Infant nutrition demand prediction method based on machine learning
Through the infant nutrition demand prediction method based on machine learning, the model is trained using intelligent optimization algorithms and deployed in the cloud computing center, the problem of personalized nutrition management in traditional methods is solved, and efficient nutrition advice and management is achieved to ensure the healthy growth of babies.
Patent Information
- Application Number
- CN202510539380.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-27
- Publication Date
- 2025-07-18
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Traditional infant nutrition guidance methods are difficult to meet personalized needs, and it is difficult for users to manage their nutritional needs on their own, so relying on professional physicians to guide is not convenient enough.
Using a machine learning-based method, we use intelligent optimization algorithms to train machine learning models by obtaining infant body characteristic data and nutritional needs indicators, and deploying them in the cloud computing center to achieve the prediction and management of infant nutrition needs.
Provide personalized nutrition suggestions to improve the accuracy of nutritional demand forecasting, help users manage their baby's nutritional needs and ensure healthy growth.
Smart Images
Figure CN120340769A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical fields of machine learning and nutritional requirement prediction, and particularly relates to a method for predicting infant nutritional requirements based on machine learning. Background Art
[0002] Predicting infant nutritional requirements is to scientifically calculate and evaluate various nutrients required by infants, such as protein, fat, carbohydrates, vitamins, and minerals, according to the growth and development characteristics of infants. Through prediction, a reasonable dietary plan is formulated to ensure that infants obtain sufficient and balanced nutrition at different stages, supporting their healthy growth. This prediction helps to guide parents to feed rationally, prevent malnutrition and overnutrition, reduce the risk of diseases, and lay a solid foundation for the health of infants throughout their lives. Infancy is a crucial stage in human growth and development, and reasonable nutritional intake is essential for the healthy growth of infants. However, due to individual differences, there are significant differences in the nutritional requirements of infants. Traditional nutritional guidance methods often rely on general rules and are difficult to meet the personalized needs of infants, or rely on professional physicians for guidance, making it difficult for users to manage the nutritional requirements of infants on their own. Therefore, it is necessary to study a method that can predict the nutritional requirements of infants. Summary of the Invention
[0003] The present invention provides a method for predicting infant nutritional requirements based on machine learning to solve the problem in the prior art that it relies on professional physicians for guidance and it is difficult for users to manage the nutritional requirements of infants on their own.
[0004] A method for predicting infant nutritional requirements based on machine learning includes: Obtaining infant physical characteristic data and nutritional requirement indicators corresponding to the infant physical characteristic data determined by professional physicians; wherein, the nutritional requirement indicators are one or more; For any one of the nutritional requirement indicators, training a machine learning model by using an intelligent optimization algorithm according to the infant physical characteristic data and its corresponding nutritional requirement indicators to obtain an infant nutritional requirement prediction model; Deploying the infant nutritional requirement prediction model to a cloud computing center, obtaining the to-be-identified infant physical characteristic data input by a user, and scheduling the infant nutritional requirement prediction model in the cloud computing center to identify the to-be-identified infant physical characteristic data to obtain an infant nutritional requirement prediction result.
[0005] In a possible implementation manner, the infant physical characteristic data includes infant gender, infant age, infant weight for consecutive days, infant height for consecutive days, and nutritional supplement indicators for consecutive days.
[0006] In a possible implementation, according to the baby's body characteristic data and its corresponding nutritional requirement indicators, a machine learning model is trained using an intelligent optimization algorithm to obtain a baby nutritional requirement prediction model, including: Based on the hyperparameters of the machine learning model, a population is generated; For any individual in the population, using the baby's body characteristic data and its corresponding nutritional requirement indicators as training data, the loss function value corresponding to the individual is obtained; Based on the loss function values corresponding to all individuals, the current optimal individual is determined; Based on the current optimal individual, the individual is trained in sequence using a collision avoidance strategy with a self-adjusting search route, a mapping information interaction strategy, and a spiral Levy search strategy to determine the target optimal individual; Based on the target optimal individual, the baby nutritional requirement prediction model is obtained.
