Method and system for constructing nutrition state outcome prediction model based on dynamic body weight

Abstract

The invention relates to a body weight dynamic-based nutrition state outcome prediction model construction method and system, and belongs to the field of medical informatization. The method comprises the following steps: firstly, establishing a crowd data set containing individual weight information and nutrition state data, and carrying out data preprocessing, including calculating a weight loss percentage and defining an end label; then, the data set is divided into a training set and a test set, and key features related to nutrition state outcome are extracted from the training set through feature screening; and based on the screened features and the training set, constructing a plurality of machine learning classifiers, and selecting an optimal model according to performance indexes of the classifiers. And finally, predicting the test set by using the optimal model, and evaluating the performance of the model. According to the method, the weight dynamic information is utilized, other information assistance is not needed, high-performance prediction of common bad nutrition state outcome can be achieved, and the method has high practical value.

Description

Technical Field

[0001] The present invention belongs to the field of medical information technology and relates to a method and system for constructing a nutritional status outcome prediction model based on body weight dynamics. Background Art

[0002] Nutrition-related problems include a large category of disease states, including malnutrition, cachexia, and sarcopenia, and have become a major global public health challenge. Nutritional problems also commonly plague the diagnosis and treatment of clinical diseases. Poor nutritional status can affect the effectiveness and safety of disease treatment, increase the use of medical resources, and lead to a variety of adverse consequences. Taking malignant tumors as an example, about 10%-20% of tumor deaths can be attributed to malnutrition. However, due to the lack of appropriate diagnostic technology and sufficient attention, poor nutritional states such as malnutrition and cachexia are often underestimated, misdiagnosed or untreated. Therefore, active diagnosis, monitoring, and intervention of poor nutritional status are crucial for patients, which can minimize or reverse its negative impact on prognosis, improve patients' quality of life and outcomes, and reduce medical costs.

[0003] The diagnostic criteria for determining poor nutritional status, such as malnutrition and cachexia, have long varied between regions and institutions, making it difficult to develop a globally unified framework for patient management. To address this challenge, a range of academic organizations have proposed different frameworks for determining nutritional status. For example, in 2019, a conceptual framework based on international consensus, the Global Leadership Initiative on Malnutrition (GLIM), was proposed for diagnosing malnutrition in adults in clinical settings. Since its publication, the GLIM framework has been shown to be of great value in diagnosing malnutrition and predicting clinical outcomes, and is applicable to a variety of diseases. The GLIM criteria use a phenotypic criterion and an etiological criterion for the diagnosis of malnutrition, and the phenotypic criteria further include weight loss calculated based on historical and current weight, body mass index, and muscle mass. Similarly, in 2011, Fearon et al. led an international Delphi consensus and developed a conceptual framework for the diagnosis of tumor cachexia, which has since been widely recognized. This framework also uses three pieces of information for diagnosing cachexia: recent weight loss, body mass index, and muscle mass.

[0004] However, the existing technical frameworks for determining poor nutritional status, whether the GLIM criteria for malnutrition or the Fearon criteria for cachexia, can only determine the nutritional status of an individual at a single time point and cannot effectively determine the future outcome of nutritional status. Given the differences in individual responses to malnutrition and its treatment, being able to identify patients who may benefit from multidisciplinary treatment and / or may return to normal nutritional status will help make individualized medical decisions. Weight and body mass index are key factors in identifying, characterizing, and grading nutritional status. Skeletal muscle weight accounts for almost half of body weight and can also be predicted by weight, height, sex, and age. Previous studies have shown that these factors also help predict the risk of future weight loss and muscle loss. Therefore, using weight-related information that is routinely available when an individual is admitted to the hospital is expected to independently predict future changes in nutritional status. By predicting changes in nutritional status, it is expected to achieve accurate patient grouping and identify patients with potential to restore nutritional status, thereby achieving more precise management strategies and optimizing patient outcomes. Summary of the invention

[0005] In view of this, the purpose of the present invention is to provide a method and system for constructing a nutritional status outcome prediction model based on weight dynamics, so as to serve as an auxiliary decision-making tool to realize nutritional status outcome assessment, patient risk stratification and efficacy estimation in various scenarios such as low cost, remote and self-assessment.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] The method for constructing a nutritional status outcome prediction model based on body weight dynamics includes:

