Health recommendation result generation method based on artificial intelligence

Abstract

The invention discloses a health recommendation result generation method based on artificial intelligence, and relates to the technical field of neural network and intelligent health recommendation, and the method comprises the steps: collecting heart rate data and height data of a user, calculating a heart rate reserve, and recognizing a vertical rising activity segment; calculating a vertically related physiological response amount based on the heart rate change and the height increment in the rising process, and carrying out statistical analysis on the response amount to obtain a quantitative parameter used for representing the user rising load sensitivity; calculating the target vertical height and the metabolic energy demand, and forming input data reflecting the individual physiological ability and the energy consumption level of the user; and constructing the input data into a feature vector and inputting the feature vector into a pre-trained feedforward neural network model to generate a health recommendation result. According to the invention, personalized health suggestions can be generated according to the vertical activity expression, and the accuracy and suitability of health management are improved.

Description

Technical Field

[0001] This invention relates to the field of neural networks and intelligent health recommendation technology, and in particular to a method for generating health recommendation results based on artificial intelligence. Background Technology

[0002] Climbing stairs and ascending heights are common activities involving significant vertical displacement, often accompanied by continuous changes in cardiac load, energy expenditure, and subsequent adjustments in metabolic demands. With the increasing prevalence of wearable devices in daily health management, users can obtain physiological responses during vertical movement through combined recording of altitude and heart rate data. However, current health management methods based on altitude and heart rate changes remain at the stage of single-data observation or empirical estimation, making it difficult to link the actual mechanical load and physiological response amplitude of vertical activities with individual metabolic needs. Furthermore, differences in weight, age, and body regulation capabilities among users result in significant individual variations in the physiological load generated by the same vertical activity, hindering the provision of continuous, quantifiable, and targeted health recommendations.

[0003] Existing technologies, when processing activity data involving vertical displacement, typically rely solely on qualitative judgments based on simple heart rate changes. They fail to establish a quantitative correlation between changes in mechanical work and cardiac load during vertical movement, making it difficult to accurately assess the actual physiological stress experienced by users during ascending activities. Consequently, health recommendations generated by existing methods often lack effectiveness for vertical movement scenarios. They cannot adjust exercise goals according to the user's physiological response level, nor can they estimate the energy required to complete the activity. This leads to a mismatch between recommended results and the user's actual metabolic level, resulting in some users experiencing insufficient or excessive energy intake when following the recommendations. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies that make it difficult to accurately assess a user's physiological load and energy needs based on changes in altitude and heart rate response during vertical movement, and to propose a health recommendation result generation method based on artificial intelligence.

[0005] To address the problems existing in the prior art, the present invention adopts the following technical solution: AI-based methods for generating health recommendation results include: S1. Collect the user's heart rate and altitude data, and calculate the heart rate reserve based on the user's heart rate data and age; S2. Identify vertical ascent activity segments based on the user's height data and calculate the corresponding height increment; S3. Calculate the vertical sensitivity sequence based on heart rate data and altitude increments within the vertical ascent activity segment; S4. Perform statistical analysis on the vertical sensitivity sequence to generate the heart rate amplification offset number; S5. Obtain the user's weight, and calculate the target vertical height and metabolic energy requirement based on heart rate reserve, user's weight and heart rate amplification offset. S6. Input the heart rate reserve, heart rate amplification offset, target vertical height, and metabolic energy requirement into the pre-trained neural network model to generate health recommendation results.

[0006] Preferably, the user's heart rate and altitude data are collected, and heart rate reserve is calculated based on the user's heart rate data and age, including: Collect users' height and heart rate data; Determine the time period during which the user is in a resting state, and use the average heart rate data during that time period as the user's resting heart rate; Obtain the user's age and determine the user's maximum heart rate based on the user's age; The user's heart rate reserve is obtained by calculating the difference between the user's maximum heart rate and resting heart rate.

[0007] Preferably, identifying vertically ascending activity segments based on the user's height data and calculating the corresponding height increment includes: The height data is divided into segments based on a preset time window length to obtain height change segments; Calculate the time mean and height mean of the height change segments; Based on the time mean and height mean of the height change segment, a linear fit is performed on the time data and height data within the height change segment to obtain the height change slope; The duration of the altitude change segment is obtained by the difference between the end time and the start time within the segment. The height increment of a height change segment is obtained by multiplying the time length of the height change segment by the slope of the height change segment. The height increment of each height change segment is judged. When the height increment is greater than zero, the corresponding height change segment is determined as a vertical ascent activity segment.