[0007] In a possible implementation, based on the hyperparameters of the machine learning model, generating a population includes: For any hyperparameter of the machine learning model, it is randomly initialized between the upper and lower limits of the hyperparameter, and the hyperparameters after random initialization are combined into a vector to obtain an individual; After multiple random initializations, multiple individuals are obtained, and the multiple individuals are combined into a population for training; or, based on the individual obtained by random initialization, a chaotic mapping sequence is used to obtain multiple individuals, and the multiple individuals are combined into a population for training.
[0008] In a possible implementation, using the baby's body characteristic data and its corresponding nutritional requirement indicators as training data to obtain the loss function value corresponding to the individual includes: after applying the hyperparameters included in the individual to the machine learning model, using the baby's body characteristic data as the actual input data of the machine learning model, using the nutritional requirement indicators corresponding to the baby's body characteristic data as the expected output data, and using the cross-entropy loss function to obtain the loss function value corresponding to the individual.
[0009] In a possible implementation, based on the current optimal individual, training the individual in sequence using a collision avoidance strategy with a self-adjusting search route, a mapping information interaction strategy, and a spiral Levy search strategy to determine the target optimal individual includes: For any individual, based on the current optimal individual, training the individual using a collision avoidance strategy with a self-adjusting search route to obtain a first target individual; For any first target individual, training the first target individual using a mapping information interaction strategy to obtain a second target individual; For any second target individual, the spiral Levy search strategy is adopted to train the second target individual to obtain a third target individual; Determine whether the current number of training times has reached the maximum number of training times. If so, determine the target optimal individual according to the third target individual in the current training process. Otherwise, use the third target individual as the input in the next training process and continue training.
[0010] In a possible implementation manner, for any individual, based on the current optimal individual, an anti-collision strategy with a self-adjusting search route is adopted to train the individual to obtain a first target individual, including: Based on the current number of training times, obtain an adaptive search route adjustment factor; For any individual, use the adaptive search route adjustment factor to perform anti-collision processing on the individual to obtain the individual after anti-collision processing; According to the adaptive search route adjustment factor, determine an information learning factor, and according to the information learning factor and the current optimal individual, obtain optimal learning information; According to the individual after anti-collision processing and the optimal learning information, train the individual to obtain a first target individual.
[0011] In a possible implementation manner, for any first target individual, a mapping information interaction strategy is adopted to train the first target individual to obtain a second target individual, including: For any first target individual, randomly match two other different first target individuals for the first target individual to obtain a first information interaction individual and a second information interaction individual corresponding to the first target individual; According to the first information interaction individual and the second information interaction individual, perform mapping information interaction to determine individual interaction information; According to the individual interaction information, train the first target individual to obtain a second target individual.
[0012] In a possible implementation manner, for any second target individual, the spiral Levy search strategy is adopted to train the second target individual to obtain a third target individual, including: For any second target individual, obtain a spiral search factor and a dynamic convergence factor; Based on the spiral search factor, perform a spiral transformation on the second target individual to obtain the second target individual after spiral transformation; Use Levy flight to generate a random individual, and obtain random search information according to the dynamic convergence factor and the random individual; Train the second target individual according to the second target individual after the spiral transformation, the current optimal individual, and the random search information to obtain a third target individual.
[0013] In a possible implementation manner, it further includes: When the predicted result of the infant's nutritional needs meets the preset warning condition, generate a nutritional needs warning message and transmit the nutritional needs warning message to the corresponding device.
[0014] A method for predicting infant nutritional needs based on machine learning provided by the present invention, by obtaining infant physical characteristic data and nutritional requirement indicators corresponding to the infant physical characteristic data determined by professional physicians, then using an intelligent optimization algorithm to train a machine learning model according to the infant physical characteristic data and its corresponding nutritional requirement indicators to obtain an infant nutritional needs prediction model, and finally obtaining the infant physical characteristic data to be recognized input by the user, scheduling the infant nutritional needs prediction model of the cloud computing center to recognize the infant physical characteristic data to be recognized, and obtaining the predicted result of the infant's nutritional needs, can effectively provide higher nutritional suggestions, assist ordinary users in realizing infant nutritional management, and ensure the healthy growth of infants. Description of the Drawings
[0015] The drawings here are incorporated into the specification and form a part of this specification, showing embodiments consistent with the present invention, and are used together with the specification to explain the principles of the present invention.
