The present invention relates to the field of medical recommendation technology, and in particular to a method and system for constructing a personalized nutritional therapy knowledge base for critically ill patients. The system is used to: obtain the basic information, medical history information, laboratory test data, and dietary intake information corresponding to critically ill patients and construct critical case entities to generate entity descriptors for each critically ill patient case; construct a corresponding personalized nutritional therapy initial knowledge base for each critically ill patient case entity descriptor in combination with the semantic knowledge association ratio between relevant scientific research literature; obtain clinical pathological characteristics and nutritional therapy knowledge entities and perform nutritional demand analysis and drug administration timing determination, while performing nutritional administration conflict assessment to obtain a clinical nutritional administration-dietary effect conflict ratio; perform nutritional therapy knowledge update and optimization on the personalized nutritional therapy initial knowledge base to generate a personalized nutritional therapy knowledge optimization library. The present invention can provide accurate nutritional support for critically ill patients.

Description

Technical Field

[0001] The present invention relates to the field of medical recommendation technology, and in particular to a method and system for constructing a personalized nutritional therapy knowledge base for critically ill patients. Background Art

[0002] Nutritional support, as an important component of critically ill patients, plays a vital role in improving patients' prognosis, reducing complications, and promoting recovery. In recent years, personalized treatment plans based on big data and artificial intelligence technologies have gradually been applied, but the current research and practice of building a knowledge base for personalized nutritional treatment of critically ill patients is still relatively scarce. Especially in the complex situation of coexistence of multiple diseases, how to provide accurate nutritional intervention plans based on each patient's specific condition, nutritional needs, metabolic characteristics and treatment response is still a difficult point in clinical practice. However, although there are some nutritional treatment guidelines and recommended plans in clinical practice, these plans are often based on group data and clinical experience summaries, lacking an accurate grasp of individual differences, so that doctors and nutritionists need to rely on a large amount of patient physiological data, laboratory test results, medical records and other information to determine the treatment plan, which leads to the problem that doctors and nutritionists often face scattered data and difficulty in making real-time and accurate nutritional treatment adjustments when formulating personalized treatment plans. Summary of the invention

[0003] Based on this, it is necessary for the present invention to provide a method and system for constructing a personalized nutritional therapy knowledge base for critically ill patients to solve at least one of the above technical problems.

[0004] To achieve the above objectives, a personalized nutritional therapy knowledge base construction system for critically ill patients includes the following modules:

[0005] The critically ill patient case entity description module is used to obtain the basic information, medical record information, laboratory test data and dietary intake information corresponding to the critically ill patients, and construct the critically ill case entity according to the basic information, medical record information, laboratory test data and dietary intake information corresponding to the critically ill patients, thereby generating each critically ill patient case entity descriptor;

[0006] The critical illness entity nutrition knowledge base construction module is used to construct the personalized nutrition treatment initial knowledge base corresponding to each critical illness patient based on the entity descriptors of each critical illness patient case and the semantic knowledge association ratio between relevant clinical medical research literature;

[0007] The module for determining nutritional needs and medication timing is used to obtain the clinical pathological characteristics and nutritional therapy knowledge entities corresponding to critically ill patients through the personalized nutritional therapy initial knowledge base, and to analyze nutritional needs and determine medication timing based on the clinical pathological characteristics and nutritional therapy knowledge entities, so as to obtain the clinical nutritional needs and nutritional medication timing corresponding to each critically ill patient;

[0008] The personalized nutritional therapy knowledge base optimization module is used to evaluate the nutritional administration conflict of the corresponding dietary intake information based on the clinical nutritional needs and nutritional administration timing of each critically ill patient, so as to obtain the clinical nutritional administration-dietary effect conflict ratio corresponding to each critically ill patient; based on the clinical nutritional administration-dietary effect conflict ratio corresponding to each critically ill patient, the personalized nutritional therapy initial knowledge base is updated and optimized for nutritional therapy knowledge, thereby generating a personalized nutritional therapy knowledge optimization base corresponding to each critically ill patient, which includes personalized nutritional knowledge practice guidelines, nutritional administration plans, nutritional intake contraindications and dietary adjustment recommendations corresponding to critically ill patients.

[0009] Furthermore, the critically ill patient case entity description module includes the following functions:

[0010] Obtain the basic information of critically ill patients through the critically ill patient information database of the target clinical hospital, including the age, gender, height, weight, condition description and clinical manifestations of the critically ill patients;

[0011] Obtain the medical record information of critically ill patients through the electronic medical records of critically ill patients in the target clinical hospital, including the basic medical records, causes of critical illness, complications and medical history of critically ill patients;

[0012] Obtain laboratory test data corresponding to critically ill patients through the target clinical hospital's corresponding automated laboratory analysis system, including heart rate, blood sugar, blood indicators, and metabolic status of critically ill patients;

[0013] Obtain the corresponding dietary intake information of critically ill patients during treatment through the target clinical hospital's corresponding dietary record documents during treatment, including the corresponding dietary habits and dietary intake records of critically ill patients during treatment;

[0014] The basic information, medical record information, laboratory test data and dietary intake information corresponding to critically ill patients are standardized and the corresponding case entity feature description vectors are extracted to generate case entity descriptors for each critically ill patient.

[0015] Furthermore, the critical care entity nutrition knowledge base construction module includes the following functions:

[0016] Obtain relevant clinical medical research literature corresponding to each critically ill patient;

[0017] Extract semantic knowledge entities from the relevant clinical medical research literature corresponding to each critically ill patient to obtain the semantic knowledge entities corresponding to the relevant medical literature corresponding to each critically ill patient;

[0018] Conduct knowledge attribute mining and analysis based on the semantic knowledge entities corresponding to the relevant medical literature of each critically ill patient, and obtain the semantic knowledge attribute relationship corresponding to the relevant medical literature of each critically ill patient;

[0019] Based on the semantic knowledge attribute relationship of the relevant medical literature corresponding to each critically ill patient, the corresponding relevant clinical medical research literature is classified for the same research purpose, so as to obtain the medical literature set with the same research purpose of the semantic knowledge corresponding to each critically ill patient;

[0020] The semantic knowledge entities corresponding to different medical documents with the same research purpose are obtained through the medical document set with the same research purpose corresponding to each critically ill patient, and the knowledge association ratio of the semantic knowledge entities corresponding to different medical documents with the same research purpose is calculated based on the critically ill patient case entity descriptor, so as to obtain the semantic knowledge association ratio between each critically ill patient case entity and different medical documents with the same research purpose;

[0021] Based on the semantic knowledge association ratio, the knowledge entity representation association relationship between the critically ill patient case entity descriptor and the semantic knowledge entities corresponding to different medical literature with the same research purpose was determined to construct an initial knowledge base of personalized nutritional treatment corresponding to each critically ill patient.

[0022] Furthermore, the calculation of the knowledge association ratio of the semantic knowledge entities corresponding to different medical documents with the same research purpose based on the entity descriptor of the critically ill patient case includes:

[0023] The semantic knowledge entities corresponding to different medical documents with the same research purpose are vectorized to obtain the knowledge entity description vectors corresponding to different medical documents with the same research purpose.

[0024] The cosine similarity is used to calculate the semantic entity similarity between the critically ill patient case entity descriptors and the knowledge entity description vectors corresponding to different medical documents with the same research purpose, so as to obtain the semantic entity similarity between each critically ill patient case entity and different medical documents with the same research purpose;

[0025] Conduct research field impact assessment analysis on the semantic knowledge entities corresponding to different medical documents with the same research purpose, and obtain the research field impact of the semantic knowledge entities corresponding to different medical documents with the same research purpose;

[0026] Based on the semantic entity similarity between each critically ill patient case entity and different medical literature with the same research purpose, as well as the corresponding semantic knowledge research field influence degree, the knowledge association ratio of the semantic knowledge entities corresponding to different medical literature with the same research purpose is calculated using the semantic knowledge association ratio calculation formula to obtain the semantic knowledge association ratio between each critically ill patient case entity and different medical literature with the same research purpose.

[0027] Furthermore, the calculation formula of the semantic knowledge association ratio is specifically:

[0028] ;

[0029] In the formula, For severe cases Corresponding medical literature with the same research purpose The semantic knowledge correlation ratio between Calculate the time window for semantic association, is the time variable parameter, is the total number of severe cases. The total number of corresponding medical literature for the same research purpose, For severe cases The corresponding semantic entity, For medical literature The corresponding semantic entity, For severe cases Middle Semantic Entities and Medical Literature Middle The semantic entity similarity between semantic entities, is the influence weight coefficient of semantic similarity association, For in time Cases of critically ill patients Middle Semantic Entities and Medical Literature Middle The semantic knowledge research field related measurement between semantic entities, The impact of semantic knowledge research field, is an exponential function, is the initial time of semantic knowledge, is the time decay rate control parameter, is the correction coefficient of the semantic knowledge association ratio.

[0030] Furthermore, the determination of the knowledge entity representation association relationship between the critically ill patient case entity descriptor and the semantic knowledge entities corresponding to different medical documents with the same research purpose based on the semantic knowledge association ratio includes:

[0031] The semantic knowledge association ratio is compared and judged according to a preset semantic knowledge association threshold. If the semantic knowledge association ratio is less than the preset semantic knowledge association threshold, it is determined that there is no knowledge entity representation association relationship between the critically ill patient case entity descriptor and the semantic knowledge entity corresponding to the current medical document with the same research purpose, and the semantic knowledge association ratio corresponding to the next medical document is compared and judged;

[0032] If the semantic knowledge association ratio is greater than or equal to the preset semantic knowledge association threshold, it is determined that there is a knowledge entity representation association relationship between the critically ill patient case entity descriptor and the semantic knowledge entity corresponding to the current medical literature with the same research purpose, and the corresponding semantic knowledge association ratio is used as the edge weight of the association relationship, so as to construct the initial knowledge base of personalized nutritional treatment corresponding to each critically ill patient based on the knowledge entity representation association relationship and the corresponding edge weight.

[0033] Furthermore, the module for determining nutritional needs and medication timing includes the following functions:

[0034] Obtain the clinical pathological characteristics of critically ill patients through the corresponding critically ill patient case entities in the initial knowledge base of personalized nutritional therapy, including clinical case types, clinical pathophysiological performance indicators, patient basal metabolic efficiency, and nutrient absorption efficiency;

[0035] Obtain the nutritional therapy knowledge entities corresponding to critically ill patients through the semantic knowledge entities corresponding to the relevant medical literature in the initial knowledge base of personalized nutritional therapy, including the nutritional supplement ingredients corresponding to critically ill patients at various clinical stages and the expected clinical nutritional effects;

[0036] According to the clinical case types and clinical pathophysiological indicators of critically ill patients, the corresponding nutritional treatment knowledge entities are analyzed for nutritional needs, so as to analyze and calculate the clinical nutritional needs of each critically ill patient, including the specific dosage requirements of protein, fat, carbohydrates and trace element supplements;

[0037] The timing of nutrient administration is determined based on the basal metabolic efficiency and nutrient absorption efficiency of each critically ill patient according to the clinical nutrient needs of each critically ill patient, so as to obtain the timing of nutrient administration for each critically ill patient.