[0008] Preferably, the vertical sensitivity sequence is calculated based on heart rate data and altitude increments within a vertically ascending activity segment, including: Calculate the mean of heart rate data within the vertical ascending activity segment, and obtain the heart rate increment of the vertical ascending activity segment based on the difference between the mean and the resting heart rate; The work done by gravity is obtained by multiplying the height increment of the vertical ascent segment by the acceleration due to gravity. When the vertical gravitational work is greater than zero, the ratio of the heart rate increment to the vertical gravitational work during the vertical ascent segment is used as the vertical sensitivity value. The vertical sensitivity values ​​of each vertically ascending activity segment are summarized to form a vertical sensitivity sequence.

[0009] Preferably, statistical analysis is performed on the vertical sensitivity sequence to generate a heart rate amplification offset, including: Arrange the vertical sensitivity sequence in ascending order to obtain an ordered sequence; Extract a preset number of element values ​​starting from the end of the ordered sequence, calculate the average value of the extracted element values, and use the average value as the heart rate amplification offset.

[0010] Preferably, the target vertical height and metabolic energy requirement are calculated based on heart rate reserve, the user's weight, and heart rate amplification offset, including: The vertical gravitational work target is calculated based on the heart rate amplification offset and heart rate reserve. The formula for calculating the vertical gravitational work target is as follows:

[0011] In the formula, For the vertical work done by gravity. For heart rate reserve, This represents the heart rate amplification offset. The vertical work target ratio coefficient is a pre-set value. The vertical height of the target is obtained by the ratio of the work done by the target under vertical gravity to the acceleration due to gravity. The user's weight is obtained, and the metabolic energy requirement is calculated by multiplying the user's weight by the vertical gravitational work target and the preset energy conversion ratio coefficient. Preferably, heart rate reserve, heart rate amplification offset, target vertical height, and metabolic energy requirement are input into a pre-trained neural network model to generate health recommendation results, including: The heart rate reserve, heart rate amplification offset, target vertical height, and metabolic energy requirement are combined to construct a feature vector; The feature vector is input into a pre-trained feedforward neural network model to obtain the nutrient supply amount, which includes: carbohydrate supply amount, protein supply amount, lipid supply amount and electrolyte supply amount; Generate health recommendations that include target vertical height and nutrient intake.

[0012] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention calculates heart rate reserve sequentially, identifies vertical ascent activity segments, quantifies the response relationship between vertical movement and heart rate changes, generates heart rate amplification offset, and estimates target vertical height and metabolic energy requirements. The quantification results are then input into a pre-trained feedforward neural network model to obtain health recommendations that include target vertical height and nutritional intake. This allows health recommendations to be individualized based on the user's actual vertical movement performance and physiological response, accurately matching the physiological load and energy requirements generated by the user during vertical activities, thus improving the relevance and reliability of health recommendations.

[0013] 2. This invention, by windowing height data, calculating the slope of height change, and identifying vertical ascent activity segments based on height increments, can extract the effective time periods of actual vertical ascent from continuous height changes, avoiding misjudgments caused by single-point height noise or multi-directional movement, ensuring that vertical load analysis is based solely on the actual ascent process. Simultaneously, it introduces the ratio between heart rate increment and vertical gravitational work, forming a continuous numerical sequence that reflects the user's sensitivity to changes in cardiac activity, and further generates a heart rate amplification offset number, enabling the quantitative extraction of high-load response segments in the user's vertical sensitivity. This allows for precise characterization of physiological load during vertical movement, accurately linking heart rate changes with vertical mechanical work.

[0014] 3. This invention constructs an input vector based on heart rate reserve, heart rate amplification offset, target vertical height, and metabolic energy demand, and inputs it into a pre-trained feedforward neural network model. This enables the model to simultaneously consider four key aspects: the user's cardiac regulation capacity, vertical motion sensitivity, physiological load, and energy demand. Through the nonlinear mapping capability of the network's internal parameters, it outputs the replenishment amounts of carbohydrates, proteins, lipids, and electrolytes. This ensures that the final health recommendation results not only accurately reflect the load of the exercise task itself but also match the individual physiological differences of the user, thereby significantly improving the scientific nature, practicality, and individual adaptability of post-exercise replenishment guidance. Attached Figure Description

[0015] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings: Figure 1 This is a flowchart illustrating a method for generating health recommendation results based on artificial intelligence, as provided in an embodiment of the present invention. Detailed Implementation