[0016] Figure 1 It is a flowchart of a method for predicting infant nutritional needs based on machine learning provided by an embodiment of the present invention.
[0017] Figure 2 It is a flowchart of obtaining an infant nutritional needs prediction model provided by an embodiment of the present invention.
[0018] Figure 3 It is a flowchart of obtaining a target optimal individual provided by an embodiment of the present invention.
[0019] Through the above drawings, specific embodiments of the present invention have been shown, and there will be more detailed descriptions later. These drawings and textual descriptions are not intended to limit the scope of the inventive concept in any way, but to illustrate the concept of the present invention to those skilled in the art by referring to specific embodiments. Detailed Embodiments
[0020] Exemplary embodiments will be described in detail herein, and examples thereof are shown in the accompanying drawings. When the following description refers to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. On the contrary, they are merely examples of devices and methods consistent with some aspects of the present invention as detailed in the appended claims.
[0021] Embodiments of the present invention will be described in detail below with reference to the accompanying drawings.
[0022] As Figure 1 shown, an embodiment of the present invention provides a method for predicting infant nutritional needs based on machine learning, including: S11. Obtain infant physical characteristic data and nutritional requirement indicators corresponding to the infant physical characteristic data determined by a professional physician; wherein, the nutritional requirement indicators are one or more; The infant physical characteristic data can be information such as the infant's growth and development data, eating habits, physiological indicators, etc. These data often indicate the infant's nutritional needs. Therefore, based on this basic data, the infant's nutritional needs can be predicted. The nutritional requirement indicators generally can be a certain nutrient (such as calcium). By establishing the relationship between the infant physical characteristic data and the nutritional requirement indicators, the prediction of the infant's nutritional needs based on machine learning can be achieved.
[0023] S12. For any one of the nutritional requirement indicators, according to the infant physical characteristic data and its corresponding nutritional requirement indicators, use an intelligent optimization algorithm to train a machine learning model to obtain an infant nutritional needs prediction model; Although machine learning algorithms can effectively achieve data prediction, the accuracy of joint prediction of multiple data will decrease. Therefore, in an embodiment of the present invention, a machine learning model is constructed for each nutritional requirement indicator, and the machine learning model is used to learn the data relationship between the infant physical characteristic data and its corresponding nutritional requirement indicators, so that an infant nutritional needs prediction model corresponding to each nutritional requirement indicator can be obtained.
[0024] The process of training a machine learning model is essentially the process of training its hyperparameters. Therefore, an intelligent optimization algorithm can be used to train the machine learning model. For example, a particle swarm optimization algorithm can be used to train the machine learning model.
[0025] S13. Deploy the infant nutritional needs prediction model to a cloud computing center, obtain the to-be-identified infant physical characteristic data input by a user, and schedule the infant nutritional needs prediction model of the cloud computing center to identify the to-be-identified infant physical characteristic data to obtain an infant nutritional needs prediction result.
[0026] The infant nutrition demand prediction model is a machine learning model after training. Deploying the infant nutrition demand prediction model in the cloud computing center allows users to directly access the cloud. It can not only effectively utilize the powerful data processing capabilities of the cloud computing center but also achieve online data management.
[0027] In a possible implementation, the infant physical characteristic data includes infant gender, infant age, infant weight for consecutive days, infant height for consecutive days, and nutrition supplement indicators for consecutive days.
[0028] It should be noted that in addition to the above-mentioned infant physical characteristic data, other data can also be used for prediction, such as bone thickness, blood test indicators, and so on.