[0008] Establishing a population data set, wherein the population data set includes a plurality of individuals, each individual including weight information at a plurality of time points and nutritional status at the corresponding time points;

[0009] Preprocessing the sample data, including calculating the weight loss percentage and defining an outcome label, wherein the outcome label is determined according to changes in the nutritional status of the individual at the multiple time points;

[0010] Dividing the population data set into a training set and a test set; dividing the population data set into a training set and a test set;

[0011] Modeling feature screening, screening out features related to the outcome label from the training set;

[0012] Constructing multiple machine learning classifiers based on the screened features and the training set;

[0013] Selecting the best model, selecting the best model according to the performance indicator of the machine learning classifier;

[0014] Performance evaluation and prediction: using the best model to predict the test set and evaluate the performance of the prediction model.

[0015] Furthermore, the weight information at multiple time points includes:

[0016] Current weight;

[0017] Historical weight, including historical weight data at multiple time points.

[0018] Furthermore, the nutritional status is diagnosed according to the GLIM standard, the Fearon standard or the SGA scale.

[0019] Furthermore, the ending tag includes:

[0020] Examples of nutritional status transitions, i.e., changes in the nutritional status of the individual at the multiple time points;

[0021] An example of a continuous nutritional state is one in which the nutritional state of the individual remains constant at the multiple time points.

[0022] Further, the machine learning classifier includes:

[0023] Random Forest;

[0024] Multilayer Perceptron;

[0025] Long short-term memory network;

[0026] Transformer.

[0027] Furthermore, the performance indicators include:

[0028] Accuracy;

[0029] Area under the curve;

[0030] Recall rate;

[0031] Accuracy;

[0032] F1 score;

[0033] Kappa index.

[0034] A nutritional status prognosis prediction system based on body weight dynamics, comprising:

[0035] A data collection module, used to collect weight information and nutritional status information of multiple individuals;

[0036] A data analysis module, used to construct a nutritional status outcome prediction model based on the weight information and nutritional status information;

[0037] The result output module is used to output the prediction result of the prediction model.

[0038] Further, the data acquisition module includes:

[0039] A model training data input submodule, used to collect training data for building the prediction model;

[0040] The individual prediction data input submodule is used to collect individual data for predicting nutritional status outcomes.

[0041] Furthermore, the data analysis module includes:

[0042] A prediction model building submodule, used to build the prediction model according to the training data;

[0043] The prediction submodule is executed to predict the individual data using the prediction model.

[0044] Further, the result output module includes:

[0045] A batch modeling result output submodule, used to output the performance indicators of the prediction model;

[0046] The individual prediction result output submodule is used to output the prediction result of the prediction model.

[0047] The beneficial effects of the present invention are:

[0048] (1) Compared with the commonly used nutritional status evaluation methods, the present invention uses only the dynamic information of body weight as the independent variable in the prediction process, which is simple and easy to understand. It does not need to rely on complex instruments and high-end medical equipment, and does not require the intervention of health professionals. All independent variable information can be obtained only through the baseline evaluation of others or individuals. Therefore, it is expected to be widely used in various hospitals, community health service institutions, family scenarios and research scenarios at all levels, and has the characteristics of simplicity, speed and low cost.

[0049] (2) The present invention does not need to consider what kind of clinical treatment or other information the patient will receive in the future, so it can determine which patients are likely to benefit from nutrition at an early stage;

[0050] (3) The present invention does not specify a specific machine learning classification algorithm, so the specific algorithm can be reasonably selected according to different populations, disease stages, data types, and sample sizes. This improves the flexibility of the present invention in the application process and can meet the needs of saving computing costs under different hardware conditions;

[0051] (4) The present invention can achieve more refined predictions such as aggravation or alleviation of abnormal nutritional status by replacing the multi-classification machine learning algorithm in the modeling step;

[0052] (5) The present invention is expected to be applicable to other systems that use body weight as the primary diagnostic criterion and has a certain degree of derivability.