[0016] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0017] Example: This example provides a method for generating health recommendation results based on artificial intelligence. See [link to example]. Figure 1 Specifically, including: S1. Collect the user's heart rate and altitude data, and calculate the heart rate reserve based on the user's heart rate data and age; In embodiments of the present invention, collecting the user's heart rate data and altitude data, and calculating heart rate reserve based on the user's heart rate data and age, includes: Collect users' height and heart rate data; Determine the time period during which the user is in a resting state, and use the average heart rate data during that time period as the user's resting heart rate; Obtain the user's age and determine the user's maximum heart rate based on the user's age; The user's heart rate reserve is obtained based on the difference between the user's maximum heart rate and resting heart rate. Specifically, altitude data refers to the instantaneous spatial height of the user relative to a reference plane measured by sensors, reflecting the user's positional changes in the vertical direction; heart rate data refers to the number of heartbeats recorded per unit time at each sampling moment, reflecting the strength and frequency of cardiac pumping activity; the resting state time period refers to a period of time during which the user remains relatively quiet while in a low-intensity activity state such as sitting or lying down, during which cardiovascular load is low and fluctuations are small; the average heart rate data refers to the arithmetic mean of all sampled heart rates within the resting state time period, representing the overall stable level of cardiac activity during that time period; resting heart rate refers to the above average heart rate value, corresponding to the cardiac output level required by the body to maintain basic life activities in a resting state; maximum heart rate is the upper limit of cardiac activity allowed under safe conditions based on age estimation, used to characterize the limit level of an individual's cardiac capacity during strenuous exercise; heart rate reserve is used to characterize the amount of cardiac regulatory capacity that can be dispatched during exercise regulation.

[0018] Specifically, the system uses a smart terminal device worn by the user to acquire ambient air pressure and the user's pulse signal in real time through built-in barometric pressure and photoplethysmography (PPG) sensors. The processor reads the sensor data at a preset sampling frequency, such as one hertz, converts the read air pressure value into relative altitude data based on the standard atmospheric pressure formula, and converts the pulse signal into heart rate data in terms of beats per minute. The system assigns the same timestamp to the altitude and heart rate data collected at the same time, forming a one-to-one time series data stored in the local memory. The one hertz sampling frequency can balance the real-time performance of data acquisition with the power consumption control of the device. The system continuously monitors the triaxial output values ​​of the accelerometer. When the variance of the acceleration value remains within a threshold range close to zero for a preset duration, such as three minutes, it determines that the user is in a relatively still and metabolically stable state. This time range is then marked as the resting state period. Subsequently, the processor extracts all heart rate data sampling points recorded within this period, calculates the arithmetic mean by summing the data and dividing by the total number of sampling points, and sets this arithmetic mean as the user's resting heart rate to eliminate random errors caused by instantaneous emotional fluctuations or minor movements on the heart rate baseline data.

[0019] Specifically, the system obtains the user's current actual age by acquiring the birth date stored in the user's personal profile module or the age value entered directly. Then, it applies a formula based on the statistical laws of physiological big data to calculate the user's maximum heart rate. Specifically, it uses the product of the baseline constant 208 and the coefficient 0.7 and the user's age to obtain the result. The baseline constant 208 and the coefficient 0.7 are widely recognized empirical parameters in the field of physiology used to correct the negative impact of aging on the heart's maximum pumping frequency. Compared with traditional formulas, this method can more accurately estimate the physiological limits of people of different age groups.

[0020] Specifically, the system retrieves the calculated maximum heart rate and resting heart rate values ​​recorded in memory and performs a subtraction operation, subtracting the resting heart rate value from the maximum heart rate value to obtain a difference. This difference is defined as the user's heart rate reserve. This heart rate reserve value objectively reflects the potential increase in heart rate when the user's heart transitions from a resting, low-metabolic state to a high-metabolic state during extreme exercise. It is used for subsequent quantitative assessment of the user's physiological load level and endurance potential during vertical exercise.

[0021] S2. Identify vertical ascent activity segments based on the user's height data and calculate the corresponding height increment; In an embodiment of the present invention, identifying vertically ascending activity segments based on the user's height data and calculating the corresponding height increment includes: The height data is divided into segments based on a preset time window length to obtain height change segments; Calculate the time mean and height mean of the height change segments; Based on the time mean and height mean of the height change segment, a linear fit is performed on the time data and height data within the height change segment to obtain the height change slope; Specifically, the height change slope refers to the average rate of change of height data over time within a height change segment. It represents the degree of increase or decrease in height per unit time and reflects the user's vertical movement trend within that segment. When height increases over time, the height change slope is positive, indicating upward displacement by the user within that segment; when height decreases over time, the height change slope is negative, indicating downward displacement by the user within that segment; when the height change slope is close to zero, it indicates that the user remains at approximately the same vertical height within that segment. The height change slope is derived from the slope value obtained by linearly fitting time and height data within a height change segment. It stably reflects the overall direction and speed of height change within the segment, providing a foundation for subsequent judgment of height increments and identification of vertically ascending activity segments. Specifically, a time window length is set. This time window length is based on the fact that during human vertical activities such as walking and climbing stairs, the height change is usually continuous between two and ten seconds. Meanwhile, the sensor sampling interval is usually one second. Therefore, the time window length is preferably set to an integer value within the range of five to twenty seconds to ensure that each window can cover a complete micro vertical displacement process. The corresponding time intervals are sequentially extracted on a continuous time axis according to this time window length. All continuous height sampling points and their corresponding timestamps contained in each time interval are defined as an independent height change segment. This discretizes the long-cycle monitoring data stream into multiple data subsets that can be used for local trend analysis, making it easier to identify the user's vertical movement trend in different time periods. For each height variation segment, the time data of all sampling moments within that segment is read, and the arithmetic mean of all time data is calculated to obtain the segment's time mean. Subsequently, the height data corresponding to all sampling moments within that segment is read, and the arithmetic mean of all height data is calculated to obtain the segment's height mean. The purpose of setting the time mean is to provide the segment's central time position, providing a stable time reference point for subsequent linear fitting. The purpose of setting the height mean is to provide a representative value of the overall height level of the segment, reducing the impact of random fluctuations in local height data on subsequent fitting. The resulting time mean and height mean will serve as a centralized reference point for the data within the segment, used to improve the robustness and noise resistance of linear fitting.