[0029] As Figure 2 shown, according to the infant physical characteristic data and its corresponding nutrition demand indicators, an intelligent optimization algorithm is used to train the machine learning model to obtain the infant nutrition demand prediction model, including: S21. Generate a population based on the hyperparameters of the machine learning model; The machine learning model can be set to a support vector machine (SVM), random forest (RF), neural network (NN), etc. Since the infant physical characteristic data collected in the embodiments of the present invention not only contains various associated data but also includes time series data, the CNN-LSTM (Convolutional Neural Network-Long Short-Term Memory Network) model is mainly used in the embodiments of the present invention for training to ensure the data recognition effect.
[0030] S22. For any individual in the population, use the infant physical characteristic data and its corresponding nutrition demand indicators as training data to obtain the loss function value corresponding to the individual; For example, the root mean square loss function or cross-entropy loss function can be used to obtain the loss function value corresponding to the individual. The smaller the loss function value, the better the position of the individual in the solution space.
[0031] S23. Determine the current optimal individual based on the loss function values corresponding to all individuals; S24. Based on the current optimal individual, sequentially use the collision avoidance strategy with self-adjusting search routes, the mapping information interaction strategy, and the spiral Levy search strategy to train the individuals to determine the target optimal individual; S25. Based on the target optimal individual, obtain the infant nutrition demand prediction model, that is, use the hyperparameters in the target optimal individual as the final hyperparameters of the machine learning model to obtain the infant nutrition demand prediction model.
[0032] In the prior art, the particle swarm algorithm is generally used to optimize the machine learning model, which often leads to poor training effects and fails to effectively achieve prediction. Therefore, the embodiments of the present invention provide a new intelligent optimization algorithm to improve the training effect and ultimately improve the accuracy of predicting the nutritional needs of infants.
[0033] In a possible implementation manner, based on the hyperparameters of the machine learning model, a population is generated, including: For any hyperparameter of the machine learning model, random initialization is performed between the upper limit and the lower limit of the hyperparameter, and the hyperparameters after random initialization are combined into a vector to obtain an individual. After multiple random initializations, multiple individuals are obtained, and the multiple individuals are combined into a population for training; alternatively, based on the individual obtained by random initialization, a chaotic mapping sequence is used to obtain multiple individuals, and the multiple individuals are combined into a population for training.
[0034] In a possible implementation manner, using the infant physical characteristic data and its corresponding nutritional requirement indicators as training data, the loss function value corresponding to the individual is obtained, including: after applying the hyperparameters included in the individual to the machine learning model, using the infant physical characteristic data as the actual input data of the machine learning model, using the nutritional requirement indicators corresponding to the infant physical characteristic data as the expected output data, and using the cross-entropy loss function to obtain the loss function value corresponding to the individual.
[0035] As Figure 3 shown, based on the current optimal individual, the individual is trained in turn using the collision avoidance strategy with self-adjusting search route, the mapping information interaction strategy, and the spiral Levy search strategy to determine the target optimal individual, including: S2401. For any individual, based on the current optimal individual, the individual is trained using the collision avoidance strategy with self-adjusting search route to obtain a first target individual. S2402. For any first target individual, the first target individual is trained using the mapping information interaction strategy to obtain a second target individual. S2403. For any second target individual, the second target individual is trained using the spiral Levy search strategy to obtain a third target individual. S2404. Determine whether the current training times reach the maximum training times. If so, based on the third target individual in the current training process, the target optimal individual is determined; otherwise, the third target individual is used as the input in the next training process and the training continues.
