Intelligent prediction method of hunger and satiety cycle based on individual digestive rhythm modeling

Abstract

The application relates to the technical field of intelligent feeding, and discloses a hunger and satiety cycle intelligent prediction method and system based on individual digestive rhythm modeling, a computer device and a storage medium, the method comprising the following steps: acquiring feeding related data, respiratory related physiological signal data and month age data of an infant; determining a gastric emptying time parameter according to the feeding related data; extracting a respiratory rhythm feature parameter according to the respiratory related physiological signal data; determining a digestive enzyme activity maturity parameter according to the month age data; determining a hunger threshold delay parameter according to food type data and the digestive enzyme activity maturity parameter; determining a respiratory feature weight parameter according to the respiratory rhythm feature parameter and the month age data; inputting the above parameters into a multi-parameter fusion prediction model to calculate a hunger signal prediction time and output a feeding reminder. The application realizes objective quantification prediction of a hunger-satiety cycle based on the individual digestive rhythm of an infant.

Description

Technical Field

[0001] This invention relates to the field of intelligent feeding technology, specifically to an intelligent prediction method, system, computer device, and storage medium for hunger and satiety cycles based on individual digestive rhythm modeling. Background Technology

[0002] With the deepening application of sensor technology and artificial intelligence algorithms in the healthcare field, infant feeding monitoring technology has made significant progress. Existing technologies can now collect physiological signals from infants using various sensors. For example, they can assess feeding behavior by collecting facial waveforms representing sucking movements, neck waveforms representing swallowing movements, and chest and abdominal waveforms representing breathing movements; or monitor neonatal gastric motility using electrogastrography (EGG) signals and abdominal sound signals; some methods also determine an infant's feeding needs by monitoring changes in respiratory rate and combining this with other monitoring indicators. Furthermore, there are also methods that integrate multi-dimensional physiological data such as infant age, weight, heart rate, and activity level to assess feeding needs.

[0003] However, existing technologies for predicting infant hunger-fullness cycles primarily rely on external behavioral characteristics (such as crying, body movements, and facial expressions) or single-dimensional physiological signals (such as the absolute value of respiratory rate), failing to incorporate microscopic physiological parameters affecting the infant's digestive process into the prediction model. Specifically, existing technologies have not established quantitative models of the relationship between gastric emptying time and feeding volume and food type, have not extracted refined respiratory rhythm characteristics such as the degree of fluctuation in respiratory rate and the frequency of abrupt changes in respiratory depth as predictive indicators of hunger, and have not considered the impact of the maturation of infant digestive enzyme activity with age on the hunger cycle. Due to the lack of quantitative modeling of the internal physiological processes of an individual infant's digestive system, existing methods struggle to achieve accurate predictions of hunger-fullness cycles for different infants, different feeding methods, and different age stages. Summary of the Invention

[0004] Therefore, it is necessary to provide a method, system, computer equipment, and storage medium for intelligent prediction of hunger and satiety cycles based on individual digestive rhythm modeling, which can achieve objective quantitative prediction of hunger-satiety cycles based on individual infant digestive rhythms.

[0005] In a first aspect, the present invention discloses an intelligent prediction method for hunger and satiety cycles based on individual digestive rhythm modeling, the method comprising: Acquire infant feeding-related data, respiratory-related physiological signal data, and age data, wherein the feeding-related data includes single feeding amount data and food type data; Based on the single feeding volume data and the food type data, the gastric emptying time parameter is determined by a preset gastric emptying quantification model; Respiratory rhythm feature parameters are extracted from the respiratory-related physiological signal data. The respiratory rhythm feature parameters include frequency feature parameters that characterize the degree of fluctuation in respiratory rate and depth feature parameters that characterize the degree of abrupt changes in respiratory depth. The digestive enzyme activity maturity parameters were determined based on the age data. Based on the food type data, the inherent digestion delay parameter of the food type is determined, and the inherent digestion delay parameter of the food type is corrected based on the digestive enzyme activity maturity parameter to obtain the hunger threshold delay parameter; The respiratory feature weight parameters are determined based on the respiratory rhythm characteristic parameters and the age data in months. The gastric emptying time parameter, the hunger threshold delay parameter, the respiratory feature weight parameter, and the respiratory rhythm feature parameter are input into a preset multi-parameter fusion prediction model to calculate the hunger signal prediction time. When the current time reaches the predicted hunger signal time, a feeding reminder signal is output.

[0006] In one embodiment, determining the digestive enzyme activity maturity parameter based on the age data includes: According to the preset digestive enzyme activity maturity function relationship, the digestive enzyme activity maturity parameter is calculated based on the age data. The digestive enzyme activity maturity parameter is positively correlated with the age data and the growth rate decreases with increasing age. The digestive enzyme activity maturity function relationship is an exponential decay function with a function value between 0 and 1, which is used to characterize the relative maturity of infant digestive enzyme activity with increasing age.

[0007] In one embodiment, the step of extracting respiratory rhythm feature parameters based on the respiratory-related physiological signal data includes: The respiratory-related physiological signal data are subjected to noise reduction preprocessing. Respiratory cycle sequence and respiratory depth sequence were extracted from the noise-reduced and preprocessed respiratory-related physiological signal data. The frequency characteristic parameters are calculated based on the respiratory cycle sequence; The depth feature parameters are calculated based on the respiratory depth sequence.

[0008] In one embodiment, determining the respiratory feature weight parameters based on the respiratory rhythm feature parameters and the age data includes: According to a preset weighting relationship, the hunger index is calculated based on the frequency feature parameters and their corresponding first weighting coefficients, the depth feature parameters and their corresponding second weighting coefficients, and the hunger index is used to characterize the infant's current hunger level. The respiratory feature weight parameters are calculated based on the hunger index and the age data according to the preset respiratory feature weight function relationship.

[0009] In one embodiment, after the step of calculating the hunger index according to a preset weighting relationship based on the frequency feature parameters and their corresponding first weighting coefficients, and the depth feature parameters and their corresponding second weighting coefficients, the method further includes: The hunger stage of an infant is determined based on the numerical range of the hunger index, which includes early hunger stage, middle hunger stage and late hunger stage; Based on the determined hunger stage, the weighting coefficients used to calculate the subsequent hunger index are updated according to a preset weighting coefficient adjustment strategy. The weighting coefficients corresponding to the early hunger stage are used to maintain baseline sensitivity when the infant is in a non-hunger state. The weighting coefficients corresponding to the middle hunger stage and the late hunger stage are gradually increased to enhance the contribution of respiratory rhythm characteristics to the hunger index.

[0010] In one embodiment, before the step of inputting the gastric emptying time parameter, the hunger threshold delay parameter, the respiratory feature weight parameter, and the respiratory rhythm feature parameter into a preset multi-parameter fusion prediction model to calculate the hunger signal prediction time, the method further includes: Based on the gastric emptying time parameter, the amount of residual gastric contents is calculated according to the functional relationship of the amount of residual gastric contents decreasing with time; When the amount of gastric contents remaining is lower than a preset residual threshold, a first candidate signal for the end of the satiety state is generated. The frequency feature parameters are obtained. When the frequency feature parameters exceed a preset variation threshold and the respiratory rate fluctuation amplitude exceeds a preset fluctuation threshold within a continuous preset duration, a second satiety state end candidate signal is generated. When the first candidate signal for the end of the satiety state and the second candidate signal for the end of the satiety state are generated simultaneously, the satiety state is determined to be over, and the end of the satiety state is used as the trigger calibration signal for the multi-parameter fusion prediction model.