[0022] Specifically, the time mean is subtracted from all time data within the height change segment, and the height mean is subtracted from the corresponding height data, so that all samples within the segment are realigned around the segment's center point. This centering process is designed to eliminate the offset caused by differences in the start times of different segments, ensuring that the fitting process only reflects the height change trend. Subsequently, the least squares method is used to establish a linear relationship between the centered time data and the height data. By solving for the slope parameter, the average rate of change of height over time within the height change segment is obtained, i.e., the height change slope. Since the height change slope reflects the trend of height increase or decrease per unit time, a positive slope indicates that the segment has a net upward movement, a negative slope indicates that the segment has a net downward movement, and a slope close to zero indicates that the height remains basically stable.

[0023] The duration of the altitude change segment is obtained by the difference between the end time and the start time within the segment. The height increment of a height change segment is obtained by multiplying the time length of the height change segment by the slope of the height change segment. The height increment of each height change segment is judged. When the height increment is greater than zero, the corresponding height change segment is determined as a vertical ascent activity segment. Specifically, a vertical ascent activity segment refers to a segment within a height change segment where the calculated height increment is greater than zero. This signifies that the user has made a continuous upward movement relative to a reference plane within that segment. Vertical ascent activity segments typically correspond to activities such as climbing stairs, ascending heights, or other actions involving vertical displacement. The criterion for identification is whether the height increment of the segment is positive, and the height increment is determined by the product of the height change slope and the segment's duration. When the height increment is greater than zero, it indicates that the user has completed an overall upward displacement within that segment, thus identifying it as a vertical ascent activity segment. Vertical ascent activity segments reflect the user's actual vertical load process.

[0024] Specifically, the system reads the start and end times of the height change segment, which are derived from the boundary information of the height change segment obtained by dividing it into time windows. The system then subtracts the start time from the end time to obtain the total duration of the height change segment. To ensure that the duration accurately reflects the actual vertical movement time within the segment, the system checks whether the segment contains continuous sampling data before calculation to avoid abnormal time differences due to missing samples. The system multiplies the duration by the height change slope to obtain the net height change of the segment within the entire window. To ensure that the height increment accurately reflects the vertical displacement, the sign of the height change slope is checked before execution. When the height change slope is positive, it indicates that the height within the window is generally increasing over time, and the product will be positive; when the height change slope is negative, it indicates that the height is decreasing, and the product will be negative. The height increment of the height change segment converts the velocity-based slope into a displacement-based height increment, which can be used to determine whether the user has experienced upward vertical movement, thus providing a direct basis for identifying vertically ascending activity segments.

[0025] Specifically, the height increments of all height change segments are read, and each increment is compared with a zero value. This zero-value threshold is set based on the fact that the positive and negative directions of the height increment directly correspond to the spatial displacement direction in the vertical direction. When the height increment is greater than zero, it indicates that the overall height within the segment is rising; when the height increment is equal to zero or less than zero, it indicates that the segment is stationary or trending downwards, respectively. Since the zero value itself represents the neutral point of vertical displacement, and vertical changes do not require offset calibration in sensor sampling, using the zero value as the classification threshold has objectivity and universality. When a height increment is determined to be greater than zero, the corresponding segment is marked as a segment of vertical upward activity, ensuring that subsequent calculations are based only on segments exhibiting genuine vertical upward behavior, thus improving the validity of the data and the reliability of the analysis results.

