Individualized method for calculating real-time energy and nutrient matrix requirements during the course of a patient's disease
By extracting data from the nutritional therapy case database and using multiple machine learning models for prediction, the problem that traditional nutrition support systems cannot perform individual calculations is solved, and more accurate and timely nutrition support is achieved, improving the treatment effect and recovery process of patients.
Patent Information
- Application Number
- CN202411724275.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-28
- Publication Date
- 2025-04-04
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
When formulating and adjusting patient nutrition support plans, the traditional nutrition support service system lacks data support and efficiency, resulting in modular, batch, homogeneous, and homogeneous planning, inability to perform individual calculations, and lacks accurate, timely and effective nutrition support.
By extracting data from the nutritional therapy case database, data cleaning and feature extraction, training using a variety of machine learning models (such as random forests, neural networks, decision trees and support vector machines), combining the patient's individual characteristics and course characteristics, predicting the patient's real-time daily energy and nutrition matrix needs.
It has achieved personalized prediction and decision-making support based on the patient's individual characteristics and disease course characteristics, improved the treatment effect and recovery process, and strengthened the accuracy and timeliness of nutritional support.
Smart Images
Figure CN119763778A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of medical care, and in particular to a method for calculating the instant energy and nutritional matrix requirements of a patient's individualized disease course. Background Art
[0002] The field of healthcare technology focuses on the research and development of technologies and methods for assessing and managing patients’ energy and nutritional needs. In this field, emphasis is placed on integrating patients’ medical information, including health status, physiological needs, underlying medical history, and the first diagnosis of the disease at the time of the visit, through technical means in order to provide a corresponding nutritional support plan.
[0003] To achieve this goal, the corresponding nutrition support service system usually integrates the patient's medical history, medical illness, physiological indicators and nutritional needs information to develop a nutrition plan that meets their target health status, thereby improving the patient's treatment effect, accelerating the recovery process and improving the quality of life.
[0004] However, when formulating and adjusting nutrition support plans for patients, traditional nutrition support service systems, based on technical means and efficiency and lack of data restrictions, often only focus on a single feature and a single disease to perform a single calculation of the nutritional needs of the same type of patients. Individualized calculations cannot be performed for different physical conditions of a single individual, different disease sets of a single individual, different first-diagnosed diseases of the same disease set, and different stages of the same first-diagnosed disease. There are defects such as modularization, batching, homogenization, and lack of precision, timeliness, and effectiveness in plan formulation. Summary of the invention
[0005] In view of this, the object of the present invention is to provide a method for calculating the instant energy and nutritional matrix requirements of a patient's individualized disease course, so as to improve the above-mentioned problems.
[0006] The embodiment of the present invention provides a method for calculating the instantaneous energy and nutrient matrix requirements of a patient's individualized disease course, which comprises:
[0007] Extract the data needed for model training from the existing nutritional therapy case database;
[0008] Performing data cleaning and feature extraction on the data to obtain feature data;
[0009] After performing feature selection, feature expansion and aggregation on the extracted feature data, a data set is obtained;
[0010] Using the data set to train multiple machine learning models to obtain multiple trained machine learning models;
[0011] Obtain input data related to the patient, input the input data into a variety of machine learning models, and obtain corresponding multiple outputs; the input data includes historical data and real-time data, the historical data includes the individual constitution and basic disease data of the patient; the real-time data includes the first diagnosed disease and disease course characteristics when seeking medical treatment;
[0012] A weighted sum is performed according to the multiple outputs to obtain a final output, and the real-time daily energy and nutrient matrix requirements of the patient are predicted according to the final output.
[0013] Preferably, data cleaning includes:
[0014] Missing value processing: use mean filling or interpolation to fill in missing values of numerical data; delete data records with completely missing key fields;
[0015] Outlier detection: Use the z-score method to detect and filter out outliers:
[0016] Data type conversion: convert the date field to datetime format to ensure the consistency of time series data; convert categorical features into numerical codes; the categorical features include gender and diagnosis results.
[0017] Preferably, the extracted feature data includes:
[0018] Numerical features, categorical features, time features, and interactive features.