[0011] In one embodiment, it further includes: Obtain the infant's swallowing and exhalation signals; Calculate the proportion of exhalation within a preset time window after each swallow; When the exhalation ratio is lower than the preset exhalation ratio threshold, a third satiety state end candidate signal is generated. When the first candidate signal for the end of the satiety state, the second candidate signal for the end of the satiety state, and the third candidate signal for the end of the satiety state are generated simultaneously, the satiety state is determined to be over.

[0012] Secondly, this application discloses a system corresponding to the aforementioned intelligent prediction method for hunger and satiety cycles based on individual digestive rhythm modeling, the system comprising: The data acquisition module is used to acquire infant feeding-related data, respiratory-related physiological signal data, and age data, wherein the feeding-related data includes single feeding amount data and food type data; The gastric emptying time determination module is used to determine the gastric emptying time parameters based on the single feeding amount data and the food type data through a preset gastric emptying quantification model. The respiratory rhythm feature extraction module is used to extract respiratory rhythm feature parameters based on the respiratory-related physiological signal data. The respiratory rhythm feature parameters include frequency feature parameters that characterize the degree of fluctuation in respiratory frequency and depth feature parameters that characterize the degree of abrupt changes in respiratory depth. A digestive enzyme activity maturity determination module is used to determine digestive enzyme activity maturity parameters based on the age data. The hunger threshold delay determination module is used to determine the inherent digestion delay parameter of the food type based on the food type data, and to correct the inherent digestion delay parameter of the food type based on the digestive enzyme activity maturity parameter to obtain the hunger threshold delay parameter; A respiratory feature weight determination module is used to determine respiratory feature weight parameters based on the respiratory rhythm feature parameters and the age data in months. The fusion prediction module is used to input the gastric emptying time parameter, the hunger threshold delay parameter, the respiratory feature weight parameter and the respiratory rhythm feature parameter into a preset multi-parameter fusion prediction model to calculate the hunger signal prediction time; The feeding reminder output module is used to output a feeding reminder signal when the current time reaches the predicted hunger signal time.

[0013] Thirdly, this application discloses a computer device, including a memory and a processor, wherein the memory is communicatively connected to the processor, and the memory stores a computer program that can be executed by the processor. When the computer program is executed by the processor, the intelligent prediction method for hunger and satiety cycles based on individual digestive rhythm modeling, as described above, is implemented.

[0014] Fourthly, this application discloses a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the intelligent prediction method for hunger and satiety cycles based on individual digestive rhythm modeling as described above.

[0015] The aforementioned intelligent prediction method, system, computer equipment, and storage medium for hunger and satiety cycles based on individual digestive rhythm modeling acquire infant feeding-related data, respiratory physiological signal data, and age data. Based on a gastric emptying quantification model, it determines gastric emptying time parameters, achieving objective quantification of individual infant digestive rates. By extracting frequency feature parameters characterizing respiratory frequency fluctuations and depth feature parameters characterizing respiratory depth abrupt changes from respiratory physiological signal data, it achieves refined feature extraction of infant respiratory rhythms, enabling more accurate capture of respiratory pattern changes during hunger. Furthermore, by determining digestive enzyme activity maturity parameters based on age data according to a preset digestive enzyme activity maturity function, it achieves precise prediction of infant digestive rates. This study quantifies the changes in infant digestive capacity with age. By determining the inherent digestive delay parameter for each food type based on food type data and correcting this parameter based on the maturity parameter of digestive enzyme activity, it accurately characterizes the differences in digestion among different food types and dynamically adjusts this parameter as the infant's digestive capacity develops. Furthermore, by determining the respiratory feature weight parameter based on respiratory rhythm characteristic parameters and age data, it adaptively adjusts the contribution of respiratory rhythm characteristics in hunger prediction according to the degree of hunger and age. Finally, by inputting gastric emptying time parameters, hunger threshold delay parameters, respiratory feature weight parameters, and respiratory rhythm characteristic parameters into a multi-parameter fusion prediction model to calculate the hunger signal prediction time, it achieves collaborative modeling and fusion prediction of infant digestive physiological parameters and respiratory rhythm characteristics. Ultimately, it enables objective and quantitative intelligent prediction of the hunger-satiety cycle based on the individual infant's digestive rhythm, overcoming the subjectivity and limitations of existing technologies that rely on external behavior or single parameters. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram illustrating an application scenario of an intelligent prediction method for hunger and satiety cycles based on individual digestive rhythm modeling in one embodiment. Figure 2 This is a flowchart of an intelligent prediction method for hunger and satiety cycles based on individual digestive rhythm modeling in one embodiment. Figure 3 This is a block diagram of an intelligent prediction system for hunger and satiety cycles based on individual digestive rhythm modeling in one embodiment. Figure 4 This is a schematic diagram of the structure of a computer device in one embodiment. Detailed Implementation

[0018] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present application.

[0019] The intelligent prediction method for hunger and satiety cycles based on individual digestive rhythm modeling in this application can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process; this system can be integrated onto server 104 or located in the cloud or on other network servers. Terminal 102 acquires infant feeding-related data, respiratory physiological signal data, and age data, and sends the collected data to server 104 via a wireless or wired network. Server 104 determines gastric emptying time parameters, respiratory rhythm characteristic parameters, digestive enzyme activity maturity parameters, hunger threshold delay parameters, and respiratory characteristic weight parameters based on the data sent by terminal 102. These parameters are then input into a multi-parameter fusion prediction model to calculate the hunger signal prediction time. When the hunger signal prediction time is reached, terminal 102 outputs a feeding reminder signal. Terminal 102 can be, but is not limited to, an infant monitor, smart bottle, smart crib, smart bracelet, personal computer, laptop, smartphone, or tablet. Server 104 can be implemented using a standalone server or a server cluster consisting of multiple servers.

[0020] In practical applications, the principles of legality, legitimacy, necessity, and good faith should be followed. The purpose, method, and scope of data collection should be clearly and explicitly communicated to the infant's guardian, and their individual consent should be obtained before processing the data. Simultaneously, de-identification measures should be taken for the collected data, limiting the data processing results to non-identifiable information such as statistics and characteristic parameters. Data should only be collected and used to the minimum extent necessary to predict the infant's hunger-satiety cycle. Data storage and transmission should employ security measures such as encryption to prevent data leakage, tampering, or loss.

[0021] Firstly, existing infant hunger prediction technologies fail to incorporate multi-dimensional digestive physiological parameters such as gastric emptying dynamics, refined respiratory rhythm characteristics, and digestive enzyme activity maturity into a unified prediction framework. This leads to prediction results relying on subjective experience and lacking objective quantitative evidence. Therefore, in one embodiment, such as... Figure 2As shown, an intelligent prediction method for hunger and satiety cycles based on individual digestive rhythm modeling is provided for application in [the following context is missing from the original text]. Figure 1 Taking terminal 102 as an example, the method includes: S1: Acquire infant feeding-related data, respiratory-related physiological signal data, and age data, including feeding-related data such as single feeding amount and food type data; S2: Based on single feeding volume data and food type data, determine the gastric emptying time parameters through a preset gastric emptying quantification model; S3: Extract respiratory rhythm feature parameters from respiratory-related physiological signal data. The respiratory rhythm feature parameters include frequency feature parameters used to characterize the degree of fluctuation in respiratory rate and the number of depth feature parameters used to characterize the degree of abrupt changes in respiratory depth. S4: Determine the maturity parameters of digestive enzyme activity based on age data; S5: Determine the inherent digestion delay parameter of food type based on food type data, and correct the inherent digestion delay parameter of food type based on digestive enzyme activity maturity parameter to obtain the hunger threshold delay parameter; S6: Determine the respiratory characteristic weight parameters based on respiratory rhythm characteristic parameters and age data; S7: Input the gastric emptying time parameter, hunger threshold delay parameter, respiratory feature weight parameter and respiratory rhythm feature parameter into the preset multi-parameter fusion prediction model to calculate the hunger signal prediction time; S8: When the hunger signal prediction time is reached at the current time, output a feeding reminder signal.