[0026] S3. Calculate the vertical sensitivity sequence based on heart rate data and altitude increments within the vertical ascent activity segment; In an embodiment of the present invention, calculating a vertical sensitivity sequence based on heart rate data and altitude increments within a vertically ascending activity segment includes: Calculate the mean of heart rate data within the vertical ascending activity segment, and obtain the heart rate increment of the vertical ascending activity segment based on the difference between the mean and the resting heart rate; The work done by gravity is obtained by multiplying the height increment of the vertical ascent segment by the acceleration due to gravity. When the vertical gravitational work is greater than zero, the ratio of the heart rate increment to the vertical gravitational work during the vertical ascent segment is used as the vertical sensitivity value. The vertical sensitivity values ​​of each vertically ascending activity segment are summarized to form a vertical sensitivity sequence; Specifically, heart rate increment refers to the increase in a user's heart rate relative to their resting heart rate during a vertical ascent segment. It is an important parameter reflecting the additional pumping load on the heart during vertical ascent; a larger heart rate increment indicates a higher cardiovascular burden on the user during this ascent. Vertical gravitational work refers to the effective mechanical work done by the user against gravity during a vertical ascent segment. Vertical gravitational work is obtained by multiplying the user's height increment in that segment by the acceleration due to gravity. The height increment represents the amount of displacement completed in the vertical direction within the segment. A larger vertical gravitational work indicates a greater upward displacement or a higher energy demand during the ascent, reflecting the degree of anti-gravity during vertical movement. Vertical sensitivity is a quantitative indicator used to characterize the relationship between a user's heart rate response and vertical gravitational work during vertical ascent activities. The vertical sensitivity value is obtained by calculating the ratio of heart rate increment to vertical gravitational work. This ratio represents the increase in heart rate caused by a unit of vertical gravitational work. The larger the vertical sensitivity value, the more sensitive the user's heart rate increase is to vertical ascent load; the smaller the vertical sensitivity value, the better the user's endurance or the smaller the heart rate response during vertical exercise.

[0027] Specifically, for each vertical ascent activity segment, firstly, the heart rate data corresponding to all sampling times within the vertical ascent activity segment is read, and an arithmetic mean is calculated on these heart rate data to obtain a heart rate mean that can represent the overall heart rate level of the segment; after obtaining the heart rate mean, the resting heart rate determined in the previous step is read, and the heart rate mean is subtracted from the resting heart rate to obtain the heart rate increment of the vertical ascent activity segment. This heart rate increment represents the amount of additional cardiac load change caused by the user's vertical ascent behavior within the segment. The vertical gravitational work is obtained by multiplying the gravitational acceleration by the height increment, where the gravitational acceleration is 9.8 m / s². This vertical gravitational work measures the amount of anti-gravity work performed by the user within a segment. This vertical gravitational work is compared to zero; if it is greater than zero, it indicates a real upward displacement in that segment. The aforementioned heart rate increment is divided by the vertical gravitational work to obtain the vertical sensitivity value. This value characterizes the sensitivity of the heart rate response intensity to the scale of vertical mechanical work output. If the vertical gravitational work is zero or less than zero, no vertical sensitivity value is generated for that segment to avoid situations where the physical meaning is invalid. The vertical sensitivity values ​​corresponding to all vertical ascent activity segments are read sequentially. Each value is stored according to the segment arrangement or chronological order, and all values ​​are combined into a sequence structure. This sequence serves as the data basis for subsequently generating heart rate amplification offset numbers, centrally presenting the user's heart rate response performance in different vertical ascent segments. Based on this sequence, individual heart rate response characteristics and adaptability can be further identified.

[0028] S4. Perform statistical analysis on the vertical sensitivity sequence to generate the heart rate amplification offset number; In an embodiment of the present invention, statistical analysis is performed on the vertical sensitivity sequence to generate a heart rate amplification offset number, including: Arrange the vertical sensitivity sequence in ascending order to obtain an ordered sequence; Extract a preset number of element values ​​starting from the end of the ordered sequence, calculate the average value of the extracted element values, and use the average value as the heart rate amplification offset. Specifically, the heart rate amplification offset refers to the value obtained by extracting a predetermined number of vertical sensitivity values ​​sequentially from the end of the ordered sequence after arranging the vertical sensitivity sequence in ascending order, and then performing an arithmetic mean on these extracted vertical sensitivity values. This value characterizes the high sensitivity level exhibited by the user during vertical ascent activities, reflecting the degree of enhancement of cardiac activity relative to vertical mechanical load in several segments of the user's strongest vertical heart rate response. By averaging the larger vertical sensitivity values ​​at the end of the ordered sequence, the high-response areas in the user's physiological response can be centrally measured, thereby generating an evaluation quantity that represents the amplification trend of the user's vertical heart rate response. This value is used as an input parameter when subsequently calculating the target vertical height and metabolic energy requirements to improve the suitability of personalized recommendations.