[0019] Preferably, Lasso regression is used for feature selection; Lasso regression adds L to the loss function. 1 The regularization term shrinks the unimportant feature coefficients to 0, thereby screening out important features:
[0020]
[0021] Among them, y i is the target value, X i is the feature vector of the i-th sample, β j is the coefficient of the jth feature, and λ is the regularization parameter;
[0022] Feature extension and aggregation include:
[0023] Time aggregation: Aggregate the patient's laboratory data by day or week to capture the trend changes of characteristics;
[0024] Sliding window: The sliding window method is used to generate time series features to reflect the dynamic changes at different stages of the disease course.
[0025] Preferably, the multiple machine learning models include at least two of random forest, neural network, decision tree and support vector machine.
[0026] Preferably, when training a machine learning model:
[0027] The dataset is split into training set and test set in a 4:1 ratio;
[0028] Use 5-fold cross validation to improve the generalization ability of the model;
[0029] Use the training set to train the model and update the parameters;
[0030] Use the validation set to evaluate model performance and adjust hyperparameters;
[0031] A final evaluation is performed on the test set to ensure the generalization performance of the model.
[0032] Preferably, in the input data:
[0033] Individual characteristics include the patient's BMI and PAL index, which reflect the patient's basic physical condition and activity level;
[0034] The basic disease data are the a priori nutritional intervention treatment data of the basic disease group experts, which are the data accumulated from the a priori consensus of experts on different individual diseases;
[0035] The first disease diagnosed when seeking medical treatment is the feature value data with the largest weight;
[0036] Disease Status includes fever, pain, redness and swelling, diarrhea, constipation, laboratory indicators, treatment plans and medications used. Treatment plans and medications used include chemotherapy, radiotherapy, surgical treatment measures and drug use.
[0037] Preferably, it also includes:
[0038] The predicted real-time daily energy and nutrient matrix requirements are corrected with the preset prior values to ensure the rationality and accuracy of the prediction; the correction formula is as follows:
[0039] Predicted Value=f(BMI,PAL,Disease Status).
[0040] Preferably, it also includes:
[0041] Monitor the patient's treatment effect and symptom changes, and update the feedback data to the database in real time to ensure that the predicted nutritional needs are consistent with the patient's actual condition; the feedback data includes daily symptom changes, laboratory test results, and the effect of treatment methods, which are used to guide the dynamic adjustment of the system; the nutritional demand adjustment formula after daily feedback is as follows:
[0042] Adjusted V alue=Predicted V alue+k(Feedback Error)
[0043] Among them, k is the feedback coefficient, which is used to control the amplitude of adjustment based on the feedback error; all feedback data will flow back to the database and serve as the basis for model optimization:
[0044]
[0045] Preferably, the database comprises:
[0046] Individual constitution + energy and nutrient matrix requirements best practice data decision base for the first day of disease diagnosis; and
[0047] Best practice data decision-making database for daily energy and nutrient matrix requirements based on individual constitution + disease course characteristics.
[0048] In summary, the embodiments of the present invention have at least the following improvements compared with the prior art:
[0049] 1. Provide personalized prediction and decision support based on the individual characteristics of the patient and the characteristics of the disease course, obtain immediate energy and nutritional matrix corresponding to the patient's disease and disease course, improve the patient's treatment effect, and accelerate the recovery process.
[0050] 2. Use multiple machine learning models for prediction and decision making, and the prediction results are more reliable.
[0051] 3. The database content is continuously updated according to clinical practice to optimize future prediction accuracy and decision-making effects, and provide patients with personalized and dynamic nutritional support. BRIEF DESCRIPTION OF THE DRAWINGS
[0052] In order to more clearly illustrate the technical solution of the present invention, the drawings required for use in the implementation mode will be briefly introduced below. Obviously, the drawings described below are only some implementation modes of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0053] Figure 1 It is a flow chart of a method for calculating the instantaneous energy and nutrient matrix requirements of a patient's individualized disease course provided by the first embodiment of the present invention.
[0054] Figure 2 This is a schematic diagram of the principle of random forest.
[0055] Figure 3 This is a schematic diagram of the principle of a neural network.
[0056] Figure 4This is a schematic diagram of the decision tree principle.
[0057] Figure 5 It is a schematic diagram of the principle of support vector machine. DETAILED DESCRIPTION
[0058] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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 ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0059] In order to better understand the technical solution of the present invention, the embodiments of the present invention are described in detail below with reference to the accompanying drawings.