[0038] Furthermore, the personalized nutritional therapy knowledge base optimization module includes the following functions:

[0039] Obtain the corresponding dietary intake food ingredients and dietary intake recording time through the dietary intake information of each critically ill patient during the corresponding period;

[0040] Determine the corresponding dietary intake action time segment according to the dietary intake recording time corresponding to each critically ill patient, and perform a nutritional administration interaction time segment analysis on the corresponding dietary intake action time segment based on the nutritional administration timing corresponding to each critically ill patient, so as to obtain the nutritional administration-dietary intake interaction time segment corresponding to each critically ill patient;

[0041] Based on the clinical nutritional needs of each critically ill patient and the nutritional administration-dietary intake interaction period, the nutritional administration conflict evaluation calculation formula is used to evaluate the nutritional administration conflict of the dietary intake food components corresponding to each critically ill patient, so as to obtain the clinical nutritional administration-dietary effect conflict ratio corresponding to each critically ill patient;

[0042] Based on the clinical nutrition administration-diet effect conflict ratio corresponding to each critically ill patient, the initial knowledge base of personalized nutrition therapy is updated and optimized to generate a personalized nutrition therapy knowledge optimization library corresponding to each critically ill patient, which includes personalized nutrition knowledge practice guidelines, nutrition administration plans, nutrition intake contraindications and dietary adjustment recommendations corresponding to critically ill patients.

[0043] Furthermore, the calculation formula for evaluating the conflict of nutrient administration is specifically as follows:

[0044] ;

[0045] In the formula, For critically ill patients The corresponding clinical nutrition administration-diet effect conflict ratio, is the total amount of clinical nutrients, The period of nutrient administration-dietary intake interaction, For the period When critically ill patients take The nutritional requirements corresponding to the clinical nutritional ingredients are For the period Severely ill patients need The dosage corresponding to the clinical nutritional ingredients, For the period When critically ill patients The drug-diet interaction coefficients corresponding to the clinical nutritional components, For the The relative contribution coefficients of the component conflicts corresponding to the clinical nutritional components, It is the correction factor for the conflict ratio of clinical nutrition administration and diet effect.

[0046] Furthermore, the present invention also provides a method for constructing a personalized nutritional therapy knowledge base for critically ill patients, which is used to execute the personalized nutritional therapy knowledge base construction system for critically ill patients as described above. The method for constructing a personalized nutritional therapy knowledge base for critically ill patients includes:

[0047] Step S1: Obtain the basic information, medical record information, laboratory test data and dietary intake information corresponding to the critically ill patients, and construct the critical case entity according to the basic information, medical record information, laboratory test data and dietary intake information corresponding to the critically ill patients, and generate the entity descriptors of each critically ill patient case;

[0048] Step S2: for each critically ill patient case entity descriptor and in combination with the semantic knowledge association ratio between relevant clinical medical research literature, an initial knowledge base of personalized nutritional therapy corresponding to each critically ill patient is constructed;

[0049] Step S3: Obtain clinical pathological characteristics and nutritional therapy knowledge entities corresponding to critically ill patients through the personalized nutritional therapy initial knowledge base, and perform nutritional demand analysis and determine medication timing based on the clinical pathological characteristics and nutritional therapy knowledge entities to obtain clinical nutritional needs and nutritional medication timing corresponding to each critically ill patient;

[0050] Step S4: Based on the clinical nutritional needs and nutritional administration timing of each critically ill patient, a nutritional administration conflict assessment is performed on the corresponding dietary intake information to obtain a clinical nutritional administration-dietary effect conflict ratio corresponding to each critically ill patient; based on the clinical nutritional administration-dietary effect conflict ratio corresponding to each critically ill patient, the nutritional therapy knowledge of the personalized nutritional therapy initial knowledge base is updated and optimized to generate a personalized nutritional therapy knowledge optimization base corresponding to each critically ill patient, which includes personalized nutritional knowledge practice guidelines, nutritional administration plans, nutritional intake contraindications and dietary adjustment recommendations corresponding to critically ill patients.

[0051] Beneficial effects of the present invention: The personalized nutritional therapy knowledge base construction system for critically ill patients proposed in the present invention has the beneficial effect of, compared with the prior art, comprehensively collecting various types of information of critically ill patients, including basic information, medical records, laboratory test results, and dietary intake during the period, thereby constructing a personalized and comprehensive case entity descriptor. The integration of this information lays the foundation for subsequent nutritional needs analysis. Specifically, basic information such as age, gender, weight, height, etc., helps to determine the basic physiological characteristics of an individual, while medical records and laboratory test data reveal key information such as the patient's health status, type and severity of the disease. By integrating and analyzing these data, the patient's main problems and needs can be identified, and then an accurate case entity descriptor, that is, the corresponding case entity feature vector, can be generated, which can reflect the patient's overall health status at a certain moment. Dietary intake information helps to understand the patient's nutritional intake and identify potential malnutrition or overnutrition problems. The key to this process is to be able to accurately tailor a case entity model for each critically ill patient, thereby avoiding the phenomenon of scattered data corresponding to critically ill patients. Secondly, by combining the semantic knowledge association ratio between the case entity descriptors of each critically ill patient and the relevant clinical medical research literature, an initial personalized nutritional treatment knowledge base corresponding to each critically ill patient is constructed. By comparing and associating the case entity descriptors with the relevant clinical medical research literature, an initial personalized nutritional treatment knowledge base is constructed, which can tailor relevant nutritional support plans for each critically ill patient. The core of this process is to form an initial knowledge base that includes the patient's pathological characteristics and nutritional treatment needs through medical literature, clinical research and the patient's actual condition information. The integration of literature data not only helps to identify the relationship between different diseases and nutritional support, but also introduces a large amount of knowledge about clinical nutrition, pathophysiology, disease progression, etc. This knowledge base not only provides a theoretical basis for the design of subsequent personalized nutritional treatment plans, but also provides reference and guidance for doctors or other clinical experts, so that the nutritional treatment plan can be more in line with the individual characteristics of the patient, thereby improving the effectiveness and efficiency of clinical nutritional intervention.Then, the clinical pathological characteristics and nutritional therapy knowledge entities corresponding to critically ill patients are obtained through the personalized nutritional therapy initial knowledge base, and nutritional needs analysis and drug administration timing are determined based on the clinical pathological characteristics and nutritional therapy knowledge entities. By utilizing the personalized nutritional therapy initial knowledge base, the clinical pathological characteristics and nutritional therapy knowledge entities related to the patient's condition can be accurately extracted. The key to this process is mainly reflected in the in-depth nutritional needs analysis by combining the patient's actual pathological data and existing medical knowledge. The clinical needs and physiological status of each critically ill patient are different. Through personalized analysis, the specific nutritional elements, supplementation timing and amount required by the patient can be determined to ensure the timeliness and effectiveness of nutritional intervention measures. At the same time, reasonable drug administration timing can effectively avoid the problem of mismatch between nutritional support plans and patients' physiological states, and maximize the optimization of patients' recovery paths. Through this analysis, personalized nutritional support plans can be formulated for patients, so that the process of nutritional therapy and disease recovery can be more closely integrated to achieve better clinical results, thereby better achieving real-time and accurate nutritional therapy adjustment processes. Finally, by evaluating the conflict of nutritional administration effects on the corresponding dietary intake information during the corresponding period based on the corresponding clinical nutritional needs and the timing of nutritional administration of each critically ill patient, the personalized nutritional treatment plan can be further optimized. Specifically, for each critically ill patient, combined with their clinical nutritional needs and the timing of administration, the interaction and conflict between nutritional administration and the patient's actual diet are evaluated. These conflicts may manifest as excessive or insufficient nutrients, poor effects caused by the interaction between dietary intake and drugs, and other problems. By evaluating the conflict ratio, the treatment plan can be adjusted and optimized to make nutritional treatment more precise and personalized. The optimization process is not just a simple elimination of conflicts, but also involves comprehensive adjustments to the patient's dietary intake and nutritional treatment, such as providing dietary compatibility recommendations and a list of taboo foods. The key to this process is to minimize the side effects caused by nutrition and diet mismatch and ensure that patients receive the best nutritional support and treatment effects. The generated personalized nutritional treatment knowledge optimization library will provide patients with a more scientific, reasonable and comprehensive practical guide for nutritional treatment and optimize the patient's overall rehabilitation plan. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments thereof made with reference to the following drawings:

[0053] Figure 1 A schematic diagram of the module flow of a system for constructing a personalized nutritional therapy knowledge base for critically ill patients of the present invention;

[0054] Figure 2 for Figure 1 Detailed functional flow chart of the entity description module for moderate and severe patients;

[0055] Figure 3 for Figure 1 Detailed functional flow diagram of the construction module of the nutrition knowledge base for moderate and severe disease entities. DETAILED DESCRIPTION

[0056] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0057] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0058] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0059] To achieve this, please refer to Figures 1 to 3 The present invention provides a system for constructing a personalized nutritional therapy knowledge base for critically ill patients, the system comprising the following modules:

[0060] The critically ill patient case entity description module is used to obtain the basic information, medical record information, laboratory test data and dietary intake information corresponding to the critically ill patients, and construct the critically ill case entity according to the basic information, medical record information, laboratory test data and dietary intake information corresponding to the critically ill patients, thereby generating each critically ill patient case entity descriptor;

[0061] The critical illness entity nutrition knowledge base construction module is used to construct the personalized nutrition treatment initial knowledge base corresponding to each critical illness patient based on the entity descriptors of each critical illness patient case and the semantic knowledge association ratio between relevant clinical medical research literature;

[0062] The module for determining nutritional needs and medication timing is used to obtain the clinical pathological characteristics and nutritional therapy knowledge entities corresponding to critically ill patients through the personalized nutritional therapy initial knowledge base, and to analyze nutritional needs and determine medication timing based on the clinical pathological characteristics and nutritional therapy knowledge entities, so as to obtain the clinical nutritional needs and nutritional medication timing corresponding to each critically ill patient;