[0053] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below in conjunction with the accompanying drawings, wherein:

[0055] Figure 1 It is a technical flow chart of the present invention;

[0056] Figure 2 The precision-recall curve and the receiver operating characteristic curve of the test set in the first embodiment of the present invention are shown;

[0057] Figure 3 The model prediction confusion matrix of the test set in the first embodiment of the present invention and the relative importance diagram of the top three variables;

[0058] Figure 4 The precision-recall curve and the receiver operating characteristic curve of the test set in the second embodiment of the present invention are shown;

[0059] Figure 5 It is the model prediction confusion matrix of the test set in the second embodiment of the present invention and the relative importance diagram of the top three variables. DETAILED DESCRIPTION

[0060] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0061] Among them, the drawings are only used for illustrative explanations, and they only represent schematic diagrams rather than actual pictures, and should not be understood as limitations on the present invention. In order to better illustrate the embodiments of the present invention, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of actual products. For those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0062] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if the terms "upper", "lower", "left", "right", "front", "rear", etc. indicate the orientation or position relationship, they are based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation. Therefore, the terms describing the position relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting the present invention. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0063] In a first aspect, the present invention provides a method for constructing a nutritional status outcome prediction model based on body weight dynamics, comprising the following steps:

[0064] 1) Establish a crowd data set:

[0065] A population of inpatients was selected, and variable information of the modeling population was collected through sources such as public databases or medical electronic medical record systems. The modeling information included two time nodes, one before and one after. The variables included at each time point were similar, including but not limited to age, gender, height, body mass index, current weight, historical weight at multiple time points, and limb skeletal muscle index as data set samples;

[0066] 2) Sample data preprocessing:

[0067] The collected sample data sets were preprocessed, including but not limited to deleting outliers, excluding samples with missing key modeling information, interpolating missing data using multiple interpolation, performing unique hot encoding on multi-classified data, standardizing continuous variables, and deleting variables with the same values ​​for all samples; the percentage of weight loss was calculated using the formula: (weight at the next time point - weight at the previous time point) ÷ weight at the previous time point × 100%; the outcome labels were defined, and the nutritional status of the two time points before and after was first diagnosed using widely recognized international standards, including but not limited to the GLIM standard (Global Leaders Malnutrition Initiative), Fearon cachexia diagnostic criteria, and SGA scale (Subjective Global Assessment Scale); instances where the nutritional status at the previous time point was normal or abnormal, but the nutritional status at the next time point changed from normal to abnormal or from abnormal to normal were defined as nutritional status transition instances and coded as 1. Instances where the nutritional status remained unchanged were defined as persistent nutritional status and coded as 0. If there is a severity classification of nutritional status, the discrete state transition possibilities such as continued abnormal state, worsening state, improved state but not completely restored to normal, and completely restored to normal state are coded separately;

[0068] 3) Divide the training set and test set:

[0069] The sample data set preprocessed in step 2) is randomly divided into a training set and a test set according to a certain ratio. The training set occupies a larger proportion and is used for model training, while the training set data occupies a smaller proportion and is used to evaluate the effectiveness of model classification;

[0070] 4) Modeling feature screening

[0071] The training set divided in step 3) is selected, and the nutritional status outcome is used as the outcome variable, and the baseline information that can be obtained from the first face-to-face consultation with the patient after preprocessing in step 1) is used as the independent variable, including but not limited to the weight loss percentage in the past month, the weight loss percentage in the past six months, the weight loss percentage in the past year, the current body mass index, the body mass index one month ago, and the body mass index six months ago; the key features that have an important impact on the prediction performance of the model are retained by ranking the importance of the independent variables or manually incorporating the variables, and irrelevant features that have a small impact on the prediction results are eliminated, so as to finally obtain a more concise and efficient feature subset;

[0072] 5) Build multiple machine learning classifiers

[0073] Based on the training set of the modeling independent variables in step 4), a supervised machine learning algorithm is used to construct a variety of classifiers, and the selection of machine learning algorithms includes but is not limited to random forest, multi-layer perceptron, long short-term memory network, and Transformer; according to the number and value range of hyperparameters of different classifiers, a random search or grid search method is used to optimize the hyperparameters of the classifier, and a cross-validation or validation in a validation set method is used to generate a variety of evaluation graphs and indicators, including but not limited to receiver operating characteristic curves, calibration curves, areas under the curves and 95% confidence intervals, and Brier scores;