[0036] In one possible implementation, for any individual, based on the current optimal individual, an anti-collision strategy with self-adjusting search route is used to train the individual to obtain a first target individual, including: S2401.1. Based on the current number of training times, obtain the self-adaptive search route adjustment factor as: ; where represents the self-adaptive search route adjustment factor, represents the preset maximum value of the self-adaptive search route adjustment factor, which can be set to 2, represents the current number of training times, represents the preset maximum number of training times; S2401.2. For any individual, use the self-adaptive search route adjustment factor to perform anti-collision processing on the individual, and the individual after anti-collision processing is ; where represents the th individual in the k th training process, k = 1, 2, …, K , and K represents the total number of individuals in the population; S2401.3. Determine the information learning factor according to the self-adaptive search route adjustment factor, and obtain the optimal learning information according to the information learning factor and the current optimal individual; The information learning factor is: ; where B represents the information learning factor, represents a random number between [0, 1]; The optimal learning information is: ; where represents the current optimal individual, represents the optimal learning information corresponding to the th individual in the k th training process; S2401.4. Train the individual according to the individual after anti-collision processing and the optimal learning information, and the first target individual obtained is: ; where represents the k th first target individual.
[0037] The anti-collision strategy with self-adjusting search route provided by the embodiments of the present invention can effectively avoid collisions during the search process, improve the traversability of the solution space, thereby enhancing the global search ability. At the same time, learning the position information of the current optimal individual can effectively improve the training speed of the algorithm.
[0038] In a possible implementation manner, for any first target individual, training the first target individual by using a mapping information interaction strategy to obtain a second target individual, including: S2402.1. For any first target individual, randomly match two other different first target individuals to the first target individual, and obtain a first information interaction individual and a second information interaction individual corresponding to the first target individual; S2402.2. According to the first information interaction individual and the second information interaction individual, perform mapping information interaction to determine that the individual interaction information is: ; where represents the first information interaction individual corresponding to the th training process and the m th first target individual, represents the second information interaction individual corresponding to the th training process and the m th first target individual, represents the individual interaction information corresponding to the th training process and the m th first target individual, = 1, 2,..., m , K represents the total number of individuals in the population, K represents a random number between [0, 1]; S2402.3. According to the individual interaction information, train the first target individual to obtain the second target individual as: ; where represents the m th second target individual.
[0039] The mapping information interaction strategy provided by the embodiments of the present invention can effectively use any two first target individuals as references to search the local solution space, can improve the regional search ability between the two first target individuals, and to a certain extent strengthen the global search ability and improve the ability to find the global optimal solution.
[0040] In a possible implementation manner, for any second target individual, training the second target individual by using a spiral Levy search strategy to obtain a third target individual, including: S2403.1. For any second target individual, obtain a spiral search factor and a dynamic convergence factor; The spiral search factor is: ; where represents the spiral search factor, represents the dynamic convergence factor, represents the spiral search radius, represents a random spiral search angle within [0, 2 . represents pi, , u represents the first spiral shape control constant, v represents the second spiral shape control constant, and both the first spiral shape control constant and the second spiral shape control constant can be set to 1, where e represents the natural constant; The dynamic convergence factor is: ; where represents the dynamic convergence factor, represents a random number within [-1, 1], represents the current training iteration, represents the maximum training iteration; S2403.2. Based on the spiral search factor, perform a spiral transformation on the second target individual to obtain the second target individual after the spiral transformation as: ; where represents the th n second target individual during the n th training process, K = 1, 2,..., and K represents the total number of individuals in the population; S2403.3. Generate a random individual using Levy flight and obtain the random search information based on the dynamic convergence factor and the random individual as: ; where represents the random individual, that is, each element of its dimensions is generated using Levy flight, and its total number of dimensions is the same as that of other individuals; S2403.4. Based on the second target individual after the spiral transformation, the current optimal individual, and the random search information, train the second target individual to obtain the third target individual as: ; where represents the current optimal individual, n represents the
[0041] The spiral Levy search strategy provided by the embodiments of the present invention can effectively learn the optimal position information in the early stage of the algorithm, perform spiral search, and at the same time add Levy flight for perturbation, which can effectively achieve global search, while gradually converting to local area search of the optimal position in the later stage of the algorithm, improving the training accuracy of the algorithm.
[0042] Through the mutual cooperation of the above-mentioned collision avoidance strategy for self-adjusting the search route, the mapping information interaction strategy, and the spiral Levy search strategy, the global training ability and training accuracy of the algorithm can be effectively improved, thereby enhancing the accuracy of predicting the nutritional needs of infants.