[0060] It should be clear that the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0061] See also Figure 1 The first embodiment of the present invention is a method for calculating the instantaneous energy and nutrient matrix requirements of a patient's individualized disease course, which comprises:
[0062] S101, extracting data required for model training from an existing nutritional therapy case database.
[0063] In this embodiment, the nutritional therapy case library C has the following characteristics:
[0064] It stores historical patient diagnosis, treatment and nutritional intervention data, providing a reference for prediction and decision-making on similar cases.
[0065] Contains the patient's initial diagnosis, individual physical data, treatment plan and implementation records of daily energy requirements.
[0066] The data in the case library is continuously updated with the treatment effects, and the model is optimized through feedback to form a dynamic adjustment mechanism to ensure that the nutritional support plans for future patients are more accurate and effective.
[0067] In this embodiment, the information required for modeling is extracted from the existing nutrition therapy case database C. The data to be extracted include:
[0068] Physical information: patient's underlying disease group, personal activity level (PAL), and first diagnosed disease.
[0069] Information on the characteristics of the course of disease: including symptoms during the onset of the disease (such as fever, pain, redness and swelling, chills, geographic tongue, loss of appetite, diarrhea, constipation, abnormal body sensation, pleural effusion and ascites, etc.).
[0070] Laboratory and imaging test results: such as C-reactive protein (CRP), tumor necrosis factor (TNF), plasma albumin, retinol protein, liver function, kidney function, imaging results, iron, hemoglobin, blood sugar and other biochemical indicators.
[0071] Treatment methods and drug use: such as chemotherapy, radiotherapy, peritoneal dialysis, hemofiltration, surgery, colonoscopy and other treatment methods and hormones, thyroid-stimulating hormone, insulin and other drugs.
[0072] Data formats include CSV, Excel, SQL, etc., which can be read, integrated and processed using pandas and SQLAlchemy libraries.
[0073] S102, performing data cleaning and feature extraction on the data to obtain feature data.
[0074] In this embodiment, data cleaning includes:
[0075] Missing value processing: use mean filling or interpolation to fill in missing values of numerical data; delete data records with completely missing key fields;
[0076] Outlier detection: Use the z-score method to detect and filter out outliers:
[0077]
[0078] Where x is the data point, μ is the mean, and σ is the standard deviation. If |z|>3, it is marked as an outlier.
[0079] Data type conversion: convert date fields to datetime format to ensure consistency of time series data; convert categorical features (such as gender and diagnosis results) to numeric codes;
[0080] The extracted feature data include:
[0081] Numerical features: such as laboratory test results (CRP, TNF, blood sugar, etc.), are directly input into the model for analysis.
[0082] Categorical features: such as gender, first diagnosed disease, and medication use, are processed using One-Hot Encoding to facilitate model parsing.
[0083] Temporal features: Convert dates into features such as months and days of the week to capture potential temporal patterns.
[0084] Interaction features: For example, the combination of a specific treatment and symptoms (such as chemotherapy and loss of appetite) is used as a new interaction feature to reflect the relationship between treatment and symptoms.
[0085] Furthermore, the extracted features can be standardized and normalized, where:
[0086] Standardization refers to standardizing numerical features such as laboratory test results so that their mean is 0 and their standard deviation is 1. Standardization helps to improve the convergence speed during model training.
[0087] Normalization (Min-Max Normalization) refers to normalizing the severity of symptoms or laboratory data values to the interval [0,1]. Normalization is applicable to data with a known and fixed range.
[0088] S103, performing feature selection, feature expansion and aggregation on the extracted feature data to obtain a data set.
[0089] In this embodiment, Lasso regression is used for feature selection. Lasso regression adds L in the loss function. 1 The regularization term shrinks the unimportant feature coefficients to 0, thereby screening out important features:
[0090]
[0091] Among them, y i is the target value, X i is the feature vector of the i-th sample, β j is the coefficient of the jth feature, and λ is the regularization parameter. Of course, other feature selection methods, such as the principal component method, etc., may also be used, and the present invention does not make specific limitations.
[0092] In this embodiment, feature expansion and aggregation include:
[0093] Time aggregation: Aggregate the patient's laboratory data by day or week to capture the trend changes of characteristics;
[0094] Sliding window: The sliding window method is used to generate time series features to reflect the dynamic changes at different stages of the disease course.