[0074] 6) Select the best model

[0075] Based on the model evaluation index obtained in step 5), a main evaluation index is selected according to the purpose, the candidate classifiers are sorted according to the index evaluation index, and the classifier with the best performance is selected as the best model;

[0076] 7) Performance evaluation and prediction

[0077] According to the best model obtained in step 6), the nutritional status of the test set samples or the samples manually input by the user is quickly predicted;

[0078] In step 1), a specific subgroup of the population is selected for stratified modeling to reduce the amount of calculation and obtain a more targeted model;

[0079] In step 2), the standardization method of continuous variables is the z-score method, and the specific formula is: (x-μ) / σ. Wherein μ represents the mean of the overall data, σ represents the standard deviation of the overall data, and x represents the individual observation value;

[0080] In step 3), a cross-validation method is used to improve sample utilization rate;

[0081] In step 4), features widely used in clinical practice are manually incorporated into the modeling.

[0082] In a second aspect, the present invention provides a method for constructing a nutritional status outcome prediction model based on body weight dynamics, see Figure 1 The system mainly includes the following modules:

[0083] Data acquisition module:

[0084] 1) Model training data input submodule: used to collect training data required for modeling from electronic medical data systems, local data sources or other data sources, including but not limited to age, gender, height, body mass index, current weight, historical weight at multiple time points, and limb skeletal muscle index at multiple time points as input data;

[0085] 2) Individual prediction data input submodule: receiving single or batch model independent variable information input by the user as input data;

[0086] Data Analysis Module:

[0087] 1) Prediction model building submodule: using the data obtained by the model training data input submodule, executing the modeling steps of step 1) to step 7) in the first aspect, and obtaining a nutritional status transition prediction model based on body weight dynamics and a model performance evaluation index;

[0088] 2) executing the prediction submodule: performing risk calculation based on the nutritional status outcome prediction model obtained by the prediction model building submodule, using the input data obtained by the data acquisition module, and obtaining the nutritional status outcome categories and the prediction probability corresponding to each category;

[0089] Result output module:

[0090] 1) Batch modeling result output submodule: output model performance evaluation indicators according to the modeling results of the prediction model modeling submodule;

[0091] 2) Individual prediction result output submodule: Obtain the nutritional status conversion to each label category and the prediction probability corresponding to each category according to the calculation result output of the execution prediction submodule.

[0092] Embodiment 1:

[0093] In this embodiment, see Figure 1 A method for constructing a nutritional status transition prediction model based on body weight dynamics comprises the following steps:

[0094] 1) Establish a crowd data set:

[0095] A total of 3,610 cases of people diagnosed with cachexia by the Fearon framework at admission were selected, and individual data including age, gender, percentage change in weight from six months ago to one month ago, percentage change in weight from one month ago to admission, body mass index at admission, and cachexia diagnosis based on the Fearon framework one month after admission were collected as data set samples;

[0096] 2) Sample data preprocessing:

[0097] The collected sample data sets were preprocessed, and the continuous variables were z-score standardized. The specific formula was: (x-μ) / σ. μ represents the mean of the overall data, σ represents the standard deviation of the overall data, and x represents the individual observation value. The cachexia prognosis one month after admission was used as the outcome label. The people whose cachexia changed from positive at baseline to negative one month later were coded as 1, otherwise they were coded as 0.

[0098] 3) Divide the training set and test set:

[0099] The sample data set preprocessed in step 2) is randomly divided into a training set and a test set in a ratio of 8:2;

[0100] 4) Modeling feature screening

[0101] The training set divided in step 3) was selected, and cachexia prognosis was used as the outcome variable. The percentage change in weight from six months ago to one month ago, the percentage change in weight from one month ago to the time of admission, and the body mass index at the time of admission were manually selected as independent variables to obtain a subset containing 4 features.