[0063] The personalized nutrition therapy knowledge base optimization module is used to evaluate the nutrition administration conflict of the corresponding period dietary intake information based on the clinical nutrition needs and nutrition administration timing of each critically ill patient, so as to obtain the clinical nutrition administration-diet effect conflict ratio corresponding to each critically ill patient; based on the clinical nutrition administration-diet effect conflict ratio corresponding to each critically ill patient, the personalized nutrition therapy initial knowledge base is optimized for nutrition therapy knowledge update, so as to generate a personalized nutrition therapy knowledge optimization base corresponding to each critically ill patient, which includes personalized nutrition knowledge practice guidelines, nutrition administration schemes, nutrition intake compatibility contraindications and dietary adjustment suggestions corresponding to critically ill patients. In the embodiments of the present invention, please refer to Figure 1 As shown, it is a module flow diagram of the personalized nutrition therapy knowledge base construction system for critically ill patients of the present invention. In this example, the personalized nutrition therapy knowledge base construction system for critically ill patients includes the following modules:

[0064] S1: Critically ill patient case entity description module, used to obtain the basic information, medical record information, laboratory test data and dietary intake information corresponding to the critically ill patients, and construct the critical case entity according to the basic information, medical record information, laboratory test data and dietary intake information corresponding to the critically ill patients, and generate each critically ill patient case entity descriptor;

[0065] In the embodiment of the present invention, the basic information of critically ill patients is comprehensively collected, including the patient's age, gender, weight, height, past medical history, drug allergy history, etc. In addition, the patient's medical history information needs to be obtained, covering diagnosis, course of disease, treatment history, severity of current condition, complications and related medical operations, etc. The collection of laboratory test data includes hematological indicators (such as blood sugar, blood lipids, electrolytes, white blood cell count, etc.), biochemical indicators (such as liver function, kidney function, intestinal function, etc.), imaging examination results (such as X-ray, CT, MRI, etc.) and microbiological examination results (such as pathogen identification, antibiotic sensitivity, etc.). At the same time, it is also necessary to collect the patient's dietary intake information during hospitalization, including detailed data on the types, quantities, energy, protein, fat, carbohydrates and other components of daily food intake. By collecting and integrating the above types of data, an accurate entity descriptor of critically ill patients can be constructed. The descriptor can fully present the clinical characteristics of each patient, ensure data support for subsequent personalized nutritional treatment, and finally generate entity descriptors of each critically ill patient case.

[0066] S2: Critical illness entity nutrition knowledge base construction module, used to construct the personalized nutrition treatment initial knowledge base corresponding to each critical illness patient based on the entity descriptor of each critical illness patient case and the semantic knowledge association ratio between relevant clinical medical research literature;

[0067] In an embodiment of the present invention, after obtaining the patient's basic information, medical history information, laboratory test data and dietary intake information, it is necessary to associate this information with the semantic knowledge in the relevant clinical medical research literature. This process can be achieved through natural language processing (NLP) technology to automatically identify and extract nutritional therapy-related knowledge in clinical literature, such as nutritional intervention strategies for different critical diseases, nutrient requirements, nutritional support for common complications, etc., and through a knowledge graph based on the medical field, the case entity descriptor of each critically ill patient is associated with the nutritional therapy plan mentioned in these documents. Through computer-assisted reasoning and weighted integration of literature knowledge, an initial personalized nutritional therapy knowledge base for each patient is established. This knowledge base includes the patient's disease type, clinical pathological characteristics, nutritional therapy recommendations, known nutritional requirements, etc., while taking into account the patient's actual condition and dietary intake to ensure the accuracy of the personalized nutritional plan, and finally constructing a personalized nutritional therapy initial knowledge base corresponding to each critically ill patient.

[0068] S3: Nutritional requirement and medication timing determination module, which is used to obtain the clinical pathological characteristics and nutritional therapy knowledge entities corresponding to critically ill patients through the personalized nutritional therapy initial knowledge base, and to analyze nutritional requirements and determine medication timing based on the clinical pathological characteristics and nutritional therapy knowledge entities, so as to obtain the clinical nutritional requirements and nutritional medication timing corresponding to each critically ill patient;

[0069] In an embodiment of the present invention, through the personalized nutritional therapy initial knowledge base, the patient's clinical pathological characteristics are deeply analyzed to identify the patient's current nutritional needs, which includes scientific calculation of each patient's energy demand, protein demand, trace element demand, vitamin demand, etc., and determining the appropriate amount of nutritional supplement based on the patient's condition and treatment goals. In addition to the analysis of nutritional needs, it is also necessary to clarify the timing of nutritional administration based on the patient's course of disease, treatment progress and other physiological characteristics. For example, patients who are receiving intensive care need intravenous nutritional supplements, while patients who have not completely lost their gastrointestinal function can be intervened through enteral nutrition. Through the reasoning of medical models and clinical pathways, combined with the patient's individualized condition, the specific timing of nutritional supplementation is determined, and a specific nutritional treatment plan is formed, ultimately obtaining the corresponding clinical nutritional needs and nutritional administration timing for each critically ill patient.

[0070] S4: Personalized nutritional therapy knowledge base optimization module is used to evaluate the nutritional administration conflict of the corresponding dietary intake information based on the clinical nutritional needs and nutritional administration timing of each critically ill patient, so as to obtain the clinical nutritional administration-dietary effect conflict ratio corresponding to each critically ill patient; based on the clinical nutritional administration-dietary effect conflict ratio corresponding to each critically ill patient, the personalized nutritional therapy initial knowledge base is optimized to update the nutritional therapy knowledge, and a personalized nutritional therapy knowledge optimization library corresponding to each critically ill patient is generated, which includes personalized nutritional knowledge practice guidelines, nutritional administration plans, nutritional intake contraindications and dietary adjustment suggestions corresponding to critically ill patients.

[0071] In an embodiment of the present invention, after confirming the nutritional needs and medication timing of each patient, the next step is to conduct a nutritional medication-diet effect conflict assessment, which aims to analyze whether there is a potential conflict between the patient's current dietary intake and the nutritional therapy plan. For example, certain nutrients may interfere with each other's absorption (such as the interaction between calcium and iron), or there may be certain contraindicated foods in the patient's dietary intake that may have adverse reactions with the medication being administered. During the assessment process, the drug interaction database and the food nutrient component database may be combined to determine the possible conflicts between nutrients, drugs and foods through an algorithm, and calculate the clinical nutritional medication-diet effect conflict ratio for each patient. For patients with greater conflicts, it is recommended to adjust the diet plan or optimize the nutritional medication plan to ensure that the treatment effect is maximized. Based on this assessment result, the personalized nutritional therapy knowledge base is further optimized, and the patient's nutritional therapy knowledge base is updated through a machine learning algorithm, specifically including personalized nutritional therapy practice guidelines, nutritional medication plans, dietary compatibility taboos, and dietary adjustment suggestions. The generated optimized knowledge base can provide a scientific basis for clinicians to guide the personalized nutritional therapy of patients, ensure the synergistic effect of nutritional therapy and dietary intake is maximized, and avoid potential adverse reactions or treatment interference, and finally generate a personalized nutritional therapy knowledge optimization library corresponding to each critically ill patient.

[0072] Furthermore, the critically ill patient case entity description module includes the following functions:

[0073] Obtain the basic information of critically ill patients through the critically ill patient information database of the target clinical hospital, including the age, gender, height, weight, condition description and clinical manifestations of the critically ill patients;

[0074] Obtain the medical record information of critically ill patients through the electronic medical records of critically ill patients in the target clinical hospital, including the basic medical records, causes of critical illness, complications and medical history of critically ill patients;

[0075] Obtain laboratory test data corresponding to critically ill patients through the target clinical hospital's corresponding automated laboratory analysis system, including heart rate, blood sugar, blood indicators, and metabolic status of critically ill patients;

[0076] Obtain the corresponding dietary intake information of critically ill patients during treatment through the target clinical hospital's corresponding dietary record documents during treatment, including the corresponding dietary habits and dietary intake records of critically ill patients during treatment;

[0077] The basic information, medical record information, laboratory test data and dietary intake information corresponding to critically ill patients are standardized and the corresponding case entity feature description vectors are extracted to generate case entity descriptors for each critically ill patient.

[0078] As an embodiment of the present invention, refer to Figure 2 As shown, Figure 1 Detailed functional flow chart of the entity description module for moderate and severe patient cases. In this embodiment, the entity description module for severe patient cases includes the following functions:

[0079] S11: Obtain the basic information of critically ill patients through the critically ill patient information database of the target clinical hospital, including the age, gender, height, weight, condition description and clinical manifestations of the critically ill patients;

[0080] In an embodiment of the present invention, the critically ill patient information database of the target clinical hospital is accessed and it is ensured that the database has a relevant critically ill patient data storage structure. The database should contain the patient's basic personal information, such as age, gender, height, weight, condition description, and clinical manifestations. The basic information data of the specified patient is obtained by connecting to the hospital's database interface and using SQL query language or API calls. In specific operations, patient screening is first performed, such as data extraction based on the patient's admission record or critically ill patient identifier. Then, the extracted data needs to be stored in a structured manner to ensure the accuracy and completeness of the information. For example, fields such as the patient's age and gender are confirmed through data format verification. Special attention should be paid to the acquisition of condition description and clinical manifestation data, which can be obtained by extracting the condition description field in the medical record or the notes information in the medical record, and finally the basic information corresponding to the critically ill patient is obtained.

[0081] S12: Obtain the medical record information of critically ill patients through the electronic medical records of critically ill patients in the target clinical hospital, including the basic medical records, causes of critical illness, complications and medical history of critically ill patients;

[0082] In an embodiment of the present invention, by obtaining detailed medical record information of critically ill patients, including basic medical records, causes of critical illness, complications and medical history, the process first needs to query data through the hospital's electronic medical record management system (such as an EMR system) to obtain the patient's historical medical record records, and by accessing the hospital's electronic medical record database, use the patient's unique identifier (such as hospitalization number, medical record number) to query relevant medical records, and filter out some information related to critical illness. In specific implementation, the patient's medical history data can be extracted through a data interface call or by using a special medical record management tool (such as the HL7 standard, FHIR protocol, etc.). In the process of obtaining medical record information, the integrity of the record content must be ensured, and relevant information such as the diagnosis of the underlying disease, disease factors that may cause critical illness, description of complications, and the patient's past medical history must be extracted. The extracted data is cleaned and classified for subsequent processing and analysis, and finally the medical record information corresponding to the critically ill patient is obtained.