[0043] In a possible implementation manner, the method for predicting the nutritional needs of infants based on machine learning further includes: When the predicted result of the infant's nutritional needs meets the preset warning condition, a nutritional needs warning message is generated and transmitted to the corresponding device.
[0044] Optionally, in the embodiments of the present invention, the predicted result of the infant's nutritional needs is essentially the predicted value of a certain nutritional need index. Therefore, the preset warning condition can be set as a threshold. When the predicted result of the infant's nutritional needs exceeds the corresponding threshold, a nutritional needs warning message can be generated, thereby realizing nutritional needs warning and assisting users to better manage the physical health of infants.
[0045] The method for predicting the nutritional needs of infants based on machine learning provided by the present invention obtains the physical characteristic data of the infant and the nutritional need index corresponding to the physical characteristic data of the infant determined by a professional doctor, and then uses an intelligent optimization algorithm to train a machine learning model according to the physical characteristic data of the infant and its corresponding nutritional need index to obtain an infant nutritional needs prediction model. Finally, the physical characteristic data of the to-be-identified infant input by the user is obtained, and the infant nutritional needs prediction model in the cloud computing center is scheduled to identify the physical characteristic data of the to-be-identified infant, and the predicted result of the infant's nutritional needs is obtained, which can effectively provide higher nutritional suggestions, assist ordinary users in realizing infant nutrition management, and ensure the healthy growth of infants.
[0046] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0047] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processors of general-purpose computers, special-purpose computers, embedded processors, or other programmable data processing devices to generate a machine, such that the instructions executed by the processors of the computer or other programmable data processing devices produce means for implementing the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.
[0048] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufactured article including instruction means that implement the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.
[0049] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are performed on the computer or other programmable device to generate a computer-implemented process, so that the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.
[0050] Those of ordinary skill in the art can understand that all or part of the steps in implementing the above facts and methods can be completed by instructing relevant hardware through a program. The involved program or the described program can be stored in a computer-readable storage medium. When the program is executed, it includes the following steps: At this time, the corresponding method steps are introduced. The storage medium can be ROM / RAM, magnetic disks, optical discs, etc.
[0051] The specific embodiments described above further elaborate on the purpose, technical solutions, and beneficial effects of the present invention. It should be understood that the above description is only the specific embodiments of the present invention and is not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims
1. A method for predicting the nutritional needs of infants based on machine learning, characterized in that, Including: Obtain the physical characteristic data of the baby and the nutritional requirement indicators corresponding to the physical characteristic data of the baby determined by professional physicians; wherein, the nutritional requirement indicators are one or more; For any one of the nutritional requirement indicators, according to the physical characteristic data of the baby and its corresponding nutritional requirement indicators, use an intelligent optimization algorithm to train a machine learning model to obtain a baby nutritional requirement prediction model; Deploy the baby nutritional requirement prediction model to the cloud computing center, and obtain the physical characteristic data of the baby to be recognized input by the user, and schedule the baby nutritional requirement prediction model in the cloud computing center to recognize the physical characteristic data of the baby to be recognized, and obtain the baby nutritional requirement prediction result.
2. The method for predicting the nutritional needs of infants based on machine learning according to claim 1, wherein The physical characteristic data of the baby includes the baby's gender, the baby's age, the baby's weight for consecutive days, the baby's height for consecutive days, and the nutritional supplement indicators for consecutive days.
3. The method for predicting the nutritional needs of infants based on machine learning according to claim 1, wherein According to the physical characteristic data of the baby and its corresponding nutritional requirement indicators, use an intelligent optimization algorithm to train a machine learning model to obtain a baby nutritional requirement prediction model, including: Generate a population based on the hyperparameters of the machine learning model; For any individual in the population, use the physical characteristic data of the baby and its corresponding nutritional requirement indicators as training data, and obtain the loss function value corresponding to the individual; Based on the loss function values corresponding to all individuals, determine the current optimal individual; Based on the current optimal individual, successively use the collision avoidance strategy with self-adjusting search route, the mapping information interaction strategy, and the spiral Levy search strategy to train the individual to determine the target optimal individual; Based on the target optimal individual, obtain the baby nutritional requirement prediction model.