[0022] Specifically, feeding-related data can be obtained in several ways. Single feeding volume data refers to the amount of milk or food an infant consumes in a single feeding. This can be manually recorded using a graduated bottle or automatically recorded using a pressure or flow sensor built into a smart bottle. Food type data refers to the category of food the infant consumes, which may include one or more combinations of breast milk, formula, and solid foods. This can be obtained through manual input by the user (e.g., selecting "breast milk," "formula," or "solid foods") or automatically identified by scanning the identification code on the food packaging.

[0023] Respiratory-related physiological signal data refers to physiological signal data related to an infant's respiratory activity. In specific data acquisition, respiratory waveform data can be obtained by directly acquiring chest and abdominal movement signals using a respiratory inductively coupled plasma plethysmography (RIP) sensor, or by acquiring pulse wave signals using a photoplethysmography (PPP) sensor and indirectly extracting respiratory rate and depth information using the baseline drift characteristics of the pulse wave signal. In the implementation using a PPP sensor, the raw signal acquired by the sensor is the PPP signal, and respiratory rate information is obtained by performing respiratory modulation extraction processing on this signal.

[0024] Monthly age data refers to the number of months that have passed since the baby's birth. It can be automatically calculated by the system after the user manually enters the baby's birth date, or it can be directly obtained by associating it with the baby's profile information already entered in the terminal device.

[0025] The specific method for determining the gastric emptying time parameter is as follows: First, determine the food type coefficient based on food type data. The food type coefficient is a dimensionless parameter used to quantify the degree of influence of different food types on gastric emptying time. Different food types have different gastric emptying rates; breast milk has a faster gastric emptying rate, formula milk has a moderate gastric emptying rate, and solid foods have a slower gastric emptying rate. Then, according to the gastric emptying time function, calculate the gastric emptying time parameter based on single feeding volume data and the food type coefficient.

[0026] The specific extraction method of respiratory rhythm feature parameters is as follows: First, the respiratory-related physiological signal data is preprocessed by noise reduction. Then, the respiratory cycle sequence and respiratory depth sequence are extracted from the preprocessed respiratory-related physiological signal data. The frequency feature parameters are calculated based on the respiratory cycle sequence, and the depth feature parameters are calculated based on the respiratory depth sequence.

[0027] The specific method for determining the digestive enzyme activity maturity parameter is as follows: According to the digestive enzyme activity maturity function relationship, the digestive enzyme activity maturity parameter is calculated based on the age data. This parameter is positively correlated with the age data and the growth rate decreases with the increase of age. Its function value is between 0 and 1, which is used to characterize the relative maturity of infant digestive enzyme activity with the increase of age.

[0028] The specific method for determining the respiratory feature weight parameters is as follows: First, according to the preset weighting relationship, the hunger index is calculated based on the frequency feature parameter and its corresponding first weighting coefficient, the depth feature parameter and its corresponding second weighting coefficient. This hunger index is used to characterize the infant's current hunger level. Then, according to the preset respiratory feature weighting function relationship, the respiratory feature weight parameters are calculated based on the hunger index and age data.

[0029] The multi-parameter fusion prediction model is a pre-established mathematical model used to fuse four input parameters—gastric emptying time, hunger threshold delay, respiratory feature weights, and respiratory rhythm features—to output the predicted time of hunger signals. There is a clear physiological correlation between the model's input and output data: the gastric emptying time parameter reflects the dynamic characteristics of gastric emptying and is the basic timescale of the hunger cycle; the hunger threshold delay parameter reflects the correction amount of food type and infant digestive enzyme activity maturity on the timing of hunger onset; the respiratory feature weights parameter reflects the contribution of respiratory rhythm features to the prediction under the current hunger level and age; and the respiratory rhythm features parameters (including frequency and depth features) reflect the fluctuation of the infant's current respiratory rate and the frequency of abrupt changes in respiratory depth, serving as a real-time physiological representation of the hunger state. These four input parameters jointly characterize the infant's hunger state from two dimensions: digestive system state (gastric emptying dynamics and enzyme maturity) and respiratory system response (respiratory rhythm features and their contribution weights), and output the predicted time of hunger signals after model fusion. As one feasible approach, the model can be implemented using a weighted summation method, where the hunger signal prediction time equals the sum of the products of the gastric emptying time parameter, the hunger threshold delay parameter, and the respiratory feature weight parameter and respiratory rhythm feature parameter (the sum of the frequency feature parameter and the depth feature parameter). Alternatively, the model can also be implemented using a linear regression model, a support vector regression model, or a neural network model. The specific model used can be determined based on the actual application scenario and accuracy requirements; this application does not limit this approach.

[0030] Specifically, the multi-parameter fusion prediction model, employing a weighted summation approach, outputs the following: , in, This parameter represents gastric emptying time, in minutes. This represents the hunger threshold delay parameter, in minutes. This represents the respiratory feature weighting parameter, which is a dimensionless parameter. This represents a frequency characteristic parameter, which is a dimensionless parameter. This represents depth feature parameters, which are dimensionless parameters. This represents the predicted hunger signal time, in minutes. In this formula, the product of the sum of the gastric emptying time parameter and the respiratory rhythm characteristic parameter and the respiratory characteristic weight parameter corresponds to the baseline digestion time and the respiratory state correction, respectively. The hunger threshold delay parameter further corrects the baseline time.

[0031] When a multiple linear regression model is used as a multi-parameter fusion prediction model, the model output is: , in, This is the intercept term (in minutes). The regression coefficients (dimensionless) corresponding to the gastric emptying time parameter. The regression coefficients (dimensionless) corresponding to the hunger threshold delay parameter are given. These are the dimensionless regression coefficients corresponding to the product of the respiratory feature weight parameters and the respiratory rhythm feature parameters. The values ​​of each regression coefficient are obtained by fitting a multiple linear regression to a historical dataset.

[0032] When using a support vector regression model as a multi-parameter fusion prediction model, the input feature vector of the model is: The output is the hunger signal prediction time. The kernel function used in this model is the radial basis function, and its expression is as follows: ,in For kernel parameters.

[0033] When using a feedforward neural network model as a multi-parameter fusion prediction model, the model structure is as follows: The input layer contains 3 neurons, each corresponding to one of the input features. , , The hidden layer contains one layer with eight neurons. The activation function is a rectified linear unit, expressed as follows: The output layer contains one neuron, which outputs the hunger signal prediction time. The activation function used is a linear activation function.