[0083] S13: Obtain laboratory test data corresponding to critically ill patients through the corresponding automated laboratory analysis system of the target clinical hospital, including heart rate, blood sugar, blood indicators and metabolic status of critically ill patients;

[0084] In an embodiment of the present invention, it is crucial to obtain laboratory test data of critically ill patients, especially experimental data related to the patient's heart rate, blood sugar, blood indicators and metabolic status. This process relies on the hospital's automated laboratory analysis system, which is usually connected to the hospital's medical information system (LIS) and stores the patient's laboratory test results. In order to obtain data, it is first necessary to query the patient's laboratory test records through an API interface or database connection, especially those key indicators that reflect the patient's physiological state and health status. Common laboratory data include but are not limited to blood routine, liver and kidney function, blood sugar level, electrocardiogram and other parameters. The system should support automatic extraction and formatted output of laboratory results. Next, the laboratory data needs to be sorted in chronological order, especially the laboratory test data of critically ill patients covers multiple time periods. Therefore, the date of each examination and the corresponding test results must be accurately marked, and the data must be correct, so as to finally obtain the corresponding laboratory test data of critically ill patients.

[0085] S14: Obtaining the dietary intake information of the critically ill patients during treatment through the dietary record documents of the target clinical hospital during treatment, including the dietary habits and dietary intake records of the critically ill patients during treatment;

[0086] In an embodiment of the present invention, by obtaining the dietary intake information of critically ill patients during treatment, it is first necessary to extract the patient's dietary records from the hospital's treatment record system or patient care system. These records are usually entered by nutritionists or nursing staff in the patient's medical records and stored in the electronic medical record system in the form of documents. By using standardized interfaces (such as the HL7 standard or the API of the hospital's own system), the dietary intake data of critically ill patients can be queried. The recorded information should include the patient's eating habits (such as food selection, intake, food type, etc.) and actual intake (such as three meals intake, special dietary needs, etc.). It should be noted that the dietary records will involve the patient's taste preferences, taboo foods, specific nutritional needs and other information, all of which should be completely extracted and stored in a classified manner. The dietary intake information will be organized in time series and associated with the patient's clinical treatment data to ensure the correlation between the dietary records and the patient's condition and treatment process, and finally obtain the corresponding period dietary intake information of the critically ill patients.

[0087] S15: Standardize the basic information, medical record information, laboratory test data, and dietary intake information corresponding to critically ill patients and extract the corresponding case entity feature description vector to generate each critically ill patient case entity descriptor.

[0088] In an embodiment of the present invention, various types of data previously extracted are standardized and feature extracted to generate case entity feature description vectors. The implementation of this requires the use of data processing and machine learning technologies. First, all data from different sources, such as basic information, medical records, laboratory test data, and dietary records, must be standardized in a unified format; for example, basic information such as age and gender should be converted into a unified numerical form, and descriptive text in medical record information (such as medical history and complications) should be word-vectorized using natural language processing technology or classified using predefined classification labels; laboratory data such as blood sugar, heart rate and other indicators must be classified in the same format. The dimensions of the diet records should be standardized to ensure that they are easy to compare on the same scale; the data in the dietary records should also be converted into digital form. For example, by converting the intake into quantitative indicators of calories or other nutrients, the standardized data can be used through feature extraction methods (such as PCA dimensionality reduction, t-SNE visualization, neural network models, etc.) to generate feature vectors of case entities. These feature vectors will cover the patient's basic physiological characteristics, clinical manifestations, laboratory data, medical history, dietary habits and other information to form a high-dimensional descriptor that can effectively describe the personalized situation of each critically ill patient, and ultimately generate a case entity descriptor for each critically ill patient.

[0089] Furthermore, the critical care entity nutrition knowledge base construction module includes the following functions:

[0090] Obtain relevant clinical medical research literature corresponding to each critically ill patient;

[0091] Extract semantic knowledge entities from the relevant clinical medical research literature corresponding to each critically ill patient to obtain the semantic knowledge entities corresponding to the relevant medical literature corresponding to each critically ill patient;

[0092] Conduct knowledge attribute mining and analysis based on the semantic knowledge entities corresponding to the relevant medical literature of each critically ill patient, and obtain the semantic knowledge attribute relationship corresponding to the relevant medical literature of each critically ill patient;

[0093] Based on the semantic knowledge attribute relationship of the relevant medical literature corresponding to each critically ill patient, the corresponding relevant clinical medical research literature is classified for the same research purpose, so as to obtain the medical literature set with the same research purpose of the semantic knowledge corresponding to each critically ill patient;

[0094] The semantic knowledge entities corresponding to different medical documents with the same research purpose are obtained through the medical document set with the same research purpose corresponding to each critically ill patient, and the knowledge association ratio of the semantic knowledge entities corresponding to different medical documents with the same research purpose is calculated based on the critically ill patient case entity descriptor, so as to obtain the semantic knowledge association ratio between each critically ill patient case entity and different medical documents with the same research purpose;

[0095] Based on the semantic knowledge association ratio, the knowledge entity representation association relationship between the critically ill patient case entity descriptor and the semantic knowledge entities corresponding to different medical literature with the same research purpose was determined to construct an initial knowledge base of personalized nutritional treatment corresponding to each critically ill patient.

[0096] As an embodiment of the present invention, refer to Figure 3 As shown, Figure 1 Detailed functional flow chart of the construction module of the medium and severe disease entity nutrition knowledge base. In this embodiment, the construction module of the severe disease entity nutrition knowledge base includes the following functions:

[0097] S21: Obtain relevant clinical medical research literature corresponding to each critically ill patient;

[0098] In an embodiment of the present invention, by determining the medical problems and treatment areas related to each critically ill patient, for example, for a patient with severe respiratory failure, the relevant clinical medical fields involved include respiratory support technology, mechanical ventilation, respiratory physiology, etc., by accessing professional medical databases (such as PubMed, Embase, Cochrane Library, etc.), a literature search is performed on the disease background of the critically ill patient, and keywords and medical subject terms (such as MeSH) are used to locate relevant literature. During the search process, the latest scientific research articles closely related to specific diseases, treatment plans, drug use, nutritional support, etc. are selected. During the search process, it is necessary to set a reasonable time interval and document type (such as randomized controlled trials, systematic reviews, clinical guidelines, etc.) to screen the most relevant documents to ensure that the obtained data is accurate and has reference value, and finally obtain the relevant clinical medical research documents corresponding to each critically ill patient.

[0099] S22: extracting semantic knowledge entities from the relevant clinical medical research literature corresponding to each critically ill patient, so as to obtain semantic knowledge entities corresponding to the relevant medical literature corresponding to each critically ill patient;

[0100] In an embodiment of the present invention, natural language processing (NLP) technology is used to perform text analysis on each document, and medical semantic knowledge entities therein are extracted. A trained medical named entity recognition (NER) model is used to identify and extract key medical terms and concepts in the document, including diseases, drugs, treatment methods, clinical symptoms, disease markers, treatment doses, patient groups, etc. The medical terms in each document are standardized, and terms with different expressions (such as the conversion between abbreviations and full names of certain disease names) are unified to ensure the consistency of information. In addition, knowledge graph-based extraction technology can also be used to assist in identifying proprietary terms and concepts related to critically ill patients through a domain knowledge base. The goal of this process is to convert the medical information in the document into standardized semantic knowledge entities, and finally obtain semantic knowledge entities corresponding to the relevant medical documents of each critically ill patient.

[0101] S23: performing knowledge attribute mining analysis based on the semantic knowledge entities corresponding to the relevant medical literature corresponding to each critically ill patient, and obtaining the semantic knowledge attribute relationship corresponding to the relevant medical literature corresponding to each critically ill patient;

[0102] In an embodiment of the present invention, knowledge attributes are mined and analyzed based on previously extracted semantic knowledge entities. In specific implementation, semantic association analysis can be used to identify the relationship between various medical entities, such as the therapeutic effect relationship between drugs and diseases, or the correlation between clinical symptoms and diseases. Graph analysis methods (such as graph neural networks, relationship extraction, etc.) are used to mine attribute relationships between entities in the literature, such as the effectiveness of "drug A treating disease B" or "the correlation between symptom C and disease D". In addition, a statistical clustering algorithm can be used to classify knowledge entities in the literature according to their attributes and relationships, thereby further revealing the key information structure in each document, forming a set of knowledge entities with clear attribute descriptions and their mutual relationships, and finally obtaining the semantic knowledge attribute relationships corresponding to the relevant medical literature of each critically ill patient.

[0103] S24: Based on the semantic knowledge attribute relationship corresponding to the relevant medical literature of each critically ill patient, the corresponding relevant clinical medical research literature is classified for the same research purpose, so as to obtain the set of medical literature with the same research purpose of the semantic knowledge corresponding to each critically ill patient;

[0104] In an embodiment of the present invention, all medical documents are classified and sorted according to their semantic knowledge attribute relationships. First, the research purpose and core issues of each document are analyzed. Through semantic analysis of the abstract and conclusion of the document, the research objectives of the document are identified, for example, whether it focuses on nutritional support for critically ill patients, drug treatment effects, complication prevention and treatment, etc., and then according to these research purposes, the documents are divided into multiple categories (such as nutritional therapy for critically ill patients, respiratory failure management, postoperative care, etc.), and through a comprehensive analysis of the keywords, research methods and conclusions of each document, they are classified into the classification that best suits its research purpose. In this process, text clustering algorithms (such as K-means, hierarchical clustering) are used in combination with the rules of experts in the medical field to ensure the accuracy and effectiveness of document classification. Each critically ill patient will have a group of related medical documents, and these documents have a high correlation under the same research objective, and finally a set of medical documents with the same research purpose and semantic knowledge corresponding to each critically ill patient is obtained.

[0105] S25: Obtain semantic knowledge entities corresponding to different medical documents with the same research purpose through the set of medical documents with the same research purpose corresponding to each critically ill patient, and calculate the knowledge association ratio of the semantic knowledge entities corresponding to different medical documents with the same research purpose based on the critically ill patient case entity descriptor, so as to obtain the semantic knowledge association ratio between each critically ill patient case entity and different medical documents with the same research purpose;

[0106] In an embodiment of the present invention, by calculating the knowledge association ratio between the clinical case descriptors of each critically ill patient and the semantic knowledge entities in the medical literature of the same research purpose, first, for each critically ill patient, the clinical case entity descriptors (including disease type, symptoms, past medical history, treatment methods, etc.) are extracted, and the case entity descriptors of the patient are compared with the semantic knowledge entities extracted from the medical literature of the same research purpose, and the similarity between the knowledge entities is calculated (such as by cosine similarity, Jaccard coefficient, etc.), and a correlation ratio is obtained based on the quantitative calculation of the knowledge entity similarity, which represents the degree of association between the clinical characteristics of the patient and the relevant knowledge entities in the literature. Through this process, the medical literature content that best matches the individual clinical characteristics of the patient can be identified, and finally the semantic knowledge association ratio between each critically ill patient case entity and different medical literature of the same research purpose is obtained.