[0029] Specifically, the generated vertical sensitivity sequence is read, and the vertical sensitivity values ​​in the sequence are sorted from smallest to largest. After sorting, an ordered sequence corresponding one-to-one with the original sequence is obtained, and the numerical positional relationship of each vertical sensitivity value is preserved in the ordered sequence. Then, the number of elements used to characterize the high-sensitivity response region is preset according to the application scenario. The preset number is set based on the total number of vertically ascending activity segments and the proportion of high-sensitivity segments to be focused on. Generally, it is an integer value in the range of one-tenth to one-third of the total number of segments, so as to highlight the influence of high-sensitivity segments while ensuring the representativeness of the sample size. After determining the preset number, the same number of vertical sensitivity values ​​as the preset number are extracted sequentially from the end of the ordered sequence. All extracted vertical sensitivity values ​​are accumulated, and the accumulated result is divided by the preset number to obtain the arithmetic mean of these high-sensitivity vertical sensitivity values. This average value is recorded as the heart rate amplification offset number, so that the heart rate amplification offset number can reflect the overall level of the strongest response segment in the user's vertical sensitivity sequence, providing a key parameter reflecting the individual's heart rate amplification trend for subsequent calculations of target vertical height and metabolic energy demand.

[0030] S5. Obtain the user's weight, and calculate the target vertical height and metabolic energy requirement based on heart rate reserve, user's weight and heart rate amplification offset. In an embodiment of the present invention, the user's weight is obtained, and the target vertical height and metabolic energy requirement are calculated based on heart rate reserve, the user's weight, and heart rate amplification offset, including: The vertical gravitational work target is calculated based on the heart rate amplification offset and heart rate reserve. The formula for calculating the vertical gravitational work target is as follows:

[0031] In the formula, For the vertical work done by gravity. For heart rate reserve, This represents the heart rate amplification offset. The vertical work target ratio coefficient is a pre-set value. Specifically, heart rate reserve reflects a user's effective cardiac output capacity when transitioning from a resting to an active state, while heart rate amplification offset reflects the additional response amplitude of heart rate with increasing height during vertical ascent. Both are directly related to the physiological load required for an individual to complete mechanical work in the direction of gravity. A higher heart rate reserve indicates a stronger mobilized physiological capacity, enabling the user to complete higher gravitational work output under the same vertical task. A larger heart rate amplification offset indicates a more significant increase in heart rate during vertical movement, requiring a higher metabolic response per unit height. Therefore, when allocating target vertical gravitational work, heart rate reserve should be used as the denominator of the carrying capacity, and heart rate amplification offset should be used as the numerator reflecting vertical load sensitivity. A proportional coefficient should be used to unify the physiological differences among individuals, ensuring that the calculated target vertical gravitational work accurately reflects the mechanical work required for a user to resist external forces in the vertical direction. The vertical work target proportionality coefficient was determined through a combination of experimental measurement and statistical regression. Under controlled conditions, multiple standardized vertical ascent tasks were arranged. Each task had its floor height, ascent speed, and total duration pre-set. Simultaneously, the participants' height, heart rate, resting heart rate, and weight before the task were recorded. Then, the actual vertical gravitational work value for each task was calculated based on the height data. Heart rate reserve was calculated based on resting heart rate and maximum heart rate. Heart rate amplification offset was obtained from the vertical sensitivity sequence using the aforementioned method. These sample data were then organized into a correspondence between the actual vertical gravitational work value and heart rate reserve and heart rate amplification offset. Based on this, regression analysis was used to fit the proportional relationship between the actual vertical gravitational work value and heart rate reserve and heart rate amplification offset in different user and task samples. The fitting result minimized the statistical error between the predicted vertical gravitational work target and the measured vertical gravitational work, thus obtaining one or more stable numerical ranges. The representative value within this range was determined as the vertical work target proportionality coefficient.

[0032] The vertical height of the target is obtained by the ratio of the work done by the target under vertical gravity to the acceleration due to gravity. The system obtains the user's weight and calculates the metabolic energy requirement by multiplying the user's weight by the target vertical gravitational work and the preset energy conversion ratio. Specifically, the target vertical height is a height quantity obtained based on the ratio between the target vertical gravitational work and gravitational acceleration. It represents the range of vertical displacement that a user needs to achieve in a complete physical activity. This height quantity reflects the displacement requirement that the user needs to overcome to complete the target movement task. Essentially, it is the equivalent displacement obtained after converting gravitational acceleration under a given energy demand, used to characterize the actual vertical ascent the user should achieve. Metabolic energy demand is the energy quantity obtained by multiplying the user's weight, the target vertical gravitational work, and the energy conversion ratio. It represents the total amount of additional metabolic energy that the body needs to output to complete the target vertical height movement. This energy reflects the physiological energy load that the body needs to supplement or consume to complete a specific mechanical work, and is used to further generate individualized nutritional supplementation or activity intensity recommendations.