[0034] Furthermore, if linear regression, support vector regression, or neural network models are used, they must be pre-trained using historical datasets. The training dataset is constructed by collecting historical feeding monitoring data from several infants as training samples, with each training sample containing an input feature vector. and the corresponding tag value , This is the time interval, in minutes, between the actual observed completion of feeding and the actual appearance of the infant's hunger signal. The training dataset should include data from infants of different ages, different food types, and different feeding amounts. The input features in the training dataset are standardized using the following formula: ,in Let be the original value of the i-th feature. Let be the mean of the i-th feature on the training dataset. Let be the standard deviation of the i-th feature on the training dataset. These are the standardized feature values. The objective function for model training is to minimize the mean squared error loss function, expressed as follows: ,in The number of training samples. Let i be the true label value of the i-th sample. Let be the model prediction value for the i-th sample. For the neural network model, backpropagation and the Adam optimizer are used for training, with an initial learning rate set to 10. -3 The exponential decay rate for the first-order moment estimation was set to 0.9, the exponential decay rate for the second-order moment estimation was set to 0.999, the batch size was set to 32, and the maximum number of training epochs was set to 200. An early stopping mechanism was triggered when the validation set loss did not decrease in 20 consecutive training epochs. After training, the model performance was evaluated using an independent test dataset, with evaluation metrics including mean absolute error and root mean square error.

[0035] Feeding reminder signals can be output in the form of sound prompts, vibration prompts, push notifications displayed on the screen, or notifications sent to associated terminals.

[0036] Through the above steps, the gastric emptying time parameter reflects the speed at which food empties from the infant's stomach, providing a basic timescale for hunger prediction; the hunger threshold delay parameter reflects the combined influence of food type and digestive enzyme activity maturity on the timing of hunger on top of gastric emptying, enabling the prediction model to adapt to infants of different feeding methods and developmental stages; the respiratory feature weight parameter adaptively adjusts the contribution of respiratory rhythm features to prediction with varying degrees of hunger and age; the respiratory rhythm feature parameter characterizes changes in respiratory patterns under hunger from two refined dimensions: fluctuations in respiratory frequency and abrupt changes in respiratory depth. By inputting these parameters into a multi-parameter fusion prediction model, collaborative modeling and fusion prediction of infant digestive physiological parameters and respiratory rhythm features are achieved, thus enabling objective quantitative prediction of the timing of hunger signals based on individual infant digestive rhythms.

[0037] In existing technologies, the changes in infant digestive enzyme activity with age are only qualitatively described, lacking quantifiable mathematical models to calculate the maturity of digestive enzyme activity in infants of different ages. This prevents its inclusion in a quantitative framework for hunger prediction. Therefore, in one embodiment, a digestive enzyme activity maturity parameter is determined based on age data, including: According to the preset digestive enzyme activity maturity function relationship, the digestive enzyme activity maturity parameter is calculated based on the age data. The digestive enzyme activity maturity parameter is positively correlated with the age data and the growth rate decreases with the increase of age. The digestive enzyme activity maturity function relationship is an exponential decay function with a function value between 0 and 1, which is used to characterize the relative maturity of infant digestive enzyme activity with the increase of age.

[0038] As an feasible approach, the maturity function relationship of digestive enzyme activity can be expressed as: , Where m represents the age data in months; This parameter represents the maturity of digestive enzyme activity, ranging from 0 to 1, and is dimensionless. The constant 0.1 in this function is a rate coefficient fitted based on clinical research data on infant digestive enzyme activity development, with dimensions of "1 / month". This coefficient reflects the average maturation rate of digestive enzyme activity with increasing age. In practical applications, this rate coefficient can be adjusted based on statistical data from different infant groups.

[0039] When the baby is 0 months old , This indicates that the digestive enzyme activity of newborns is not yet fully developed; as they grow older, The value gradually decreases. A gradual increase in the value and its approximation to 1 indicates that the digestive enzyme activity is gradually approaching maturity. The growth rate of this function decreases with increasing age: it matures faster in the early stages (0 to 6 months) and gradually slows down in the later stages (after 6 months), which is consistent with the physiological laws of infant digestive system development.

[0040] By using the aforementioned exponential decay function relationship, the easily accessible data of infant age in months can be transformed into a quantitative parameter of digestive enzyme activity maturity. This provides a quantifiable input basis for subsequent calculation of the hunger threshold delay, enabling the hunger prediction model to reflect the dynamic changes in infant digestive capacity as the infant grows older.

[0041] When extracting respiratory rhythm feature parameters from respiratory-related physiological signal data, the raw respiratory-related physiological signal data usually contains various noise components such as power frequency interference, motion artifacts, and high-frequency noise. Directly extracted feature parameters have low accuracy and cannot accurately reflect the true changes in respiratory rhythm. Therefore, in one embodiment, extracting respiratory rhythm feature parameters from respiratory-related physiological signal data includes: Noise reduction preprocessing was performed on respiratory-related physiological signal data; Respiratory cycle sequence and respiratory depth sequence were extracted from the noise-reduced and preprocessed respiratory-related physiological signal data. Calculate frequency characteristic parameters based on respiratory cycle sequences; Depth feature parameters are calculated based on the respiratory depth sequence.

[0042] Specifically, a respiratory cycle sequence refers to a sequence of duration values ​​for each respiratory cycle arranged in chronological order, with each element corresponding to the duration of one respiratory cycle. The duration of each respiratory cycle can be obtained by detecting the time interval between two adjacent inspiratory initiations or two adjacent expiratory initiations in the respiratory signal. Taking the detection of the expiratory initiation as an example, the inflection point where the respiratory signal waveform changes from a rising edge to a falling edge is marked as the expiratory initiation point, and the time interval between two adjacent expiratory initiations is the duration of one respiratory cycle.

[0043] A respiratory depth sequence is a sequence of respiratory depth values ​​for each respiratory cycle arranged in chronological order, with each element corresponding to the respiratory depth value of one respiratory cycle. Respiratory depth can be obtained by detecting the amplitude difference between the peak value (highest point of the waveform) of the inspiratory phase and the trough value (lowest point of the waveform) of the expiratory phase in each respiratory cycle; this amplitude difference is the respiratory depth of that respiratory cycle.

[0044] Frequency characteristic parameters are used to characterize the degree of fluctuation in respiratory rate. One feasible approach is to use the ratio of the standard deviation to the mean of the respiratory cycle sequence, i.e., the respiratory rate coefficient of variation, which is a dimensionless parameter. The standard deviation of the respiratory cycle sequence reflects the dispersion of the respiratory cycle duration; a larger standard deviation indicates greater fluctuation in respiratory rate. The mean of the respiratory cycle sequence reflects the central tendency of the respiratory cycles. Dividing the standard deviation by the mean can eliminate differences in baseline respiratory rate levels between different infants, making the frequency characteristic parameters comparable.

[0045] Depth feature parameters are used to characterize the frequency of abrupt changes in respiratory depth. One feasible approach is to use the ratio of the number of respiratory depth abrupt changes to the total number of breaths, i.e., the respiratory depth abrupt change rate, which is dimensionless. The method for determining a respiratory depth abrupt change is as follows: calculate the change in respiratory depth between the current respiratory cycle and the previous respiratory cycle. If this change exceeds a preset abrupt change threshold, the respiratory event is marked as a respiratory depth abrupt change. The preset abrupt change threshold can be set to 30% of the respiratory depth of the previous respiratory cycle, and this threshold can be adjusted based on clinical experience.

[0046] The frequency and depth feature parameters extracted in the above manner can characterize the refined features of an infant's respiratory rhythm from two dimensions: the degree of fluctuation in respiratory rate and the frequency of abrupt changes in respiratory depth. This provides richer quantitative input for the subsequent calculation of the hunger index than a single respiratory rate value.

[0047] When establishing a quantitative relationship between respiratory rhythm features and infant hunger levels, relying solely on a single dimension of either the frequency or depth feature parameters is insufficient to comprehensively reflect changes in breathing patterns during hunger. Furthermore, the contribution of respiratory rhythm features to hunger prediction varies with the infant's age and hunger level. Therefore, in one embodiment, respiratory feature weighting parameters are determined based on respiratory rhythm feature parameters and age data, including: According to the preset weighting relationship, the hunger index is calculated based on the frequency feature parameters and their corresponding first weighting coefficients, the depth feature parameters and their corresponding second weighting coefficients. The hunger index is used to characterize the current hunger level of the infant. According to the preset respiratory feature weight function relationship, the respiratory feature weight parameters are calculated based on the hunger index and age data.