[0095] S104, using the data set to train multiple machine learning models to obtain multiple trained machine learning models.
[0096] In this embodiment, the multiple machine learning models include at least two of random forest, neural network, decision tree and support vector machine.
[0097] like Figure 2 As shown in Figure 1, random forest is an ensemble learning-based algorithm consisting of multiple decision trees, each of which is trained independently on a different subset of data. The final prediction result is the average of all tree prediction results (regression task) or the majority voting result (classification task).
[0098] The formula is as follows:
[0099]
[0100] Where T is the number of decision trees, f t (x) is the predicted value of the tth tree.
[0101] Its working steps are:
[0102] 1. Generate multiple data subsets by random sampling from the training set.
[0103] 2. Train a decision tree for each subset.
[0104] 3. Average or vote the predictions of all trees as the final output.
[0105] like Figure 3 As shown in Figure 1, the neural network extracts features from input data and makes predictions through multiple layers of nonlinear transformations. Each layer of the network includes multiple neurons, which are calculated through activation functions (such as ReLU).
[0106] The loss function is the mean square error (MSE):
[0107]
[0108] Its working steps are:
[0109] 1. The input layer passes in the patient’s BMI, PAL and other data.
[0110] 2. Multiple hidden layers perform nonlinear transformations and output prediction results.
[0111] 3. Use the back-propagation algorithm to update the weights and minimize the loss function.
[0112] like Figure 4 As shown in the figure, the decision tree recursively divides the feature space to generate a tree structure, where each node represents a division point of the feature. The final leaf node represents the predicted value.
[0113] Information gain or Gini coefficient is used for calculation when dividing:
[0114] IG(T,X)=H(T)-H(T|X)
[0115] Among them, H(T) is the entropy of the current node, and H(T|X) is the entropy after partitioning.
[0116] Its working steps are:
[0117] 1. Select features and determine the best partitioning points.
[0118] 2. Recursively build the tree until the stopping condition is met.
[0119] 3. Use the generated tree to make predictions.
[0120] like Figure 5 As shown, SVM classifies or regresses data points by finding the hyperplane that maximizes the margin. For nonlinear problems, SVM uses a kernel function (such as the RBF kernel) to map to a high-dimensional space.
[0121] The optimization goals are as follows:
[0122]
[0123] Among them, C is the regularization parameter, y i is the label, x i As a feature.
[0124] Its working steps are:
[0125] 1. Choose a suitable kernel function to map the feature space.
[0126] 2. Train the SVM model and find the optimal hyperplane.
[0127] 3. Use the trained model for classification or regression.
[0128] In this embodiment, when training the above machine learning model:
[0129] The dataset is split into training set and test set in a 4:1 ratio;
[0130] Use 5-fold cross validation to improve the generalization ability of the model;
[0131] Use the training set to train the model and update the parameters;
[0132] Use the validation set to evaluate model performance and adjust hyperparameters;
[0133] A final evaluation is performed on the test set to ensure the generalization performance of the model.
[0134] S105, obtaining input data related to the patient, inputting the input data into a plurality of machine learning models, and obtaining corresponding plurality of outputs.
[0135] In this embodiment, the input data includes historical data and real-time data. The historical data includes the individual constitution and underlying disease data of the patient; the real-time data includes the first diagnosed disease and disease course characteristics of the current medical treatment.
[0136] Among them, individual characteristics include the patient's BMI and PAL index, which reflect the patient's basic physical condition and activity level.
[0137] In this embodiment, for normal people, different BMI and PAL indexes correspond to different daily energy requirements and nutritional matrix, which can be obtained based on the daily energy requirements and nutritional matrix database A of the living distribution of the Chinese population. The daily energy requirements and nutritional matrix database A of the living distribution of the Chinese population includes:
[0138] The BMI of a specific person's energy and nutritional reserve status will adjust the weight of the actual weight in different ranges. For example, if it is 20-24, the weight weight is 1*weight; if it is >24, the weight is calculated as (height-105).
[0139] Different daily activities have different PAL indexes, such as lying in bed 1.2 and running 2.0.
[0140] The recommended daily basic energy requirement BEE for Chinese people, such as adult males (20Kcal / day / kg), will be different for other groups such as women, children, pregnant women, postpartum women, and the elderly due to different characteristics.