[0102] 5) Build multiple classifiers:

[0103] Based on the training set after feature selection in step 4), supervised machine learning algorithms were used to construct three classifiers in Python 3.9.11 environment, including Gaussian process classifier, multilayer perceptron and gradient boosting machine algorithm. The hyperparameters of each model were found by random search iteration 1000 times, and the model performance parameters in the training set were calculated by 10-fold cross validation, including accuracy, area under the curve, recall rate, precision, F1 score and Kappa index, as shown in Table 1. The model classification performance of Example 1 in the training set after 10-fold cross validation.

[0104] Table 1

[0105] Model Accuracy Area under the curve Recall Accuracy F1 score Kappa Gaussian Process Classifier 0.864 0.916 0.913 0.838 0.874 0.726 Multilayer Perceptron 0.860 0.916 0.907 0.836 0.870 0.718 Gradient Boosting Machine 0.856 0.915 0.901 0.835 0.866 0.712

[0106] 6) Select the best model

[0107] On the basis of the model evaluation index obtained in step 5), the area under the cross-validated curve is used as the main evaluation index, and the candidate classifiers are sorted according to the index evaluation index, and the Gaussian process classifier model with the best performance is selected as the best model, see Table 1;

[0108] 7) Performance evaluation and prediction

[0109] According to the best model obtained in step 6), the cachexia prognosis of the test set samples was quickly predicted, and the precision-recall curve, receiver operating characteristic curve, confusion matrix and the relative importance diagram of the top three variables were drawn respectively. Figure 2 and Figure 3 .

[0110] Embodiment 2:

[0111] This embodiment is substantially the same as the first embodiment. Figure 1 The data analysis module and result output module in the above method are different, but different input features and different machine learning algorithms are used for modeling. A method for constructing a nutritional status transition prediction model based on body weight dynamics includes the following steps:

[0112] 1) Establish a crowd data set:

[0113] A total of 4197 cases of people diagnosed with malnutrition by the GLIM framework at admission were selected, and individual data including age, gender, body mass index six months ago, body mass index one month ago, body mass index at admission, predicted limb skeletal muscle index six months ago, predicted limb skeletal muscle index one month ago, predicted limb skeletal muscle index at admission, and malnutrition diagnosis based on the GLIM framework one month after admission were collected as data set samples;

[0114] 2) Sample data preprocessing:

[0115] The collected sample data sets were preprocessed, and continuous variables were z-score standardized. The specific formula is: (x-μ) / σ. μ represents the mean of the overall data, σ represents the standard deviation of the overall data, and x represents the individual observation value. The malnutrition outcome one month after admission was used as the outcome label. The population whose malnutrition changed from positive at baseline to negative one month later was coded as 1, otherwise it was coded as 0.

[0116] 3) Divide the training set and test set:

[0117] The sample data set preprocessed in step 2) is randomly divided into a training set and a test set in a ratio of 8:2;

[0118] 4) Modeling feature screening

[0119] The training set divided in the step 3) was selected, malnutrition outcome was used as the outcome variable, and the body mass index six months ago, the body mass index one month ago, the body mass index at admission, the predicted limb skeletal muscle index six months ago, the predicted limb skeletal muscle index one month ago, and the predicted limb skeletal muscle index at admission were manually selected as independent variables to obtain a subset containing 7 features.

[0120] 5) Build multiple classifiers:

[0121] Based on the training set after feature selection in step 4), supervised machine learning algorithms were used in the Python 3.9.11 environment to build three classifiers, including gradient boosting machine, lightweight gradient boosting machine and extreme gradient boosting machine. The hyperparameters of each model were found by random search iteration 1000 times, and the model performance parameters in the training set were calculated by 10-fold cross validation, including accuracy, area under the curve, recall rate, precision, F1 score and Kappa index, as shown in Table 2. The model classification performance of Example 2 in the training set after 10-fold cross validation.