[0033] Specifically, after obtaining the target vertical gravitational work, the gravitational acceleration value used to convert the target vertical height is first determined. The gravitational acceleration is taken as the national basic physical constant standard value of 9.81 meters per square second. Then, the target vertical gravitational work is divided by the gravitational acceleration to obtain the target vertical height, which is used to characterize the amount of vertical displacement that the user needs to complete in future activities. The user's weight data is obtained, which can be obtained from wearable devices, smart scales, or user input, ensuring that it belongs to the same cycle as the heart rate data. Then, an energy conversion ratio coefficient is selected. The energy conversion ratio coefficient is determined based on the quantification of the metabolic energy required for the human body to complete a unit of vertical gravitational work in vertical ascent. It is usually set between two and four, and is obtained by experimentally measuring the statistical proportional relationship between the metabolic consumption and vertical gravitational work of a user group in a standard climbing task. Then, the user's weight, the target vertical gravitational work, and the energy conversion ratio coefficient are multiplied sequentially to obtain the metabolic energy requirement. The metabolic energy requirement is used to represent the total amount of energy that the user needs to supplement to complete the target vertical height, and can be used as a basis parameter for subsequent health recommendations.

[0034] S6. Input the heart rate reserve, heart rate amplification offset, target vertical height, and metabolic energy requirement into the pre-trained neural network model to generate health recommendation results; In embodiments of the present invention, heart rate reserve, heart rate amplification offset, target vertical height, and metabolic energy requirements are input into a pre-trained neural network model to generate health recommendation results, including: The heart rate reserve, heart rate amplification offset, target vertical height, and metabolic energy requirement are combined to construct a feature vector; The feature vector is input into a pre-trained feedforward neural network model to obtain the nutrient supply amount, which includes: carbohydrate supply amount, protein supply amount, lipid supply amount and electrolyte supply amount; Generate health recommendations that include target vertical height and nutrient intake; Specifically, after obtaining heart rate reserve, heart rate amplification offset, target vertical height, and metabolic energy demand, heart rate reserve is placed at the beginning of the feature vector, heart rate amplification offset is placed after it, target vertical height is placed in the third position, and metabolic energy demand is placed in the fourth position. By concatenating these elements sequentially, a one-dimensional data structure containing four numerical elements is formed, which can then be used as a unified input for the neural network model.

[0035] Specifically, in the process of inputting feature vectors into a pre-trained feedforward neural network model to obtain nutrient supply amounts, a feature vector containing heart rate reserve, heart rate amplification offset, target vertical height, and metabolic energy requirement is first prepared. Based on the numerical range and scale used by the pre-trained feedforward neural network model during training, normalization or scaling is performed on each parameter in the feature vector, consistent with the training phase. This ensures that heart rate, height, and energy parameters remain numerically stable and fall within the effective range learned by the model, avoiding numerical imbalances caused by differences in units. The pre-trained feedforward neural network model refers to a multi-layer nonlinear mapping model that has been trained using historical sample data on the weights and bias parameters of each layer before the method runs. This model adopts a unidirectional propagation structure from input to output, consisting of an input layer, hidden layers, and an output layer. Numerical transfer and nonlinear transformation are achieved between layers through fixed connection weights. In the online application phase, only forward computation is performed without updating the internal parameters. To ensure accurate mapping capabilities during the training phase, training samples were selected that covered user data of different ages, weights, vertical activity levels, and heart rate amplification offsets. This allowed the model parameters to adapt to the physiological distribution of different individuals. Each sample was also assigned a nutritional supplementation amount matching the actual consumption, including carbohydrate, protein, lipid, and electrolyte supplementation. The network weights and biases were repeatedly adjusted to minimize the error between the predicted output and the actual labeling until the model's fitting error to the sample data was controlled within the target range. This resulted in a set of network parameters that stably characterizes the relationship between input features and nutritional requirements. During method execution, the organized feature vector was input element-wise into the model's input layer, with the four input nodes corresponding to heart rate reserve, heart rate amplification offset, target vertical height, and metabolic energy requirement, respectively. After receiving the input, the input layer passed the values ​​to the first hidden layer according to their respective connection weights. The hidden layer performed nonlinear transformations through linear weighted sum operations and activation functions, allowing the latent relationships between the input parameters to be expressed in a high-dimensional space. Subsequently, the output of the first hidden layer continues to be passed to the second hidden layer, repeating the weighting and activation operations, causing the model to generate higher-level descriptions of combinatorial relationships layer by layer. This process continues layer by layer in the same way, making the output of each layer the input of the next layer, until the data reaches the output layer. The output layer contains four output nodes, corresponding to carbohydrate replenishment, protein replenishment, lipid replenishment, and electrolyte replenishment, respectively. Each output node generates a continuous value based on the weighted summation and activation function result of the final layer. This value represents the specific nutritional replenishment requirement corresponding to the target vertical activity.The four nutrient supply values ​​generated by the output layer are used as nutrient supply vectors to obtain quantitative supplementation suggestions for four types of nutrients: carbohydrates, protein, lipids, and electrolytes. After obtaining the nutrient supply values, the target vertical height is used as the first element of the recommendation results, and the carbohydrate, protein, lipid, and electrolyte supply values ​​are used as the second to fifth elements of the recommendation results, respectively, forming a complete set of recommendation results. This allows users to receive quantitative nutrient intake guidance while performing exercise tasks corresponding to the target vertical height, ensuring metabolic recovery, energy compensation, and physiological balance after activity.