[0107] S26: Based on the semantic knowledge association ratio, determine the knowledge entity representation association relationship between the critically ill patient case entity descriptor and the semantic knowledge entities corresponding to different medical literature with the same research purpose, so as to construct an initial knowledge base of personalized nutritional treatment corresponding to each critically ill patient.

[0108] In an embodiment of the present invention, a personalized nutritional therapy knowledge base is further constructed through the semantic knowledge association ratio. By summarizing the previously calculated association ratios, a knowledge entity representation association relationship is established between each critically ill patient and different medical documents. In specific implementation, a graph-based method can be used to map the case entity descriptors of critically ill patients and the knowledge entities in the medical documents into the same multidimensional knowledge graph. Each node represents a knowledge entity, and the edge represents the relationship between the entities. The weight of the edge is the association ratio between the knowledge entities. Through this process, a personalized nutritional therapy knowledge base is constructed. The base not only contains the individual characteristics of the patients, but also reflects the specific guiding significance of different medical documents for the treatment of patients. This knowledge base provides a basis for the subsequent design of personalized nutritional therapy plans, helps doctors formulate personalized intervention plans according to the specific conditions of critically ill patients, and finally constructs an initial personalized nutritional therapy knowledge base corresponding to each critically ill patient.

[0109] Furthermore, the calculation of the knowledge association ratio of the semantic knowledge entities corresponding to different medical documents with the same research purpose based on the entity descriptor of the critically ill patient case includes:

[0110] The semantic knowledge entities corresponding to different medical documents with the same research purpose are vectorized to obtain the knowledge entity description vectors corresponding to different medical documents with the same research purpose.

[0111] In an embodiment of the present invention, medical documents under the same research purpose are preprocessed to extract semantic knowledge entities therein, which include medical terms related to critically ill patients, such as diseases, symptoms, treatment methods, drugs, and nutritional requirements. Natural language processing (NLP) tools, such as SpaCy and BERT, are used to perform word segmentation and named entity recognition (NER) on the documents to identify the medical entities therein. Then, a word embedding-based method (such as Word2Vec and GloVe) or a more advanced semantic model (such as BERT and BioBERT) is used to convert each extracted knowledge entity into a high-dimensional vector. These vectors can capture the semantic information of the entity. For example, the BioBERT model can be used to obtain a vector representation of each entity based on its training in the medical field. The vectors of these entities can retain the semantic relationship between medical concepts. For example, the similarity between "diabetes" and "insulin" is high, while the similarity between "diabetes" and "antibiotics" is low. All semantic knowledge entities in each document will be converted into corresponding knowledge entity description vectors, thereby forming a knowledge vector representation of the document, and finally obtaining knowledge entity description vectors corresponding to different medical documents with the same research purpose.

[0112] Preferably, the semantic entity similarity is calculated using cosine similarity for the critically ill patient case entity descriptors and the knowledge entity description vectors corresponding to different medical documents with the same research purpose, so as to obtain the semantic entity similarity between each critically ill patient case entity and different medical documents with the same research purpose;

[0113] In an embodiment of the present invention, relevant entity descriptors are extracted from critically ill patient cases. These descriptors include information such as the patient's symptoms, diseases, medical history, current treatment plan, nutritional needs, etc. These information are also subjected to entity recognition and vectorization through natural language processing technology to obtain a semantic description vector of each entity in the patient case. Then, the cosine similarity formula is used to calculate the similarity between the entity description vector of the critically ill patient case and the vectors of semantic knowledge entities in different medical literature. Specifically, assuming that an entity vector in the critically ill patient case is , the semantic entity vector in a document is , then the cosine similarity calculation formula is: , is the dot product of two vectors, and are the modulus lengths of the vectors respectively. By calculating the cosine similarity between all critically ill patient case entities and medical literature entities, the semantic entity similarity between each patient case and different documents can be obtained. This process can help determine which knowledge entities in the documents are most relevant to the condition, symptoms, and treatment needs of specific patients, and ultimately the semantic entity similarity between each critically ill patient case entity and different medical documents with the same research purpose can be obtained.

[0114] Preferably, the research field impact assessment analysis is performed on the semantic knowledge entities corresponding to different medical documents with the same research purpose to obtain the research field impact degree of the semantic knowledge corresponding to different medical documents with the same research purpose;

[0115] In an embodiment of the present invention, by evaluating the influence of semantic knowledge entities in different medical documents under the same research purpose in the research field, the evaluation of this influence can be based on factors such as the number of citations of the document, the impact factor of the journal in which the document is published, and the publication time of the document. In specific implementation, the number of citations and journal impact factors of each document can be obtained by consulting databases (such as PubMed, Google Scholar) to obtain indicators such as the number of citations and journal impact factors of each document, thereby quantifying the academic influence of each document. In addition, text mining technology can also be used to analyze the frequency of occurrence and citation frequency of semantic knowledge entities in the document in the same field. For example, if the nutritional treatment plan or disease treatment method involved in a certain document is widely cited in other studies, then the influence of the semantic knowledge entity corresponding to the document can be evaluated as high. By combining these influence evaluation indicators, the influence of the semantic knowledge entity corresponding to each document in the same research field can be calculated, and finally the influence of the semantic knowledge research field corresponding to different medical documents with the same research purpose can be obtained.

[0116] Preferably, based on the semantic entity similarity between each critically ill patient case entity and different medical documents with the same research purpose and the corresponding semantic knowledge research field influence degree, the knowledge association ratio of the semantic knowledge entities corresponding to different medical documents with the same research purpose is calculated using the semantic knowledge association ratio calculation formula to obtain the semantic knowledge association ratio between each critically ill patient case entity and different medical documents with the same research purpose.

[0117] In an embodiment of the present invention, a suitable semantic knowledge association ratio calculation formula is formed by combining the semantic association calculation time window, time variable parameters, the total number of critically ill patient cases, the total number of medical documents corresponding to the same research purpose, the semantic entities corresponding to the critically ill patient cases, the semantic entities corresponding to the medical documents, the semantic entity similarity, the semantic similarity association influence weight coefficient, the semantic knowledge research field related measurement, the degree of influence of the semantic knowledge research field, the semantic knowledge initial time, the time decay rate control parameter and related parameters to calculate the knowledge association ratio of the semantic knowledge entities corresponding to different medical documents for the same research purpose, so as to determine the knowledge association ratio between each critically ill patient case entity and different medical documents, and finally obtain the semantic knowledge association ratio between each critically ill patient case entity and different medical documents for the same research purpose.

[0118] Furthermore, the calculation formula of the semantic knowledge association ratio is specifically:

[0119] ;

[0120] In the formula, For severe cases Corresponding medical literature with the same research purpose The semantic knowledge correlation ratio between Calculate the time window for semantic association, is the time variable parameter, is the total number of severe cases. The total number of corresponding medical literature for the same research purpose, For severe cases The corresponding semantic entity, For medical literature The corresponding semantic entity, For severe cases Middle Semantic Entities and Medical Literature Middle The semantic entity similarity between semantic entities, is the influence weight coefficient of semantic similarity association, For in time Cases of critically ill patients Middle Semantic Entities and Medical Literature Middle The semantic knowledge research field related measurement between semantic entities, The impact of semantic knowledge research field, is an exponential function, is the initial time of semantic knowledge, is the time decay rate control parameter, is the correction coefficient of the semantic knowledge association ratio.

[0121] The present invention obtains a semantic knowledge association ratio calculation formula by using a specific mathematical model and verifying it, which is used to calculate the knowledge association ratio of semantic knowledge entities corresponding to different medical documents with the same research purpose. The semantic knowledge association ratio calculation formula can quantify the similarity between semantic entities in critically ill patient cases and semantic entities in medical documents by calculating the similarity between them at the content level, which helps to identify which information in medical documents is more relevant to critically ill patient cases and improves the accuracy of the research. The introduction of time-varying factors and the consideration of the relevance of knowledge entities in different fields can help identify which research areas in medical literature have a greater impact on the understanding and treatment of critically ill patients' cases, improving the comprehensive interpretation of cases across literature. By introducing dynamic time factors, including time variables and exponential decay functions , this formula can capture the changes in the correlation of knowledge entities at different time points. The correlation between medical research and severe cases often changes over time. This time-sensitive modeling method can effectively reflect the development trend of medical knowledge or the replacement of treatment methods. The time decay coefficient makes the contribution of older knowledge entities to the correlation ratio gradually decrease, which helps to emphasize the applicability of recent and updated literature to current severe cases. By introducing weight coefficients, the influence of different factors can be adjusted. Specifically, controls the weight of semantic similarity in the association ratio, and The degree of influence of the research field is controlled. Through these weight coefficients, researchers can make adjustments according to different research purposes or the specific context of the data to optimize the relevance and accuracy of the results. This formula takes into account the combined relationship between critical cases and multiple semantic entities in medical literature. This processing method can effectively reflect the complex relationship between multiple medical literature and multiple case entities, rather than just isolated individual comparisons. The formula not only focuses on the match between a single case and a document, but also involves multi-level association measurements across documents and cases, which helps to comprehensively consider the complex interactions between multiple research fields, multiple cases and multiple documents, and reveal potential implicit associations. In addition, the introduction of correction coefficients can adjust the semantic knowledge association ratio, which is very helpful for compensating for errors in the model, optimizing calculation results and improving the robustness of the model, especially when dealing with complex data. It can avoid the excessive impact of extreme values ​​on the results. In summary, this formula fully takes into account the cases of critically ill patients. Corresponding medical literature with the same research purpose Semantic knowledge correlation ratio , semantic association calculation time window , time variable parameter , the total number of severe cases , the total number of medical literature corresponding to the same research purpose , severe cases The corresponding Semantic Entity , Medical Literature The corresponding Semantic Entity , severe cases Middle Semantic Entities and Medical Literature Middle The semantic entity similarity between semantic entities , semantic similarity association influence weight coefficient , at time Cases of critically ill patients Middle Semantic Entities and Medical Literature Middle Semantic knowledge research field related measurement between semantic entities , the impact of semantic knowledge research field , exponential function , semantic knowledge initial time , time decay rate control parameter , correction coefficient of semantic knowledge association ratio According to the cases of severe patients Corresponding medical literature with the same research purpose Semantic knowledge correlation ratio The correlation between the above parameters constitutes a functional relationship This formula can realize the calculation process of the knowledge association ratio of semantic knowledge entities corresponding to different medical documents with the same research purpose. At the same time, the correction coefficient of the semantic knowledge association ratio is The introduction of can be adjusted according to the errors that occur in the calculation process, thereby improving the accuracy and applicability of the calculation formula for the semantic knowledge association ratio.

