Early warning system for obesity risk of teenagers

Abstract

The invention discloses a teenager obesity risk early warning system, which relates to the technical field of teenager human body health weight risk prediction and evaluation, and comprises an electronic scale, a data acquisition module, a dynamic growth and development correction module, a behavior-physiology linkage analysis module, an edge calculation early warning module and a family-school linkage feedback module. Through dynamic evaluation, multi-source data integration, real-time early warning and collaborative intervention, the limitation that a traditional method is inaccurate in evaluation, lagged in early warning and single in intervention is overcome, the accuracy, timeliness and prevention and control effect of adolescent obesity risk monitoring are remarkably improved, and the method is easy and convenient to operate and suitable for long-term use.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of adolescent human health weight risk prediction and evaluation, in particular to an adolescent obesity risk early warning system. BACKGROUND

[0002] With the increasing severity of obesity problems among adolescents, obesity risk monitoring and prevention have become an important issue in the field of health management. In the prior art, the commonly used obesity assessment method mainly relies on body mass index (BMI) as a single indicator, which measures body weight by a body scale and calculates BMI in combination with height, compares it with a fixed threshold to determine obesity risk. In addition, some intelligent devices (such as wearable wristbands) can collect physiological data such as heart rate and step count, and some health management applications allow users to record dietary and exercise behavior, and provide health advice in combination with cloud analysis. However, these technologies have several shortcomings. First, the traditional BMI assessment uses a static threshold, which does not fully consider the dynamic characteristics of adolescent growth and development, for example, rapid height growth during puberty can cause temporary high BMI but not obesity, which can easily lead to misjudgment. Second, existing systems are mainly based on a single data source, lacking deep integration and causal analysis of physiological data and behavioral data, and unable to accurately identify the driving factors of obesity, such as dietary habits or lack of sleep. Third, most monitoring devices rely on cloud computing, and there is a delay in data transmission and processing, making it difficult to achieve real-time warning and affecting the timeliness of intervention. In addition, existing technologies usually only target individual users, lacking a multi-party collaborative intervention mechanism such as family and school, resulting in limited health management effectiveness.

[0003] In view of the above problems, there is an urgent need for an adolescent obesity risk early warning system to solve these problems. SUMMARY

[0004] The purpose of the present application is to provide an adolescent obesity risk early warning system to solve the problems raised in the background art.

[0005] To achieve the above purpose, the present application provides the following technical solution: an adolescent obesity risk early warning system, characterized in that it comprises an electronic scale, a data acquisition module, a dynamic growth and development correction module, a behavior-physiology linkage analysis module, an edge computing early warning module, and a family-school linkage feedback module, wherein:

[0006] The electronic scale is used for regular weighing and acquisition of adolescent body weight data;

[0007] The data acquisition module acquires physiological data P d and behavioral data B d through the electronic scale and wearable devices, the wearable devices including smart wristbands or smart watches;

[0008] The dynamic growth and development correction module dynamically adjusts the obesity risk threshold T based on growth curve and time series prediction technology. adj ;

[0009] The behavior-physiology linkage analysis module calculates the conditional probability P(P d |B d ), identifying key drivers of obesity risk;

[0010] The edge computing warning module monitors and calculates the risk index R(t) in real time on the electronic scale to trigger an early warning;

[0011] The home-school linkage feedback module calculates the intervention priority I according to the risk index R(t) p And push it to parents and schools.

[0012] As a preferred technical solution of the present invention, the electronic scale integrates a weighing sensor, a microprocessor and a wireless communication module, supports a periodic weighing function, and the user can automatically record the weight data W(t) at preset time intervals, and the data is uploaded to a mobile terminal or a cloud database in real time via Bluetooth or Wi-Fi.

[0013] As a preferred technical solution of the present invention, the physiological data Pd collected by the data acquisition module includes weight W(t) and BMI(t) measured by electronic scales, as well as heart rate, sleep duration and number of steps collected by wearable devices, and behavioral data B d Including dietary calorie intake, exercise frequency and screen usage time, all data is stored in time series form and updated regularly.

[0014] As a preferred technical solution of the present invention, the dynamic growth and development correction module uses the following formula to calculate the adjusted obesity risk threshold T adj :

[0015] T adj =T base +α·ΔG(t)+β·P(t)

[0016] Among them, T base is the standard BMI threshold based on age and gender, ΔG(t) is the standardized height growth rate, P(t) is the BMI change trend predicted in the next 6-12 months based on the regular weighing data W(t) of the electronic scale, α and β are correction coefficients, and P(t) is calculated using the short-term memory network model.

[0017] As a preferred technical solution of the present invention, the behavior-physiology linkage analysis module calculates the behavior data B through the Bayesian causal network. d With physiological data P d The conditional probability P(P d |Bd ):

[0018]

[0019] wherein P d represents physiological data, B d represents behavior data, P(P d |B d ), P(B d |P d ), P(P d ) and P(B d ) are probability values for identifying the causal relationship between behavior and weight change.