[0048] As an feasible approach, the hunger index can be calculated using the following formula: , in, The frequency characteristic parameter (i.e., the coefficient of variation of respiratory rate) is a dimensionless parameter. The depth characteristic parameter (i.e., respiratory depth mutation rate) is a dimensionless parameter. This represents the first weighting coefficient, which is a dimensionless parameter. This represents the second weighting coefficient, which is a dimensionless parameter. This represents the hunger index, which is a dimensionless parameter. The initial values ​​of the first and second weighting coefficients can be set empirically.

[0049] As an feasible approach, the respiratory feature weighting parameters can be calculated using the following formula: , in, The respiratory feature weighting parameter is a dimensionless parameter. In this formula, the respiratory feature weighting parameter consists of three parts: a baseline weight of 0.5 (dimensionless), representing the basic contribution of respiratory rhythm features in the absence of hunger index and age information; a dynamic adjustment term based on the hunger index; and a weighting parameter of 0.5. (Dimensionless), indicating that the contribution of respiratory rhythm features to prediction increases with increasing hunger index; age-based developmental adjustment term. (Dimensionless) indicates that as infants grow older, their breathing patterns become more mature and stable, and the contribution of respiratory rhythm characteristics to prediction increases accordingly. The constant coefficients (0.5, 0.3, 0.2, 0.05) in the above terms are empirical values ​​obtained by fitting experimental data, and their dimensions are the fitting coefficients of the corresponding terms, making the final calculation results dimensionless parameters. In practical applications, these constant coefficients can be adjusted according to the statistical characteristics of different infant groups.

[0050] The hunger index is calculated by the above weighted relationship, and the respiratory feature weight parameters are calculated by the respiratory feature weight function relationship. This realizes the adaptive adjustment of the respiratory feature weight parameters according to the infant's current hunger level and age, so that the contribution of respiratory rhythm features in hunger prediction can change dynamically with the changes in the infant's state.

[0051] After calculating the hunger index, an infant's hunger state is a gradual evolution from non-hunger to high hunger, with different breathing pattern characteristics at different stages. Using fixed weighting coefficients is insufficient to accurately characterize this dynamic evolution. Therefore, in one embodiment, after calculating the hunger index according to a preset weighting relationship and based on frequency feature parameters and their corresponding first weighting coefficients, and depth feature parameters and their corresponding second weighting coefficients, the method further includes: The hunger stage of an infant is determined by the range of values ​​in the hunger index. Hunger stages include early hunger stage, middle hunger stage, and late hunger stage. Based on the determined hunger stage, the weighting coefficients used to calculate the subsequent hunger index are updated according to a preset weighting coefficient adjustment strategy. The weighting coefficients corresponding to the early hunger stage are used to maintain the baseline sensitivity when the infant is not hungry. The weighting coefficients corresponding to the middle and late hunger stages are gradually increased to enhance the contribution of respiratory rhythm characteristics to the hunger index.

[0052] As an feasible approach, the determination of the starvation stage and the adjustment of the weighting coefficient can be carried out according to the following rules: When the hunger index is less than 0.3, it is determined to be the early hunger stage. The first weighting coefficient is set to 0.7 (dimensionless), and the second weighting coefficient is set to 0.3 (dimensionless). In the early hunger stage, the infant is not yet hungry or the degree of hunger is low, and the changes in respiratory rhythm characteristics are not obvious. Therefore, a lower weighting coefficient combination is used to keep the hunger index less sensitive to changes in respiratory rhythm characteristics.

[0053] When the hunger index is greater than or equal to 0.3 and less than 0.7, it is determined to be the intermediate hunger stage. The first weighting coefficient is set to 0.7 (dimensionless), and the second weighting coefficient is set to 0.3 (dimensionless). In the intermediate hunger stage, infants begin to show more obvious signs of hunger, and their respiratory rhythm characteristics begin to change. At this time, the same weighting coefficient combination as in the early hunger stage is used, but the sensitivity to changes in respiratory rhythm characteristics has begun to increase.

[0054] When the hunger index is greater than or equal to 0.7, it is determined to be in the late stage of hunger. The first weighting coefficient is set to 1.0 (dimensionless), and the second weighting coefficient is set to 0.5 (dimensionless). In the late stage of hunger, the infant is already in a high degree of hunger, and the respiratory rhythm characteristics change significantly. At this time, a higher weighting coefficient combination is used to make the hunger index more sensitive to changes in respiratory rhythm characteristics, so as to accurately reflect the intensification of hunger.

[0055] The above-mentioned threshold values ​​(0.3 and 0.7) and weighting coefficient values ​​(0.7, 0.3, 1.0, 0.5) are example values ​​set based on experimental data. In practical applications, they can be adjusted according to the statistical characteristics of different infant groups.

[0056] Through the aforementioned stage determination and weighted coefficient dynamic adjustment mechanism, the gradual evolution of infant hunger from early to late stages can be more precisely characterized, and the contribution of respiratory rhythm characteristics to the hunger index calculation can be adaptively adjusted according to the characteristics of different hunger stages, thereby improving the accuracy of infant hunger assessment.

[0057] Before inputting the various parameters into the multi-parameter fusion prediction model to calculate the hunger signal prediction time, it is necessary to determine the start time of the hunger cycle, i.e., the time when the satiety state ends. Existing technologies lack objective quantitative markers for the end time of the satiety state, leading to inaccurate start points for the prediction cycle. Therefore, in one embodiment, before inputting the gastric emptying time parameter, hunger threshold delay parameter, respiratory feature weight parameter, and respiratory rhythm feature parameter into a preset multi-parameter fusion prediction model to calculate the hunger signal prediction time, the following step is also included: Based on the gastric emptying time parameter, the amount of residual gastric contents is calculated according to the functional relationship of the amount of residual gastric contents decreasing over time; When the amount of stomach contents remaining is lower than a preset threshold, a candidate signal for the end of the first satiety state is generated. Acquire frequency characteristic parameters. When the frequency characteristic parameters exceed a preset variation threshold and the respiratory rate fluctuation amplitude exceeds a preset fluctuation threshold within a preset duration, generate a second candidate signal for the end of the satiety state. When the first candidate signal for the end of the satiety state and the second candidate signal for the end of the satiety state are generated simultaneously, the satiety state is determined to be over, and the end of the satiety state is used as the trigger calibration signal for the multi-parameter fusion prediction model.

[0058] Specifically, gastric contents residual amount refers to the proportion of unemptied food in the stomach after a certain period of time following feeding, expressed as a percentage. The functional relationship between gastric contents residual amount and time reflects the dynamic process of gastric emptying. As an implementable method, gastric contents residual amount can be calculated using the following formula: , in, This indicates the elapsed time since the feeding was completed, in minutes. This represents the amount of gastric contents remaining after time tt, a dimensionless parameter ranging from 0 to 1. When expressed as a percentage, it is multiplied by 100%. In this formula, e is the base of the natural logarithm (approximately 2.71828). The ratio of elapsed time to gastric emptying time is a dimensionless parameter. This formula reflects the exponential decay of residual gastric contents over time, with the decay rate determined by the gastric emptying time parameter: the smaller the gastric emptying time parameter, the faster the decay; the larger the gastric emptying time parameter, the slower the decay.