[0141] In this embodiment, the basic disease data are the prior nutritional intervention treatment data of the basic disease group experts, which are the data accumulated from the prior consensus of experts on different individual diseases.
[0142] Different diseases have different requirements for different energies and nutritional matrices. The same person may have several basic diseases at the same time, which may lead to contradictions. The decision-making method is that if a disease requires a certain value < less than a certain limit, the value is in the value range of this item. If they are all > greater than or = equal to, the maximum value is used as the value range. These can be obtained based on the prior nutrition intervention treatment database B of the basic disease group experts.
[0143] In this embodiment, the first disease diagnosed at the time of medical treatment is the eigenvalue data with the largest weight in the input data. This is because the first disease diagnosed at the time of medical treatment is the main disease of the patient's current medical treatment and is also the main contradiction in treatment. Therefore, it needs to be given a higher weight. At the same time, the first disease diagnosed at the time of medical treatment can be combined with the previous two factors to form a database for the training model.
[0144] Disease Status includes fever, pain, redness and swelling, diarrhea, constipation, laboratory indicators, treatment plan and medications used.
[0145] In this embodiment, the characteristics of the course of disease include symptoms of the course of disease, treatment plans and drugs used, symptoms of the course of disease include fever, pain, redness and swelling, chills, geographic tongue, appetite, diarrhea, constipation, body sensation, pleural effusion and ascites, pathological laboratory imaging examinations reflecting the course of disease after hospitalization such as C-reactive protein, tumor necrosis factor, plasma albumin, retinol protein, liver function, kidney function and others such as, imaging results, iron, hemoglobin, blood sugar; treatment plans and drugs used include chemotherapy, radiotherapy, surgical treatment measures and drug use such as chemotherapy, radiotherapy, peritoneal dialysis, hemofiltration and hemodialysis, surgery, colonoscopy, and specific drug treatments such as hormones, thyroid-stimulating hormone, and insulin.
[0146] At each stage of the disease process, the system calculates the patient's current needs according to the formula:
[0147] y(t) = f(disease course characteristics(t), nutritional intervention(t))
[0148] The formula combines symptoms, laboratory data and treatment measures to dynamically adjust daily nutritional requirements, thereby better responding to the patient's real-time needs.
[0149] In this embodiment, when constructing a training database based on the characteristics of the course of disease, the course of disease is used as the main line for construction. This is mainly because different courses of disease may have obvious differences in the types and amounts of nutrition required. Different treatment plans also have different metabolic changes and nutritional requirements for the body. The use of some specific drugs will also affect nutritional metabolism and requirements. These are all strongly associated feature values for nutritional demand analysis in the course of disease.
[0150] In summary, for input data, the overall structure of this embodiment is individual characteristics + different basic chronic diseases + the first diagnosed disease at the time of medical treatment + different stages of the disease. These data also include historical data related to the individual situation of the patient and real-time data related to the patient's current real-time status. Therefore, the output judgment and value based on these input data can ensure that the output data can meet the patient's personal physical condition and current real-time physical condition, and better meet the patient's actual needs, thereby improving the patient's treatment effect and accelerating the recovery process.
[0151] S106, performing weighted summation according to the multiple outputs to obtain a final output, and predicting the patient's real-time daily energy and nutrient matrix requirements according to the final output.
[0152] In this embodiment, multiple machine learning models are used to estimate daily energy and nutritional requirements in combination with the individual characteristics of the patient. After each model is trained independently, a weighted integration method is used in the prediction stage to integrate the output results of all models. According to the conflict between multiple disease requirements, the system will dynamically adjust the final output energy requirements according to the decision rules.
[0153] Specifically, if there are conflicts between the requirements of multiple diseases, the following rules are adopted:
[0154] Restrictive demand priority: If a disease demand has a restriction, the restriction value is selected as the final demand value.
[0155] Final Value = min (limit value, other predicted values)
[0156] Maximum value first: If there is no explicit restriction, the maximum value among all requirements is selected:
[0157] Final Value = max(demand 1, demand 2,…, demand n).
[0158] In this embodiment, based on the individual characteristics, stage of disease and course of disease of the patient, the system outputs real-time daily energy and nutrient matrix requirement values. These outputs include:
[0159] Energy Requirement: Dynamically adjust the daily calorie requirement based on the patient's disease course.