[0122] Furthermore, the determination of the knowledge entity representation association relationship between the critically ill patient case entity descriptor and the semantic knowledge entities corresponding to different medical documents with the same research purpose based on the semantic knowledge association ratio includes:

[0123] The semantic knowledge association ratio is compared and judged according to a preset semantic knowledge association threshold. If the semantic knowledge association ratio is less than the preset semantic knowledge association threshold, it is determined that there is no knowledge entity representation association relationship between the critically ill patient case entity descriptor and the semantic knowledge entity corresponding to the current medical document with the same research purpose, and the semantic knowledge association ratio corresponding to the next medical document is compared and judged;

[0124] In an embodiment of the present invention, by utilizing a natural language processing tool (such as a BERT model or other semantic analysis tools) to parse entity descriptors of critically ill patient cases in medical literature, medical semantic entities related to the research purpose are extracted, including symptoms, treatment methods, symptoms, drugs, etc. Then, by calculating the semantic knowledge association ratio between the critically ill patient case descriptors and the semantic entities in the current medical literature, a numerical result is obtained. This association ratio reflects the degree of semantic matching between the case descriptors and the relevant entities in the literature, and the calculated semantic knowledge association ratio is compared with a preset semantic knowledge association threshold. If the association ratio is less than the threshold, it can be determined that the current case descriptor does not have sufficient association with the entity in the medical literature, and thus does not constitute a valid knowledge entity representation association relationship. At this time, it is necessary to skip the current document and proceed to compare and judge the next document. If the association ratio meets the threshold condition, proceed to the next step.

[0125] Preferably, if the semantic knowledge association ratio is greater than or equal to a preset semantic knowledge association threshold, it is determined that there is a knowledge entity representation association relationship between the critically ill patient case entity descriptor and the semantic knowledge entity corresponding to the current medical literature with the same research purpose, and the corresponding semantic knowledge association ratio is used as the edge weight of the association relationship, so as to construct an initial knowledge base of personalized nutritional treatment corresponding to each critically ill patient based on the knowledge entity representation association relationship and the corresponding edge weight.

[0126] In an embodiment of the present invention, if the corresponding semantic knowledge association ratio is greater than or equal to a preset semantic knowledge association threshold, it means that there is a strong knowledge association relationship between the current critically ill patient case entity descriptor and the relevant semantic entity in the medical literature. In this step, the relationship between the case descriptor of the critically ill patient and the semantic entity in the literature is first confirmed, and the calculated semantic knowledge association ratio is used as the edge weight of the association relationship. The weight value reflects the strength and reliability of the association relationship. When constructing the initial knowledge base of personalized nutritional therapy, the relationship between all qualified cases and entities in the literature will be used as part of the knowledge graph, and a multi-dimensional, personalized nutritional therapy knowledge base will be constructed through the associated edges and their weights. At this time, the knowledge base contains the mapping relationship between different symptoms and pathophysiological states of critically ill patients and nutritional therapy recommendations in relevant medical literature. This initial knowledge base will play a core role in the subsequent formulation of personalized treatment plans, provide a basis for treatment decisions for different patients, and finally obtain a personalized nutritional therapy initial knowledge base corresponding to critically ill patients.

[0127] Furthermore, the module for determining nutritional needs and medication timing includes the following functions:

[0128] Obtain the clinical pathological characteristics of critically ill patients through the corresponding critically ill patient case entities in the initial knowledge base of personalized nutritional therapy, including clinical case types, clinical pathophysiological performance indicators, patient basal metabolic efficiency, and nutrient absorption efficiency;

[0129] In an embodiment of the present invention, the clinical medical record data of each critically ill patient is first retrieved through the personalized nutritional therapy initial knowledge base. These data usually come from the patient's electronic health record (EHR) system, medical record files or medical database. The patient's case entity information is extracted from structured or unstructured data through semantic parsing and natural language processing technology. Specifically, the patient's clinical case type (such as traumatic shock, acute respiratory distress syndrome, etc.) and clinical pathophysiological performance indicators (such as blood sugar level, body temperature, blood pressure, liver and kidney function, etc.) are identified and extracted from the case, and the patient's basal metabolic efficiency data (usually measured by clinical basal metabolic rate BMR, or estimated by prediction formula) and nutrient absorption efficiency evaluation indicators (such as gastrointestinal function, intestinal absorption capacity, the presence or absence of digestive system diseases, etc.) are further obtained, and finally the corresponding clinical pathological characteristics of the critically ill patients are obtained.

[0130] Obtain the nutritional therapy knowledge entities corresponding to critically ill patients through the semantic knowledge entities corresponding to the relevant medical literature in the initial knowledge base of personalized nutritional therapy, including the nutritional supplement ingredients corresponding to critically ill patients at various clinical stages and the expected clinical nutritional effects;

[0131] In an embodiment of the present invention, semantic knowledge entities in relevant medical literature are obtained from an initial knowledge base of personalized nutritional therapy, and specific knowledge related to nutritional therapy for critically ill patients is deeply mined. Through professional medical knowledge graph construction technology, various nutrients that should be supplemented for critically ill patients in different clinical stages (such as acute stage, recovery stage, post-discharge stage, etc.) are systematically extracted from medical literature, clinical guidelines and expert consensus. These nutrients include but are not limited to protein, fat, carbohydrates, vitamins, trace elements, etc. Through data mining and natural language processing methods, nutritional therapy recommendations with high relevance are identified from a large amount of medical literature, and are associated with clinical stages and patient characteristics to form knowledge entities for nutritional therapy. These nutritional therapy knowledge entities will help build a personalized nutritional therapy plan for each critically ill patient, and ultimately obtain nutritional therapy knowledge entities corresponding to the critically ill patient.

[0132] According to the clinical case types and clinical pathophysiological indicators of critically ill patients, the corresponding nutritional treatment knowledge entities are analyzed for nutritional needs, so as to analyze and calculate the clinical nutritional needs of each critically ill patient, including the specific dosage requirements of protein, fat, carbohydrates and trace element supplements;

[0133] In an embodiment of the present invention, a clinical nutritional needs analysis of critically ill patients is performed by combining the patient's clinical case type, pathophysiological performance indicators, and extracted nutritional therapy knowledge entities. Specifically, first, according to the patient's clinical case type, such as acute respiratory distress syndrome (ARDS) or severe trauma, the nutritional therapy requirements corresponding to the disease are identified. Then, through a calculation model, the patient's nutritional needs are analyzed based on pathophysiological indicators (such as metabolic rate, course of disease, weight change, etc.), especially the quantitative needs of protein, fat, and carbohydrates. For example, for a severe trauma patient, more protein may be needed to support wound healing and muscle repair. At the same time, the supplemented trace elements (such as zinc, selenium, etc.) are also calculated according to the patient's condition and nutritional absorption capacity. The nutritional needs calculation uses a special nutritional needs model (such as REE formula, Mifflin-St Jeor equation, etc.) to accurately estimate the patient's energy and nutritional needs, ensuring that each critically ill patient can obtain a personalized nutritional supplement plan, and finally calculate the clinical nutritional needs corresponding to each critically ill patient.

[0134] The timing of nutrient administration is determined based on the basal metabolic efficiency and nutrient absorption efficiency of each critically ill patient according to the clinical nutrient needs of each critically ill patient, so as to obtain the timing of nutrient administration for each critically ill patient.

[0135] In an embodiment of the present invention, the timing of nutritional administration for critically ill patients is determined based on the clinical nutritional needs calculated in the previous step, as well as the patient's basal metabolic efficiency and nutritional absorption efficiency. First, the patient's current energy consumption level and metabolic rate are evaluated based on their basal metabolic efficiency. The patient's metabolic status is measured by combining clinical data (such as body temperature, blood oxygen level, etc.) with metabolic monitoring equipment (such as a dynamic metabolic monitor) to further refine the timing of nutritional administration. The evaluation of nutritional absorption efficiency is based on indicators such as the patient's gastrointestinal function and intestinal absorption capacity. In particular, in patients with digestive and absorption disorders, it is necessary to adjust the nutritional supplement method and timing based on their gastrointestinal emptying time, digestive and absorption capacity, etc. On this basis, the optimal timing of nutritional administration is determined through an optimization algorithm (such as time window optimization, dynamic adjustment algorithm) to maximize the effect of nutritional intervention, ensure that patients receive the best nutritional support during treatment, and ultimately obtain the corresponding nutritional administration timing for each critically ill patient.

[0136] Furthermore, the personalized nutritional therapy knowledge base optimization module includes the following functions:

[0137] Obtain the corresponding dietary intake food ingredients and dietary intake recording time through the dietary intake information of each critically ill patient during the corresponding period;

[0138] In an embodiment of the present invention, by obtaining detailed dietary intake data of critically ill patients, which usually come from the patient's electronic health record (EHR) system or through the records of nutritionists and nursing staff, covering the patient's meal content within a specific time period, each food intake information includes the food name, food amount, intake time and corresponding food ingredient information (such as energy, protein, fat, carbohydrates, minerals, etc.), by integrating this information, the dietary intake of each critically ill patient is classified and archived based on the recording time. In this process, the food ingredients need to be analyzed in detail to ensure that the nutritional components of each food are connected to its standard database, so that the ingredient data of each meal can be accurately calibrated. At the same time, the dietary intake timestamp related to the patient will be automatically extracted, and the time of each intake behavior will be recorded, and finally the corresponding dietary intake food ingredients and dietary intake recording time will be obtained.

[0139] Determine the corresponding dietary intake action time segment according to the dietary intake recording time corresponding to each critically ill patient, and perform a nutritional administration interaction time segment analysis on the corresponding dietary intake action time segment based on the nutritional administration timing corresponding to each critically ill patient, so as to obtain the nutritional administration-dietary intake interaction time segment corresponding to each critically ill patient;

[0140] In an embodiment of the present invention, the dietary intake information is divided into different time segments according to the patient's dietary intake recording time. These time segments are subdivided according to the dietary intake patterns and disease status of critically ill patients, including morning, lunch, dinner, and snack time periods. Next, the corresponding nutritional administration time is determined according to the patient's clinical treatment plan. These nutritional administration times are usually set by doctors or nutrition teams according to the patient's clinical needs, including daily scheduled oral nutritional supplements, enteral or parenteral nutritional supplements, etc. For each patient, the dietary intake time is compared with the nutritional administration time, and the overlap between the two is analyzed through an algorithm to calculate the interaction period between the dietary intake action time segment and the nutritional administration action time segment. The interaction period analysis helps to identify possible conflicts or synergistic effects, and finally obtains the nutritional administration-dietary intake interaction period corresponding to each critically ill patient.