[0059] The preset residual amount threshold can be set based on physiological data. For example, it can be set to 5%, meaning that when the residual amount of gastric contents is less than 5% of the initial feeding amount, the stomach is considered to be basically empty. This threshold represents the percentage of residual gastric contents and can be adjusted in practical applications.

[0060] The preset variation threshold, preset duration, and preset fluctuation threshold can be set based on experimental data. As an feasible approach, the preset variation threshold can be set to 0.3 (dimensionless), indicating that the trigger condition occurs when the respiratory rate characteristic parameter exceeds 0.3; the preset duration can be set to 3 minutes, indicating continuous monitoring within a 3-minute time window; and the preset fluctuation threshold can be set to 20%, indicating that the trigger condition occurs when the respiratory rate fluctuation exceeds 20% of the baseline level. Specifically, when the respiratory rate characteristic parameter is greater than 0.3 and the respiratory rate fluctuation exceeds 20% within 3 consecutive minutes, a candidate signal for the end of the second satiety state is generated.

[0061] The first and second candidate signals for the end of the satiety state are logical marker signals, indicating that the conditions "the amount of gastric contents remaining is below a preset threshold" and "the degree of respiratory rate fluctuation meets a preset condition" have been met, respectively. When both candidate signals are generated simultaneously, that is, when both conditions are met, the satiety state is determined to have ended.

[0062] The above-mentioned dual-condition joint judgment mechanism determines the end of the satiety state only when the amount of gastric contents remaining is lower than the preset residual amount threshold and the degree of respiratory rate fluctuation meets the preset conditions. This achieves accurate marking of the end of the satiety state and avoids misjudgment that may occur if only a single condition is relied upon. It provides a reliable starting reference point for the subsequent calculation of the hunger signal prediction time.

[0063] While determining the end of satiety based on the amount of gastric contents remaining and the degree of fluctuation in respiratory rate, swallowing-exhalation coordination is also an important physiological indicator reflecting an infant's satiety status. However, existing technologies do not incorporate this indicator into the satiety determination system. Therefore, in one embodiment, the step of determining the end of satiety also involves acquiring the infant's swallowing and exhalation signals; calculating the exhalation ratio within a preset time window after each swallow; generating a third candidate signal for the end of satiety when the exhalation ratio is lower than a preset exhalation ratio threshold; and determining the end of satiety when the first, second, and third candidate signals for the end of satiety are generated simultaneously.

[0064] Specifically, swallowing signals refer to signals that characterize the occurrence of an infant's swallowing action, which can be detected by collecting electromyographic, acoustic, or pressure signals from the infant's neck or jaw area. For example, a microphone placed under the infant's jaw can be used to collect swallowing sound signals, and a swallowing event can be marked when an acoustic signal within a specific frequency range is detected; or a surface electromyography (SEMG) sensor can be used to collect activity signals of the jaw muscles, and a swallowing event can be marked when the amplitude of the SEMG signal exceeds a preset threshold.

[0065] Expiratory signals are signals that characterize an infant's exhalation action and can be collected using respiratory sensors or airflow sensors. For example, a thermal sensor placed near the infant's mouth and nose can be used to detect temperature changes in the expiratory airflow, or a piezoelectric sensor can be used to detect pressure changes generated by the expiratory airflow.

[0066] The proportion of exhalation within a preset time window after each swallow refers to the percentage of exhalation time within the preset time window, calculated from the moment of each swallowing event. The preset time window can be set based on physiological data, for example, it can be set to 2 seconds. The specific calculation method is as follows: starting from the moment of the swallowing event... ,statistics to The length of time the expiratory signal appears within a preset time window (seconds). The exhalation ratio is The percentage of exhalation is expressed as a percentage in seconds. This exhalation percentage reflects the recovery of the breathing pattern after swallowing; healthy infants can quickly return to a normal expiratory rhythm after swallowing.

[0067] The preset exhalation ratio threshold can be set based on the physiological data of a healthy infant, for example, it can be set to 40%. When the exhalation ratio after swallowing is lower than 40%, it indicates that the infant's swallowing-exhalation coordination is reduced, which usually occurs when the infant is full.

[0068] By introducing swallowing-exhalation coordination as a third criterion, the satiety state was further verified from the dimension of breathing-swallowing coordination, which improved the accuracy and reliability of the judgment of the end of satiety state.

[0069] Based on the introduction of swallowing-expiration coordination as a condition for determining satiety, the switching of swallowing frequency patterns from continuous sucking phases to intermittent phases is also an important behavioral marker reflecting an infant's satiety status. However, existing technologies do not incorporate this behavioral marker into the satiety determination system. Therefore, in one embodiment, in the step of determining the end of satiety, the swallowing frequency is also detected based on the swallowing signal; when a switch from a first frequency pattern representing the continuous sucking phase to a second frequency pattern representing the intermittent phase is detected, a fourth satiety end candidate signal is generated; when the first, second, third, and fourth satiety end candidate signals are generated simultaneously, the satiety status is determined to have ended.

[0070] Swallowing frequency refers to the number of swallows per unit of time, measured in "swallowing times per minute". It can be obtained by detecting and counting swallowing signals. The specific calculation method is as follows: taking the current moment as the endpoint, take a preset time window (e.g., 30 seconds) backward, count the number of swallowing events within this time window, and then convert it into the number of swallowing events per minute.

[0071] During infant feeding, swallowing frequency exhibits a typical pattern of change: In the initial feeding phase, the infant is in a continuous sucking phase, with a high and regular swallowing frequency, for example, approximately 50 to 70 times per minute; as feeding progresses, the infant gradually enters an intermittent phase, with a lower and more irregular swallowing frequency, for example, approximately 20 to 40 times per minute. The first frequency pattern corresponds to the continuous sucking phase, characterized by a high swallowing frequency (e.g., greater than 45 times per minute); the second frequency pattern corresponds to the intermittent phase, characterized by a lower swallowing frequency (e.g., less than 45 times per minute). When the swallowing frequency switches from the first frequency pattern to the second frequency pattern, it indicates that the infant is nearing satiety. The aforementioned frequency threshold (45 times per minute) is an example value set based on clinical observation data and can be adjusted according to individual differences among infants in practical applications.

[0072] By introducing swallowing frequency pattern switching as the fourth criterion, the satiety state was further verified from the dimension of changes in feeding behavior patterns, which further improved the accuracy of the judgment of the end of satiety state.

[0073] When determining gastric emptying time parameters using a gastric emptying quantification model, the gastric emptying rates of different food types vary significantly. Existing technologies lack mathematical models that simultaneously incorporate feeding volume and food type into the gastric emptying time quantification calculation. Therefore, in one embodiment, when determining the gastric emptying time parameters using a preset gastric emptying quantification model based on single feeding volume data and food type data, firstly, a food type coefficient corresponding to each food type is determined based on the food type data, with different food type coefficient values ​​for different food types; then, according to a preset gastric emptying time function relationship, the gastric emptying time parameter is calculated based on the single feeding volume data and the food type coefficient. This gastric emptying time parameter is positively correlated with the single feeding volume data and changes monotonically with the change in the food type coefficient.

[0074] Specifically, the food type coefficient is a dimensionless parameter used to quantify the impact of different food types on gastric emptying time. As an feasible approach, the value of the food type coefficient can be determined based on the food type: 0.8 for breast milk, 1.2 for formula, and 1.5 for solid foods. These values ​​reflect the differences in gastric emptying rates among different food types: breast milk has the fastest gastric emptying rate (lowest coefficient), followed by formula, and solid foods have the slowest (highest coefficient). These values ​​are empirical values ​​derived from fitting clinical gastric emptying study data and can be adjusted in practical applications based on the statistical characteristics of different infant populations.