[0160] Nutrient Requirements: The proportion of nutrients such as protein, fat, carbohydrates, etc. is allocated according to the treatment plan and the characteristics of the disease course. The formula is as follows:
[0161] Adjusted Energy=Predicted Energy+k×(Feedback Error)
[0162] Among them, k is the feedback coefficient, which is dynamically adjusted according to the patient's daily feedback.
[0163] After obtaining the ratio of nutrients such as protein, fat, carbohydrates, etc., the appropriate food combination can be matched according to the nutritional composition of the food in the food database to meet the nutritional needs of the patient.
[0164] Preferably, it also includes:
[0165] The predicted real-time daily energy and nutrient matrix requirements are corrected with the preset prior values to ensure the rationality and accuracy of the prediction; the correction formula is as follows:
[0166] Predicted Value=f(BMI,PAL,Disease Status).
[0167] Preferably, it also includes:
[0168] Monitor the patient's treatment effect and symptom changes, and update the feedback data to the database in real time to ensure that the predicted nutritional needs are consistent with the patient's actual condition; the feedback data includes daily symptom changes, laboratory test results, and the effect of treatment methods, which are used to guide the dynamic adjustment of the system; the nutritional demand adjustment formula after daily feedback is as follows:
[0169] Adjusted V alue=Predicted V alue+k(Feedback Error)
[0170] Among them, k is the feedback coefficient, which is used to control the amplitude of adjustment based on the feedback error; all feedback data will flow back to the database and serve as the basis for model optimization:
[0171]
[0172] This mechanism ensures that the database content is continuously updated in line with clinical practice, optimizes future prediction accuracy and decision-making effects, and provides patients with personalized and dynamic nutritional support.
[0173] The databases here include: individual constitution + energy and nutrient matrix requirements on the first day of disease diagnosis best practice data decision library; and individual constitution + energy and nutrient matrix requirements on the characteristic day of disease course best practice data decision library.
[0174] In summary, the embodiments of the present invention have at least the following improvements compared with the prior art:
[0175] 1. Provide personalized prediction and decision support based on the individual characteristics of the patient and the characteristics of the disease course, obtain immediate energy and nutritional matrix corresponding to the patient's disease and disease course, improve the patient's treatment effect, and accelerate the recovery process.
[0176] 2. Use multiple machine learning models for prediction and decision making, and the prediction results are more reliable.
[0177] 3. The database content is continuously updated according to clinical practice to optimize future prediction accuracy and decision-making effects, and provide patients with personalized and dynamic nutritional support.
[0178] The second embodiment of the present invention also provides a device for calculating the immediate energy and nutritional matrix requirements of a patient's individualized disease course, which includes a memory and a processor, wherein a computer program is stored in the memory, and the computer program can be executed by the processor to implement the method for calculating the immediate energy and nutritional matrix requirements of a patient's individualized disease course as described above.
[0179] The third embodiment of the present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program can be executed by a processor of a device where the computer-readable storage medium is located to implement the method for calculating the instant energy and nutritional matrix requirements of a patient's individualized disease course as described above.
[0180] Exemplarily, the above-mentioned devices and process steps can be implemented by a computer program. The computer program can be divided into one or more units. The one or more units are stored in the memory and executed by the processor to complete the present invention.
[0181] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.
[0182] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the present invention by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (FlashCard), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.
[0183] Wherein, if the electronic device or printer integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.
[0184] It should be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the accompanying drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art may understand and implement it without paying any creative effort.
[0185] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.
Claims
1. A method for calculating the instantaneous energy and nutrient matrix requirements of a patient's individualized disease course, characterized in that: include: Extract the data needed for model training from the existing nutritional therapy case database; Performing data cleaning and feature extraction on the data to obtain feature data; After performing feature selection, feature expansion and aggregation on the extracted feature data, a data set is obtained; Using the data set to train multiple machine learning models to obtain multiple trained machine learning models; Obtain input data related to the patient, input the input data into a variety of machine learning models, and obtain corresponding multiple outputs; the input data includes historical data and real-time data, the historical data includes the individual constitution and basic disease data of the patient; the real-time data includes the first diagnosed disease and disease course characteristics when seeking medical treatment; A weighted sum is performed according to the multiple outputs to obtain a final output, and the real-time daily energy and nutrient matrix requirements of the patient are predicted according to the final output.