4. The method for predicting the nutritional needs of infants based on machine learning according to claim 3, characterized in that, Generate a population based on the hyperparameters of the machine learning model, including: For any hyperparameter of the machine learning model, perform random initialization between the upper limit and the lower limit of the hyperparameter, and form a vector with the hyperparameters after random initialization to obtain an individual; After multiple random initializations, obtain multiple individuals, and form a population with the multiple individuals for training; or, based on the individual obtained by random initialization, use a chaotic mapping sequence to obtain multiple individuals, and form a population with the multiple individuals for training.
5. The method for predicting the nutritional needs of infants based on machine learning according to claim 3, wherein Use the physical characteristic data of the baby and its corresponding nutritional requirement indicators as training data, and obtain the loss function value corresponding to the individual, including: after applying the hyperparameters included in the individual to the machine learning model, use the physical characteristic data of the baby as the actual input data of the machine learning model, use the nutritional requirement indicators corresponding to the physical characteristic data of the baby as the expected output data, and use the cross-entropy loss function to obtain the loss function value corresponding to the individual.
6. The method for predicting infant nutritional needs based on machine learning according to claim 3, wherein, Based on the current optimal individual, successively use the collision avoidance strategy with self-adjusting search route, the mapping information interaction strategy, and the spiral Levy search strategy to train the individual to determine the target optimal individual, including: For any individual, based on the current optimal individual, use the collision avoidance strategy with self-adjusting search route to train the individual to obtain the first target individual; For any first target individual, the first target individual is trained using a mapping information interaction strategy to obtain a second target individual; For any second target individual, the second target individual is trained using a spiral Levy search strategy to obtain a third target individual; Determine whether the current number of training times has reached the maximum number of training times. If so, based on the third target individual in the current training process, determine the target optimal individual. Otherwise, use the third target individual as the input in the next training process and continue training.
7. The method for predicting infant nutritional requirements based on machine learning according to claim 6, wherein, For any individual, based on the current optimal individual, the individual is trained using a collision avoidance strategy with self-adjusting search routes to obtain a first target individual, including: Based on the current number of training times, obtain an adaptive search route adjustment factor; For any individual, use the adaptive search route adjustment factor to perform collision avoidance processing on the individual to obtain the individual after collision avoidance processing; Based on the adaptive search route adjustment factor, determine an information learning factor, and based on the information learning factor and the current optimal individual, obtain optimal learning information; Based on the individual after collision avoidance processing and the optimal learning information, train the individual to obtain a first target individual.
8. The method for predicting infant nutritional needs based on machine learning according to claim 7, wherein For any first target individual, the first target individual is trained using a mapping information interaction strategy to obtain a second target individual, including: For any first target individual, randomly match two other different first target individuals to the first target individual to obtain a first information interaction individual and a second information interaction individual corresponding to the first target individual; Based on the first information interaction individual and the second information interaction individual, perform mapping information interaction to determine individual interaction information; Based on the individual interaction information, train the first target individual to obtain a second target individual.
9. The method for predicting infant nutritional requirements based on machine learning according to claim 8, wherein For any second target individual, the second target individual is trained using a spiral Levy search strategy to obtain a third target individual, including: For any second target individual, obtain a spiral search factor and a dynamic convergence factor; Based on the spiral search factor, perform a spiral transformation on the second target individual to obtain the second target individual after spiral transformation; Use Levy flight to generate a random individual, and obtain random search information based on the dynamic convergence factor and the random individual; Based on the second target individual after spiral transformation, the current optimal individual, and the random search information, train the second target individual to obtain a third target individual.
10. The method for predicting the nutritional needs of infants based on machine learning according to claim 1, wherein It also includes: When the predicted result of the infant's nutritional needs meets the preset warning condition, a nutritional needs warning message is generated and the nutritional needs warning message is transmitted to the corresponding device.