[0075] The gastric emptying time function reflects the quantitative relationship between gastric emptying time and feeding volume and food type coefficient. As an implementable method, the gastric emptying time parameter can be calculated using the following formula: , in, This indicates the amount of food given per feeding, in milliliters. The food type coefficient is a dimensionless parameter ranging from 0.8 to 1.5. In this formula, the constant 0.8 is in units of minutes per milliliter, representing the basic contribution of each milliliter of feeding volume to gastric emptying time; the constant 0.3, also in units of minutes per milliliter, represents the additional increase in gastric emptying time per unit of feeding volume for every unit increase in the food type coefficient. The values ​​of these constants can be obtained through linear regression fitting based on existing techniques using clinical research data on infant gastric emptying.

[0076] For example, when the amount of food given at one feeding Taking milliliters as an example, when the food type is breast milk (k=0.8), the corresponding gastric emptying time is approximately 124.8 minutes; when the food type is formula milk (k=1.2), the corresponding gastric emptying time is approximately 139.2 minutes. These calculations show that, under the same feeding volume, the gastric emptying time for formula milk is longer than that for breast milk, consistent with common physiological knowledge.

[0077] The above linear approximation formula enables the quantitative calculation of gastric emptying time based on single feeding volume and food type data. The formula is concise and easy to calculate, providing objective gastric emptying dynamics parameters for subsequent multi-parameter fusion prediction. In practical applications, the constant coefficients (0.8 and 0.3) in the formula can be refitted and adjusted based on measured gastric emptying data from different infant groups; this embodiment does not impose such limitations.

[0078] In the denoising preprocessing of respiratory-related physiological signal data, the noise components mixed in the original signal are complex and diverse, and a single denoising method is insufficient to effectively remove all types of noise. Therefore, denoising preprocessing is performed on respiratory-related physiological signal data, including: Wavelet transform noise reduction was performed on the respiratory-related physiological signal data to obtain the noise-reduced respiratory-related physiological signal data. Respiratory cycle sequence and respiratory depth sequence were extracted from the noise-reduced respiratory-related physiological signal data; Calculate frequency characteristic parameters based on respiratory cycle sequences, and calculate depth characteristic parameters based on respiratory depth sequences; The calculated frequency and depth characteristic parameters are then processed using Kalman filtering.

[0079] Specifically, wavelet transform denoising is a signal denoising method based on wavelet analysis. Its basic principle is to decompose the signal into components of different scales, threshold or zero out the noise-dominant components, and then reconstruct the signal. As an implementable approach, wavelet transform denoising can use the db4 wavelet from the Daubechies wavelet family as the basis function, performing a three-level decomposition denoising. The db4 wavelet has good regularity and tight support, making it suitable for time-frequency analysis of respiratory signals. Three levels of decomposition are chosen because the main energy of the respiratory signal is concentrated in the low-frequency components, and three levels of decomposition are sufficient to separate the signal from the noise.

[0080] After extracting the respiratory cycle sequence and respiratory depth sequence from the denoised respiratory-related physiological signal data, frequency feature parameters and depth feature parameters are calculated, respectively. The specific calculation methods for the frequency feature parameters and depth feature parameters are the same as those described above.

[0081] The calculated frequency and depth characteristic parameters are processed by Kalman filtering. First, the frequency and depth characteristic parameters are input as observations into two independent one-dimensional Kalman filters. The input to each one-dimensional Kalman filter is the calculated time series of the frequency (or depth) characteristic parameters, and the output is the filtered frequency (or depth) characteristic parameters. The initial parameters of the Kalman filter include the initial state estimate, initial error covariance, process noise variance, and measurement noise variance. These parameters can be set according to the signal characteristics and noise level. The specific calculation process of the Kalman filter is existing technology, and those skilled in the art can implement it based on the above description.

[0082] By cascading the application of wavelet transform denoising and Kalman filtering, wavelet transform is first used to remove sudden noise and power frequency interference from respiratory-related physiological signals. Then, respiratory cycle sequences and respiratory depth sequences are extracted from the denoised signals, and frequency and depth feature parameters are calculated. Finally, Kalman filtering is used to smooth the extracted feature parameters, thereby effectively extracting stable and reliable respiratory rhythm feature parameters from the original respiratory-related physiological signal data, providing high-quality input data for subsequent hunger index calculation.

[0083] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps.

[0084] Secondly, based on the same inventive concept, this application also provides a system for implementing the intelligent prediction method for hunger and satiety cycles based on individual digestive rhythm modeling as described above. The solution provided by this system is similar to the implementation scheme described in the above method; therefore, the specific limitations in one or more system embodiments provided below can be found in the limitations of the intelligent prediction method for hunger and satiety cycles based on individual digestive rhythm modeling described above, and will not be repeated here.

[0085] In one embodiment, such as Figure 3 As shown, an intelligent prediction system for hunger and satiety cycles based on individual digestive rhythm modeling is provided. The system includes: The data acquisition module is used to acquire infant feeding-related data, respiratory-related physiological signal data, and age data, including feeding-related data such as single feeding amount data and food type data; The gastric emptying time determination module is used to determine the gastric emptying time parameters based on single feeding volume data and food type data through a preset gastric emptying quantification model. The respiratory rhythm feature extraction module is used to extract respiratory rhythm feature parameters based on respiratory-related physiological signal data. The respiratory rhythm feature parameters include frequency feature parameters that characterize the degree of fluctuation in respiratory rate and depth feature parameters that characterize the degree of abrupt changes in respiratory depth. The digestive enzyme activity maturity determination module is used to determine the digestive enzyme activity maturity parameters based on age data. The hunger threshold delay determination module is used to determine the inherent digestion delay parameter of food type based on food type data, and to correct the inherent digestion delay parameter of food type based on digestive enzyme activity maturity parameter to obtain the hunger threshold delay parameter; The respiratory feature weight determination module is used to determine respiratory feature weight parameters based on respiratory rhythm feature parameters and age data. The fusion prediction module is used to input gastric emptying time parameters, hunger threshold delay parameters, respiratory feature weight parameters, and respiratory rhythm feature parameters into a preset multi-parameter fusion prediction model to calculate the hunger signal prediction time. The feeding reminder output module is used to output a feeding reminder signal when the current time reaches the predicted hunger signal time.

[0086] The aforementioned modules can be integrated into the same data processing device or distributed and work collaboratively through network communication. The specific hardware implementation of each module can adopt any of the existing technologies such as general-purpose processors, digital signal processors, application-specific integrated circuits, and field-programmable gate arrays; this application embodiment does not limit this.

[0087] The modules in the aforementioned intelligent prediction system for hunger and satiety cycles based on individual digestive rhythm modeling can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, allowing the processor to call and execute the corresponding operations. Furthermore, it should be noted that the specific limitations regarding the implementation of the intelligent prediction method for hunger and satiety cycles based on individual digestive rhythm modeling in this system can be found in the limitations described above for the intelligent prediction method for hunger and satiety cycles based on individual digestive rhythm modeling, and will not be repeated here.

[0088] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 4As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database stores data to be processed by the server. The network interface communicates with external terminals via a network connection. When executed by the processor, the computer program implements an intelligent prediction method for hunger and satiety cycles based on individual digestive rhythm modeling.

[0089] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0090] Thirdly, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of any of the methods described in the above method embodiments.

[0091] Fourthly, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the steps of any of the methods described in the above method embodiments.