2. The method for calculating the instantaneous energy and nutrient matrix requirements of a patient's individualized disease course according to claim 1, characterized in that: Data cleaning includes: Missing value processing: use mean filling or interpolation to fill in missing values of numerical data; delete data records with completely missing key fields; Outlier detection: Use the z-score method to detect and filter out outliers: Data type conversion: convert the date field to datetime format to ensure the consistency of time series data; convert categorical features into numerical codes; the categorical features include gender and diagnosis results.
3. The method for calculating the instantaneous energy and nutrient matrix requirements of a patient's individualized disease course according to claim 1, characterized in that: The extracted feature data include: Numerical features, categorical features, time features, and interactive features.
4. The method for calculating the instantaneous energy and nutritional matrix requirements of a patient's individualized disease course according to claim 1, characterized in that: When selecting features, Lasso regression is used for feature selection; Lasso regression shrinks unimportant feature coefficients to 0 by adding an L1 regularization term to the loss function, thereby screening out important features: Among them, y i is the target value, X i is the feature vector of the i-th sample, β j is the coefficient of the jth feature, and λ is the regularization parameter; Feature extension and aggregation include: Time aggregation: Aggregate the patient's laboratory data by day or week to capture the trend changes of characteristics; Sliding window: The sliding window method is used to generate time series features to reflect the dynamic changes at different stages of the disease course.
5. The method for calculating the instantaneous energy and nutrient matrix requirements of a patient's individualized disease course according to claim 1, characterized in that: The multiple machine learning models include at least two of random forest, neural network, decision tree and support vector machine.
6. The method for calculating the instantaneous energy and nutrient matrix requirements of a patient's individualized disease course according to claim 1, characterized in that: When training a machine learning model: The dataset is split into training set and test set in a 4:1 ratio; Use 5-fold cross validation to improve the generalization ability of the model; Use the training set to train the model and update the parameters; Use the validation set to evaluate model performance and adjust hyperparameters; A final evaluation is performed on the test set to ensure the generalization performance of the model.
7. The method for calculating the instantaneous energy and nutrient matrix requirements of a patient's individualized disease course according to claim 1, characterized in that: In the input data: Individual characteristics include the patient's BMI and PAL index, which reflect the patient's basic physical condition and activity level; The basic disease data are the a priori nutritional intervention treatment data of the basic disease group experts, which are the data accumulated from the a priori consensus of experts on different individual diseases; The first disease diagnosed when seeking medical treatment is the feature value data with the largest weight; Disease Status includes fever, pain, redness and swelling, diarrhea, constipation, laboratory indicators, treatment plans and medications used. Treatment plans and medications used include chemotherapy, radiotherapy, surgical treatment measures and drug use.
8. The method for calculating the instantaneous energy and nutritional matrix requirements of a patient's individualized disease course according to claim 7, characterized in that: Also includes: The predicted real-time daily energy and nutrient matrix requirements are corrected with the preset prior values to ensure the rationality and accuracy of the prediction; the correction formula is as follows: Predicted Value=f(BMI,PAL,Disease Status).
9. The method for calculating the instantaneous energy and nutritional matrix requirements of a patient's individualized disease course according to claim 8, characterized in that: Also includes: Monitor the patient's treatment effect and symptom changes, and update the feedback data to the database in real time to ensure that the predicted nutritional needs are consistent with the patient's actual condition; the feedback data includes daily symptom changes, laboratory test results, and the effect of treatment methods, which are used to guide the dynamic adjustment of the system; the nutritional demand adjustment formula after daily feedback is as follows: Adjusted V alue=Predicted V alue+k(Feedback Error) Among them, k is the feedback coefficient, which is used to control the amplitude of adjustment based on the feedback error; all feedback data will flow back to the database and serve as the basis for model optimization:
10. The method for calculating the instantaneous energy and nutritional matrix requirements of a patient's individualized disease course according to claim 9, characterized in that: The database includes: Individual constitution + energy and nutrient matrix requirements best practice data decision base for the first day of disease diagnosis; and Best practice data decision-making database for daily energy and nutrient matrix requirements based on individual constitution + disease course characteristics.