[0036] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for generating health recommendation results based on artificial intelligence, characterized in that, Includes the following steps: S1. Collect the user's heart rate and altitude data, and calculate the heart rate reserve based on the user's heart rate data and age; S2. Identify vertical ascent activity segments based on the user's height data and calculate the corresponding height increment; S3. Calculate the vertical sensitivity sequence based on heart rate data and altitude increments within the vertical ascent activity segment; S4. Perform statistical analysis on the vertical sensitivity sequence to generate the heart rate amplification offset number; S5. Obtain the user's weight, and calculate the target vertical height and metabolic energy requirement based on heart rate reserve, user's weight and heart rate amplification offset. S6. Input the heart rate reserve, heart rate amplification offset, target vertical height, and metabolic energy requirement into the pre-trained neural network model to generate health recommendation results.

2. The method for generating health recommendation results based on artificial intelligence according to claim 1, characterized in that, Collects user's heart rate and altitude data, and calculates heart rate reserve based on the user's heart rate data and age, including: Collect user's altitude and heart rate data; Determine the time period during which the user is in a resting state, and use the average heart rate data during that time period as the user's resting heart rate; Obtain the user's age and determine the user's maximum heart rate based on the user's age; The user's heart rate reserve is obtained by calculating the difference between the user's maximum heart rate and resting heart rate.

3. The method for generating health recommendation results based on artificial intelligence according to claim 1, characterized in that, Identify vertically ascending activity segments based on user height data and calculate the corresponding height increments, including: The height data is divided into segments based on a preset time window length to obtain height change segments; Calculate the time mean and height mean of the height change segments; Based on the time mean and height mean of the height change segment, a linear fit is performed on the time data and height data within the height change segment to obtain the height change slope; The duration of the altitude change segment is obtained by the difference between the end time and the start time within the segment. The height increment of a height change segment is obtained by multiplying the time length of the height change segment by the slope of the height change segment. The height increment of each height change segment is judged. When the height increment is greater than zero, the corresponding height change segment is determined as a vertical ascent activity segment.

4. The method for generating health recommendation results based on artificial intelligence according to claim 1, characterized in that, Vertical sensitivity sequences are calculated based on heart rate data and altitude increments within vertical ascent segments, including: Calculate the mean of heart rate data within the vertical ascending activity segment, and obtain the heart rate increment of the vertical ascending activity segment based on the difference between the mean and the resting heart rate; The work done by gravity is obtained by multiplying the height increment of the vertical ascent segment by the acceleration due to gravity. When the vertical gravitational work is greater than zero, the ratio of the heart rate increment to the vertical gravitational work during the vertical ascent segment is used as the vertical sensitivity value. The vertical sensitivity values ​​of each vertically ascending activity segment are summarized to form a vertical sensitivity sequence.

5. The method for generating health recommendation results based on artificial intelligence according to claim 1, characterized in that, Statistical analysis of the vertical sensitivity sequence was performed to generate heart rate amplification offset numbers, including: The vertical sensitivity sequence is sorted in ascending order to obtain an ordered sequence. Extract a preset number of element values ​​starting from the end of the ordered sequence, calculate the average value of the extracted element values, and use the average value as the heart rate amplification offset.

6. The method for generating health recommendation results based on artificial intelligence according to claim 1, characterized in that, Target vertical height and metabolic energy requirements are calculated based on heart rate reserve, user weight, and heart rate amplification offset, including: The vertical gravitational work target is calculated based on the heart rate amplification offset and heart rate reserve. The formula for calculating the vertical gravitational work target is:

7. In the formula, For the vertical work done by gravity For heart rate reserve, This represents the heart rate amplification offset. The vertical work target ratio coefficient is a pre-set value. The vertical height of the target is obtained by the ratio of the work done by the target under vertical gravity to the acceleration due to gravity. The system obtains the user's weight and calculates the metabolic energy requirement by multiplying the user's weight by the target vertical gravitational work and a preset energy conversion ratio.

8. The method for generating health recommendation results based on artificial intelligence according to claim 1, characterized in that, Heart rate reserve, heart rate amplification offset, target vertical height, and metabolic energy requirements are input into a pre-trained neural network model to generate health recommendation results, including: The heart rate reserve, heart rate amplification offset, target vertical height, and metabolic energy requirement are combined to construct a feature vector; The feature vector is input into a pre-trained feedforward neural network model to obtain the nutrient supply amount, which includes: carbohydrate supply amount, protein supply amount, lipid supply amount and electrolyte supply amount; Generate health recommendations that include target vertical height and nutrient intake.

Images