[0092] Those skilled in the art will understand that all or part of the processes of the above method embodiments can be implemented by computer program instructions and related hardware. This computer program can be stored in a computer-readable storage medium, and its execution can include the processes of the above method embodiments. The storage medium involved in each embodiment includes at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory, magnetic tape, floppy disk, flash memory, or optical memory, etc.; volatile memory may include random access memory or external cache memory, etc. The database involved in each embodiment may include at least one of relational database and non-relational database. The processor involved in each embodiment may be a general-purpose processor, central processing unit, graphics processing unit, or digital signal processor, etc.

[0093] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0094] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for intelligent prediction of hunger and satiety cycles based on individual digestive rhythm modeling, characterized in that, The method includes: Acquire infant feeding-related data, respiratory-related physiological signal data, and age data, wherein the feeding-related data includes single feeding amount data and food type data; Based on the single feeding volume data and the food type data, the gastric emptying time parameter is determined by a preset gastric emptying quantification model; Respiratory rhythm feature parameters are extracted from the respiratory-related physiological signal data. The respiratory rhythm feature parameters include frequency feature parameters that characterize the degree of fluctuation in respiratory rate and depth feature parameters that characterize the degree of abrupt changes in respiratory depth. The digestive enzyme activity maturity parameters were determined based on the age data. Based on the food type data, the inherent digestion delay parameter of the food type is determined, and the inherent digestion delay parameter of the food type is corrected based on the digestive enzyme activity maturity parameter to obtain the hunger threshold delay parameter; The respiratory feature weight parameters are determined based on the respiratory rhythm characteristic parameters and the age data in months. The gastric emptying time parameter, the hunger threshold delay parameter, the respiratory feature weight parameter, and the respiratory rhythm feature parameter are input into a preset multi-parameter fusion prediction model to calculate the hunger signal prediction time. When the current time reaches the predicted hunger signal time, a feeding reminder signal is output.

2. The method according to claim 1, characterized in that, The determination of digestive enzyme activity maturity parameters based on the age data includes: According to the preset digestive enzyme activity maturity function relationship, the digestive enzyme activity maturity parameter is calculated based on the age data. The digestive enzyme activity maturity parameter is positively correlated with the age data and the growth rate decreases with increasing age. The digestive enzyme activity maturity function relationship is an exponential decay function with a function value between 0 and 1, which is used to characterize the relative maturity of infant digestive enzyme activity with increasing age.

3. The method according to claim 1, characterized in that, The step of extracting respiratory rhythm feature parameters based on the respiratory-related physiological signal data includes: The respiratory-related physiological signal data are subjected to noise reduction preprocessing. Respiratory cycle sequence and respiratory depth sequence were extracted from the noise-reduced and preprocessed respiratory-related physiological signal data. The frequency characteristic parameters are calculated based on the respiratory cycle sequence; The depth feature parameters are calculated based on the respiratory depth sequence.

4. The method according to claim 1, characterized in that, The step of determining respiratory feature weight parameters based on the respiratory rhythm feature parameters and the age data includes: According to a preset weighting relationship, the hunger index is calculated based on the frequency feature parameters and their corresponding first weighting coefficients, the depth feature parameters and their corresponding second weighting coefficients, and the hunger index is used to characterize the infant's current hunger level. The respiratory feature weight parameters are calculated based on the hunger index and the age data according to the preset respiratory feature weight function relationship.

5. The method according to claim 4, characterized in that, After the step of calculating the hunger index according to a preset weighting relationship, based on the frequency feature parameters and their corresponding first weighting coefficients, and the depth feature parameters and their corresponding second weighting coefficients, the method further includes: The hunger stage of an infant is determined based on the numerical range of the hunger index, which includes early hunger stage, middle hunger stage and late hunger stage; Based on the determined hunger stage, the weighting coefficients used to calculate the subsequent hunger index are updated according to a preset weighting coefficient adjustment strategy. The weighting coefficients corresponding to the early hunger stage are used to maintain baseline sensitivity when the infant is in a non-hunger state. The weighting coefficients corresponding to the middle hunger stage and the late hunger stage are gradually increased to enhance the contribution of respiratory rhythm characteristics to the hunger index.

6. The method according to claim 1, characterized in that, Before the step of inputting the gastric emptying time parameter, the hunger threshold delay parameter, the respiratory feature weight parameter, and the respiratory rhythm feature parameter into a preset multi-parameter fusion prediction model to calculate the hunger signal prediction time, the method further includes: Based on the gastric emptying time parameter, the amount of residual gastric contents is calculated according to the functional relationship of the amount of residual gastric contents decreasing with time; When the amount of gastric contents remaining is lower than a preset residual threshold, a first candidate signal for the end of the satiety state is generated. The frequency feature parameters are obtained. When the frequency feature parameters exceed a preset variation threshold and the respiratory rate fluctuation amplitude exceeds a preset fluctuation threshold within a continuous preset duration, a second satiety state end candidate signal is generated. When the first candidate signal for the end of the satiety state and the second candidate signal for the end of the satiety state are generated simultaneously, the satiety state is determined to be over, and the end of the satiety state is used as the trigger calibration signal for the multi-parameter fusion prediction model.

7. The method according to claim 6, characterized in that, Also includes: Obtain the baby's swallowing and exhalation signals; Calculate the proportion of exhalation within a preset time window after each swallow; When the exhalation ratio is lower than the preset exhalation ratio threshold, a third satiety state end candidate signal is generated. When the first candidate signal for the end of the satiety state, the second candidate signal for the end of the satiety state, and the third candidate signal for the end of the satiety state are generated simultaneously, the satiety state is determined to be over.

8. An intelligent prediction system for hunger and satiety cycles based on individual digestive rhythm modeling, characterized in that, The system includes: The data acquisition module is used to acquire infant feeding-related data, respiratory-related physiological signal data, and age data, wherein the feeding-related data includes single feeding amount data and food type data; The gastric emptying time determination module is used to determine the gastric emptying time parameters based on the single feeding amount data and the food type data through a preset gastric emptying quantification model. The respiratory rhythm feature extraction module is used to extract respiratory rhythm feature parameters based on the respiratory-related physiological signal data. The respiratory rhythm feature parameters include frequency feature parameters that characterize the degree of fluctuation in respiratory rate and depth feature parameters that characterize the degree of abrupt changes in respiratory depth. A digestive enzyme activity maturity determination module is used to determine digestive enzyme activity maturity parameters based on the age data. The hunger threshold delay determination module is used to determine the inherent digestion delay parameter of the food type based on the food type data, and to correct the inherent digestion delay parameter of the food type based on the digestive enzyme activity maturity parameter to obtain the hunger threshold delay parameter; A respiratory feature weight determination module is used to determine respiratory feature weight parameters based on the respiratory rhythm feature parameters and the age data in months. The fusion prediction module is used to input the gastric emptying time parameter, the hunger threshold delay parameter, the respiratory feature weight parameter and the respiratory rhythm feature parameter into a preset multi-parameter fusion prediction model to calculate the hunger signal prediction time; The feeding reminder output module is used to output a feeding reminder signal when the current time reaches the predicted hunger signal time.

9. A computer device comprising a memory and a processor, the memory being communicatively connected to the processor, and the memory storing a computer program executable by the processor, characterized in that, When the computer program is executed by the processor, it implements the intelligent prediction method for hunger and satiety cycles based on individual digestive rhythm modeling as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the intelligent prediction method for hunger and satiety cycles based on individual digestive rhythm modeling as described in any one of claims 1 to 7.

Images