[0141] Based on the clinical nutritional needs of each critically ill patient and the nutritional administration-dietary intake interaction period, the nutritional administration conflict evaluation calculation formula is used to evaluate the nutritional administration conflict of the dietary intake food components corresponding to each critically ill patient, so as to obtain the clinical nutritional administration-dietary effect conflict ratio corresponding to each critically ill patient;

[0142] In an embodiment of the present invention, a suitable nutritional administration conflict assessment calculation formula is formed by combining the nutritional administration-dietary intake interaction period, the nutritional requirement corresponding to the clinical nutritional components, the dosage corresponding to the clinical nutritional components, the administration-diet interaction coefficient, the relative contribution coefficient of the component conflict and related parameters to evaluate the nutritional administration conflict of the food components in the dietary intake corresponding to each critically ill patient, so as to evaluate whether the various food components in the dietary intake conflict with the timing of administration. For example, if the intake time of a certain food overlaps with the optimal absorption time of a certain drug, it will lead to reduced efficacy or adverse reactions. The conflict ratio between nutritional administration and dietary intake of each critically ill patient will be calculated, which reflects the optimization space of nutritional administration and dietary intake, and finally the clinical nutritional administration-dietary effect conflict ratio corresponding to each critically ill patient is obtained.

[0143] Based on the clinical nutrition administration-diet effect conflict ratio corresponding to each critically ill patient, the initial knowledge base of personalized nutrition therapy is updated and optimized to generate a personalized nutrition therapy knowledge optimization library corresponding to each critically ill patient, which includes personalized nutrition knowledge practice guidelines, nutrition administration plans, nutrition intake contraindications and dietary adjustment recommendations corresponding to critically ill patients.

[0144] In an embodiment of the present invention, the personalized nutrition therapy knowledge base is optimized and updated based on the previously quantified conflict ratio between clinical nutrition administration and dietary effects. First, the knowledge base is updated based on critically ill patients with a high conflict ratio, and combined with the latest nutrition research and clinical treatment plans, more accurate nutrition administration timing and dietary intake recommendations are provided for each patient. These updates will form a personalized nutrition therapy knowledge optimization library for patients, including a personalized nutrition knowledge practice guide, which provides specific dietary guidance, details the nutritional needs of each critically ill patient, the types of food intake and their timing, so as to avoid conflicts with the timing of drug administration; nutrition administration The system will recommend the most suitable timing, method and dosage of nutritional administration according to the patient's clinical status to ensure that the drugs and nutrients are absorbed to the maximum extent and exert their therapeutic effects. For each patient, the system will mark the foods or nutrients that should not be consumed at the same time and the contraindications between drugs. For example, some minerals (such as calcium and iron) will interact with specific drugs and affect their efficacy. For dietary adjustment suggestions, specific suggestions for dietary adjustments will be given based on the conflict ratio. Suggestions include changing meal times, changing food types, etc., so as to provide patients with a diet that is more suitable for their treatment plan, and finally generate a personalized nutritional treatment knowledge optimization library corresponding to each critically ill patient.

[0145] Furthermore, the calculation formula for evaluating the conflict of nutrient administration is specifically as follows:

[0146] ;

[0147] In the formula, For critically ill patients The corresponding clinical nutrition administration-diet effect conflict ratio, is the total amount of clinical nutrients, The period of nutrient administration-dietary intake interaction, For the period When critically ill patients take The nutritional requirements corresponding to the clinical nutritional ingredients are For the period Severely ill patients need The dosage corresponding to the clinical nutritional ingredients, For the period When critically ill patients The drug-diet interaction coefficients corresponding to the clinical nutritional components, For the The relative contribution coefficients of the component conflicts corresponding to the clinical nutritional components, It is the correction factor for the conflict ratio of clinical nutrition administration and diet effect.

[0148] The present invention obtains a nutritional administration conflict evaluation calculation formula by using a specific mathematical model and verifying it, which is used to evaluate the nutritional administration conflict of the food components of the dietary intake corresponding to each critically ill patient. The nutritional administration conflict evaluation calculation formula involves quantitative analysis of the interaction between the dietary intake of critically ill patients and clinical nutritional administration. Its purpose is to optimize the patient's personalized nutritional treatment plan by evaluating the possible conflicts between the overlapping parts of the administration and dietary intake time periods of different nutrients, and ensure the effect and safety of nutritional treatment. By quantifying the conflict between the dietary intake and the timing of administration of each critically ill patient, it can help doctors better identify potential nutritional conflicts, which makes clinical nutritional treatment more accurate, avoids the conflict between unreasonable nutritional intake and improper administration time, and compares the demand and actual administration amount of each clinical nutrient component, thereby evaluating the overlap or mutual interference between the dietary intake time period and the administration time period, and obtaining a conflict ratio. This ratio provides doctors with specific and quantitative indicators to help make reasonable decisions during the treatment process. This formula involves multiple variables, including the amount of clinical nutrients that critically ill patients need to consume in a certain period of time, the amount of medication that critically ill patients need during this period, the interaction coefficient between medication and dietary intake, and the relative contribution coefficient of each clinical nutrient in the conflict, which reflects the degree of influence of the component in the total conflict, taking into account the mutual influence between different food components and nutritional medication. By combining these factors, the formula can accurately evaluate the interaction between nutritional medication and dietary intake, and determine which time periods and nutrients will cause inappropriate conflicts. The n in this formula represents the total number of clinical nutrients. For each nutrient, there are independent evaluation criteria, allowing doctors to analyze the impact of different components in multiple dimensions. Each component has a different mechanism of action and interaction with other components. Therefore, individual calculations can effectively avoid adverse interactions between certain food components and drugs. In summary, this formula fully takes into account the needs of critically ill patients. Corresponding clinical nutrition administration-diet effect conflict ratio , the total number of clinical nutritional ingredients , Nutritional administration-dietary intake interaction period , in the period When critically ill patients take Nutritional requirements corresponding to clinical nutritional ingredients , in the period Severely ill patients need Dosage of clinical nutritional ingredients , in the period When critically ill patients Drug-diet interaction coefficients corresponding to clinical nutritional components , No. Relative contribution coefficients of ingredient conflicts corresponding to clinical nutritional ingredients , Correction coefficient of clinical nutrition administration-diet effect conflict ratio According to the critically ill patients Corresponding clinical nutrition administration-diet effect conflict ratio The correlation between the above parameters constitutes a functional relationship This formula can realize the nutritional administration conflict assessment process of the food ingredients in the diet intake of each critically ill patient. At the same time, the correction coefficient of the clinical nutritional administration-dietary effect conflict ratio is used. The introduction of can be adjusted according to the errors that occur in the calculation process, thereby improving the accuracy and applicability of the calculation formula for evaluating the conflict of nutrient administration.

[0149] Furthermore, the present invention also provides a method for constructing a personalized nutritional therapy knowledge base for critically ill patients, which is used to execute the personalized nutritional therapy knowledge base construction system for critically ill patients as described above. The method for constructing a personalized nutritional therapy knowledge base for critically ill patients includes:

[0150] Step S1: Obtain the basic information, medical record information, laboratory test data and dietary intake information corresponding to the critically ill patients, and construct the critical case entity according to the basic information, medical record information, laboratory test data and dietary intake information corresponding to the critically ill patients, and generate the entity descriptors of each critically ill patient case;

[0151] Step S2: for each critically ill patient case entity descriptor and in combination with the semantic knowledge association ratio between relevant clinical medical research literature, an initial knowledge base of personalized nutritional therapy corresponding to each critically ill patient is constructed;

[0152] Step S3: Obtain clinical pathological characteristics and nutritional therapy knowledge entities corresponding to critically ill patients through the personalized nutritional therapy initial knowledge base, and perform nutritional demand analysis and determine medication timing based on the clinical pathological characteristics and nutritional therapy knowledge entities to obtain clinical nutritional needs and nutritional medication timing corresponding to each critically ill patient;

[0153] Step S4: Based on the clinical nutritional needs and nutritional administration timing of each critically ill patient, a nutritional administration conflict assessment is performed on the corresponding dietary intake information to obtain a clinical nutritional administration-dietary effect conflict ratio corresponding to each critically ill patient; based on the clinical nutritional administration-dietary effect conflict ratio corresponding to each critically ill patient, the nutritional therapy knowledge of the personalized nutritional therapy initial knowledge base is updated and optimized to generate a personalized nutritional therapy knowledge optimization base corresponding to each critically ill patient, which includes personalized nutritional knowledge practice guidelines, nutritional administration plans, nutritional intake contraindications and dietary adjustment recommendations corresponding to critically ill patients.