[0122] Table 2

[0123] Model Accuracy Area under the curve Recall Accuracy F1 score Kappa Gradient Boosting Machine 0.885 0.828 0.471 0.752 0.578 0.515 Lightweight Gradient Boosting Machine 0.884 0.820 0.503 0.726 0.593 0.528 Extreme Gradient Boosting Machine 0.881 0.818 0.501 0.702 0.583 0.516

[0124] 6) Select the best model

[0125] On the basis of the model evaluation index obtained in step 5), the area under the cross-validation curve is used as the main evaluation index, and the candidate classifiers are sorted according to the index evaluation index, and the gradient boosting machine model with the best performance is selected as the best model, see Table 2;

[0126] 7) Performance evaluation and prediction

[0127] According to the best model obtained in step 6), the cachexia prognosis of the test set samples was quickly predicted, and the precision-recall curve, receiver operating characteristic curve, confusion matrix and the relative importance diagram of the top three variables were drawn respectively. Figure 4 and Figure 5 .

[0128] The results show that the present invention can use the dynamic weight information to achieve high-performance prediction of common poor nutritional status without the assistance of other information. The method and system for constructing a nutritional status prognosis prediction model based on weight dynamics proposed in the present invention is expected to be used as an auxiliary decision-making tool to achieve nutritional status prognosis prediction in various scenarios such as medical institutions, public health, and medical care institutions and families.

[0129] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.

Claims

1. A method for constructing a nutritional status outcome prediction model based on body weight dynamics, characterized in that: include: Establishing a population data set, wherein the population data set includes a plurality of individuals, each individual including weight information at a plurality of time points and nutritional status at the corresponding time points; Preprocessing the sample data, including calculating the weight loss percentage and defining an outcome label, wherein the outcome label is determined according to changes in the nutritional status of the individual at the multiple time points; Dividing the population data set into a training set and a test set; dividing the population data set into a training set and a test set; Modeling feature screening, screening out features related to the outcome label from the training set; Constructing multiple machine learning classifiers based on the screened features and the training set; Selecting the best model, selecting the best model according to the performance indicator of the machine learning classifier; Performance evaluation and prediction: using the best model to predict the test set and evaluate the performance of the prediction model.

2. The method for constructing a nutritional status outcome prediction model based on body weight dynamics according to claim 1, characterized in that: The weight information at multiple time points includes: Current weight; Historical weight, including historical weight data at multiple time points.

3. The method for constructing a nutritional status outcome prediction model based on body weight dynamics according to claim 1, characterized in that: The nutritional status was diagnosed according to the GLIM standard, Fearon standard or SGA scale.

4. The method for constructing a nutritional status outcome prediction model based on body weight dynamics according to claim 1, characterized in that: The ending tags include: Examples of nutritional status transitions, i.e., changes in the nutritional status of the individual at the multiple time points; An example of a continuous nutritional state is one in which the nutritional state of the individual remains constant at the multiple time points.

5. The method for constructing a nutritional status outcome prediction model based on body weight dynamics according to claim 1, characterized in that: The machine learning classifier includes: Random Forest; Multilayer Perceptron; Long short-term memory network; Transformer.

6. The method for constructing a nutritional status outcome prediction model based on body weight dynamics according to claim 1, characterized in that: The performance indicators include: Accuracy; Area under the curve; Recall rate; Accuracy; F1 score; Kappa index.

7. A nutritional status prognosis prediction system based on body weight dynamics, characterized by: include: A data collection module, used to collect weight information and nutritional status information of multiple individuals; A data analysis module, used to construct a nutritional status outcome prediction model based on the weight information and nutritional status information; The result output module is used to output the prediction result of the prediction model.

8. The nutritional status prognosis prediction system based on body weight dynamics according to claim 7, characterized in that: The data acquisition module comprises: A model training data input submodule, used to collect training data for building the prediction model; The individual prediction data input submodule is used to collect individual data for predicting nutritional status outcomes.

9. The nutritional status prognosis prediction system based on body weight dynamics according to claim 7, characterized in that: The data analysis module includes: A prediction model building submodule, used to build the prediction model according to the training data; The prediction submodule is executed to predict the individual data using the prediction model.

10. The nutritional status prognosis prediction system based on body weight dynamics according to claim 7, characterized in that: The result output module includes: A batch modeling result output submodule, used to output the performance indicators of the prediction model; The individual prediction result output submodule is used to output the prediction result of the prediction model.

Images