[0154] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0155] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for constructing a personalized nutritional therapy knowledge base for critically ill patients, characterized in that: The following steps are involved: Step S1: Obtain the basic information, medical record information, laboratory test data and dietary intake information corresponding to the critically ill patients, and construct the critical case entity according to the basic information, medical record information, laboratory test data and dietary intake information corresponding to the critically ill patients, and generate the entity descriptors of each critically ill patient case; Step S2: for each critically ill patient case entity descriptor and in combination with the semantic knowledge association ratio between relevant clinical medical research literature, an initial knowledge base of personalized nutritional therapy corresponding to each critically ill patient is constructed; wherein step S2 includes the following steps: Step S21: Obtain relevant clinical medical research literature corresponding to each critically ill patient; Step S22: extracting semantic knowledge entities from the relevant clinical medical research literature corresponding to each critically ill patient, so as to obtain semantic knowledge entities corresponding to the relevant medical literature corresponding to each critically ill patient; Step S23: performing knowledge attribute mining analysis based on the semantic knowledge entities corresponding to the relevant medical documents corresponding to each critically ill patient, and obtaining the semantic knowledge attribute relationship corresponding to the relevant medical documents corresponding to each critically ill patient; Step S24: Based on the semantic knowledge attribute relationship corresponding to the relevant medical literature corresponding to each critically ill patient, the corresponding relevant clinical medical research literature is classified for the same research purpose, so as to obtain a set of medical literature with the same research purpose of semantic knowledge corresponding to each critically ill patient; Step S25: obtaining semantic knowledge entities corresponding to different medical documents with the same research purpose through the set of medical documents with the same research purpose corresponding to each critically ill patient, and calculating the knowledge association ratio of the semantic knowledge entities corresponding to different medical documents with the same research purpose based on the critically ill patient case entity descriptor, so as to obtain the semantic knowledge association ratio between each critically ill patient case entity and different medical documents with the same research purpose; wherein step S25 includes the following steps: The semantic knowledge entities corresponding to different medical documents with the same research purpose are vectorized to obtain the knowledge entity description vectors corresponding to different medical documents with the same research purpose. The cosine similarity is used to calculate the semantic entity similarity between the critically ill patient case entity descriptors and the knowledge entity description vectors corresponding to different medical documents with the same research purpose, so as to obtain the semantic entity similarity between each critically ill patient case entity and different medical documents with the same research purpose; Conduct research field impact assessment analysis on the semantic knowledge entities corresponding to different medical documents with the same research purpose, and obtain the research field impact of the semantic knowledge entities corresponding to different medical documents with the same research purpose; Based on the semantic entity similarity between each critically ill patient case entity and different medical documents with the same research purpose and the corresponding semantic knowledge research field influence degree, the semantic knowledge entity corresponding to different medical documents with the same research purpose is calculated using the semantic knowledge association ratio calculation formula to obtain the semantic knowledge association ratio between each critically ill patient case entity and different medical documents with the same research purpose; wherein, the semantic knowledge association ratio calculation formula is specifically: ； In the formula, For severe cases Corresponding medical literature with the same research purpose The semantic knowledge correlation ratio between Calculate the time window for semantic association, is the time variable parameter, is the total number of severe cases. The total number of corresponding medical literature for the same research purpose, For severe cases The corresponding semantic entity, For medical literature The corresponding semantic entity, For severe cases Middle Semantic Entities and Medical Literature Middle The semantic entity similarity between semantic entities, is the semantic similarity association influence weight coefficient, For in time Cases of critically ill patients Middle Semantic Entities and Medical Literature Middle The semantic knowledge research field related measurement between semantic entities, is the impact degree of semantic knowledge research field, is an exponential function, is the initial time of semantic knowledge, is the time decay rate control parameter, is the correction coefficient of semantic knowledge association ratio; Step S26: determining the knowledge entity representation association relationship between the critically ill patient case entity descriptor and the semantic knowledge entities corresponding to different medical documents with the same research purpose based on the semantic knowledge association ratio, so as to construct an initial knowledge base of personalized nutritional therapy corresponding to each critically ill patient; Step S3: Obtain clinical pathological characteristics and nutritional therapy knowledge entities corresponding to critically ill patients through the personalized nutritional therapy initial knowledge base, and perform nutritional demand analysis and determine the timing of drug administration based on the clinical pathological characteristics and nutritional therapy knowledge entities, so as to obtain clinical nutritional needs and nutritional drug administration timing corresponding to each critically ill patient; wherein, step S3 includes the following steps: Step S31: Obtaining clinical pathological characteristics corresponding to critically ill patients through corresponding critically ill patient case entities in the personalized nutritional therapy initial knowledge base, including clinical case types, clinical pathophysiological performance indicators, patient basal metabolic efficiency, and nutrient absorption efficiency; Step S32: Obtaining nutritional therapy knowledge entities corresponding to critically ill patients through semantic knowledge entities corresponding to relevant medical literature in the personalized nutritional therapy initial knowledge base, including nutritional supplement ingredients corresponding to critically ill patients at various clinical stages and expected clinical nutritional effects; Step S33: Performing a nutritional requirement analysis on the corresponding nutritional therapy knowledge entity according to the clinical case type and clinical pathophysiological performance indicators corresponding to the critically ill patients, so as to analyze and calculate the clinical nutritional requirements corresponding to each critically ill patient, including the specific dosage requirements corresponding to the protein, fat, carbohydrate and trace element supplementation components; Step S34: determining the timing of nutrient administration for each critically ill patient based on the basal metabolic efficiency and nutrient absorption efficiency of the critically ill patient, so as to obtain the timing of nutrient administration for each critically ill patient; Step S4: Based on the clinical nutritional needs and nutritional administration timing of each critically ill patient, the nutritional administration conflict assessment is performed on the corresponding dietary intake information during the period to obtain the clinical nutritional administration-dietary effect conflict ratio corresponding to each critically ill patient; based on the clinical nutritional administration-dietary effect conflict ratio corresponding to each critically ill patient, the nutritional therapy knowledge is updated and optimized on the personalized nutritional therapy initial knowledge base to generate a personalized nutritional therapy knowledge optimization base corresponding to each critically ill patient, which includes the personalized nutritional knowledge practice guidelines, nutritional administration schemes, nutritional intake compatibility contraindications and dietary adjustment suggestions corresponding to the critically ill patients; wherein, step S4 includes the following steps: Step S41: obtaining corresponding dietary intake food ingredients and dietary intake recording time through the dietary intake information of each critically ill patient during the corresponding period; Step S42: determining the corresponding dietary intake action time segment according to the dietary intake recording time corresponding to each critically ill patient, and performing a nutritional administration interaction time segment analysis on the corresponding dietary intake action time segment based on the nutritional administration timing corresponding to each critically ill patient, so as to obtain the nutritional administration-dietary intake interaction time segment corresponding to each critically ill patient; Step S43: Based on the clinical nutritional needs of each critically ill patient and the nutritional administration-dietary intake interaction period, the nutritional administration conflict evaluation calculation formula is used to evaluate the nutritional administration conflict of the dietary intake food components corresponding to each critically ill patient, so as to obtain the clinical nutritional administration-dietary effect conflict ratio corresponding to each critically ill patient; wherein the nutritional administration conflict evaluation calculation formula is specifically: ； In the formula, For critically ill patients The corresponding clinical nutrition administration-diet effect conflict ratio, is the total amount of clinical nutrients, The period of nutrient administration-dietary intake interaction, For the period When critically ill patients take The nutritional requirements corresponding to the clinical nutritional ingredients are For the period Severely ill patients need The dosage corresponding to the clinical nutritional ingredients, For the period When critically ill patients The drug-diet interaction coefficients corresponding to the clinical nutritional components, For the The relative contribution coefficients of the component conflicts corresponding to the clinical nutritional components, It is the correction factor of the conflict ratio of clinical nutrition administration and diet effect; Step S44: Based on the clinical nutrition administration-diet effect conflict ratio corresponding to each critically ill patient, the personalized nutrition therapy initial knowledge base is updated and optimized for nutrition therapy knowledge, and a personalized nutrition therapy knowledge optimization base corresponding to each critically ill patient is generated, which includes personalized nutrition knowledge practice guidelines, nutrition administration plans, nutrition intake contraindications and dietary adjustment suggestions corresponding to critically ill patients.

2. The method for constructing a personalized nutritional therapy knowledge base for critically ill patients according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: obtaining basic information corresponding to the critically ill patient through the critically ill patient information database corresponding to the target clinical hospital, including the age, gender, height, weight, condition description and clinical manifestations of the critically ill patient; Step S12: Obtaining medical record information corresponding to the critically ill patient through the electronic medical record of the critically ill patient corresponding to the target clinical hospital, including the basic medical record, cause of the critical illness, complications and medical history corresponding to the critically ill patient; Step S13: Obtain laboratory test data corresponding to the critically ill patient through the automated laboratory analysis system corresponding to the target clinical hospital, including the heart rate, blood sugar, blood indexes and metabolic status corresponding to the critically ill patient; Step S14: obtaining the dietary intake information of the critically ill patients during treatment through the dietary record document of the target clinical hospital during treatment, including the dietary habits and dietary intake records of the critically ill patients during treatment; Step S15: Standardize the basic information, medical history information, laboratory test data and dietary intake information corresponding to the critically ill patients and extract the corresponding case entity feature description vector to generate each critically ill patient case entity descriptor.

3. The method for constructing a personalized nutritional therapy knowledge base for critically ill patients according to claim 1, characterized in that: Step S26 includes the following steps: The semantic knowledge association ratio is compared and judged according to a preset semantic knowledge association threshold. If the semantic knowledge association ratio is less than the preset semantic knowledge association threshold, it is determined that there is no knowledge entity representation association relationship between the critically ill patient case entity descriptor and the semantic knowledge entity corresponding to the current medical document with the same research purpose, and the semantic knowledge association ratio corresponding to the next medical document is compared and judged; If the semantic knowledge association ratio is greater than or equal to the preset semantic knowledge association threshold, it is determined that there is a knowledge entity representation association relationship between the critically ill patient case entity descriptor and the semantic knowledge entity corresponding to the current medical literature with the same research purpose, and the corresponding semantic knowledge association ratio is used as the edge weight of the association relationship, so as to construct the initial knowledge base of personalized nutritional treatment corresponding to each critically ill patient based on the knowledge entity representation association relationship and the corresponding edge weight.

4. A system for constructing a personalized nutritional therapy knowledge base for critically ill patients, characterized in that: Used to execute the method for constructing a personalized nutritional therapy knowledge base for critically ill patients as claimed in claim 1, the personalized nutritional therapy knowledge base construction system for critically ill patients comprises: The critically ill patient case entity description module is used to obtain the basic information, medical record information, laboratory test data and dietary intake information corresponding to the critically ill patients, and construct the critically ill case entity according to the basic information, medical record information, laboratory test data and dietary intake information corresponding to the critically ill patients, thereby generating each critically ill patient case entity descriptor; The critical illness entity nutrition knowledge base construction module is used to construct the personalized nutrition treatment initial knowledge base corresponding to each critical illness patient based on the entity descriptors of each critical illness patient case and the semantic knowledge association ratio between relevant clinical medical research literature; The module for determining nutritional needs and medication timing is used to obtain the clinical pathological characteristics and nutritional therapy knowledge entities corresponding to critically ill patients through the personalized nutritional therapy initial knowledge base, and to analyze nutritional needs and determine medication timing based on the clinical pathological characteristics and nutritional therapy knowledge entities, so as to obtain the clinical nutritional needs and nutritional medication timing corresponding to each critically ill patient; The personalized nutritional therapy knowledge base optimization module is used to evaluate the nutritional administration conflict of the corresponding dietary intake information based on the clinical nutritional needs and nutritional administration timing of each critically ill patient, so as to obtain the clinical nutritional administration-dietary effect conflict ratio corresponding to each critically ill patient; based on the clinical nutritional administration-dietary effect conflict ratio corresponding to each critically ill patient, the personalized nutritional therapy initial knowledge base is updated and optimized for nutritional therapy knowledge, thereby generating a personalized nutritional therapy knowledge optimization base corresponding to each critically ill patient, which includes personalized nutritional knowledge practice guidelines, nutritional administration plans, nutritional intake contraindications and dietary adjustment recommendations corresponding to critically ill patients.

Images