[0020] As a preferred technical solution of the present application, the edge computing early warning module runs a lightweight risk assessment algorithm on the microprocessor of the electronic scale to calculate a real-time risk index R(t):

[0021] R(t) = w1·BMI(t) + w2·ΔW(t) + w3·E(t)

[0022] wherein BMI(t) is the current BMI, ΔW(t) is the weight change rate recorded by the electronic scale during regular weighing, E(t) is the exercise index, w1, w2, w3 are prediction weights, and R(t) is a dimensionless risk index that triggers a local warning when it exceeds a preset threshold.

[0023] As a preferred technical solution of the present application, the home-school linkage feedback module supports multi-user permission management and calculates an intervention priority I p :

[0024] I p = γ·R(t) + δ·F(t)

[0025] wherein R(t) is the risk index, F(t) is the execution feasibility score of the feedback from the family and school, γ and δ are dimensionless weighting coefficients, and I p is a dimensionless priority score.

[0026] As a preferred technical solution of the present application, the long short-term memory network model is based on the electronic scale regular weighing data W(t) for prediction, and the formula is:

[0027] P(t+1) = σ(W f ·[h(t), x(t)] + b f )

[0028] wherein P(t+1) is the BMI prediction value at the next time step, h(t) is the hidden state at the previous time step, x(t) is the current input weight data after standardization, W fand b f are weight matrix and bias respectively, and sigma is a dimensionless activation function.

[0029] As a preferred technical solution of the present application, the behavior-physiology linkage analysis module combines group comparison learning, and optimizes the causal model through the following loss function:

[0030]

[0031] wherein y i is a real causal label, is a predicted probability based on the body weight data W(t) of the electronic scale, D group is the difference between the individual and the group weight distribution, and lambda is a regularization parameter, and L is a dimensionless loss value.

[0032] As a preferred technical solution of the present application, the electronic scale has a low-power design, built-in voice prompt function and LED display screen, and supports users to set a regular weighing plan through a key or a mobile application.

[0033] Compared with the prior art, the present application has the following beneficial effects:

[0034] The present application provides a youth obesity risk early warning system, which has significant technical innovation advantages compared with the prior art. Firstly, through the dynamic growth and development correction module, combined with the growth curve and long and short term memory network prediction technology, the system dynamically adjusts the obesity risk threshold according to the height growth rate of the youth growth stage and the future BMI change trend, overcomes the limitations of the traditional static BMI threshold, and significantly improves the individualization accuracy of risk assessment. Secondly, the data acquisition module integrates physiological data (such as weight, heart rate, sleep, and steps) and behavior data (such as calorie intake, exercise frequency, and screen time) collected by the electronic scale and wearable devices, and calculates the conditional probability through the Bayesian causal network, realizing multi-source data fusion and accurate identification of obesity driving factors, which is more scientific than single index monitoring. In addition, the edge computing early warning module calculates the risk index in real time on the electronic scale and triggers the early warning, which overcomes the delay problem of traditional cloud computing, ensuring the timeliness of the early warning.

[0035] The practicality and application value of the present invention are also outstanding. The family-school linkage feedback module calculates the intervention priority according to the risk index, and by pushing specific suggestions to parents and schools, a closed-loop intervention system with multi-party collaboration is formed, which significantly improves the obesity prevention and control effect, and is better than the existing individualized monitoring methods. The system design is user-friendly, the electronic scale supports regular weighing plans and is equipped with an LED display and voice prompts, and the supporting application provides a food calorie database and photo recognition functions, which are easy to operate and enhance the compliance of adolescents and their families. At the same time, the electronic scale and wearable device adopt a low-power design, combined with efficient cloud analysis, taking into account both performance and cost, and are suitable for long-term use. In summary, the present invention provides an accurate, timely and efficient solution for adolescent obesity risk management through technological innovation and practical design. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is a system block diagram of a teenager obesity risk early warning system proposed by the present invention. DETAILED DESCRIPTION

[0037] The following is a combination of the embodiments of the present invention Figure 1 The technical solutions in the embodiments of the present invention are clearly and completely described. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0038] like Figure 1 As shown, the present invention provides a teenager obesity risk warning system, and its implementation is described in detail below so that those skilled in the art can understand and implement the present invention.

[0039] 1. Implementation of an electronic scale: This electronic scale is equipped with four high-precision strain gauge sensors and integrates Bluetooth 5.0 and Wi-Fi. The scale is used for regular weighing and weight data collection. After the user steps on the scale, the sensor samples the weight every 0.5 seconds, and the average of 10 times is used as the weight data W(t) (for example, 70.02 kg). Users can set a regular weigh-in schedule through the accompanying mobile app, for example, at 8:00 AM ± 15 minutes daily. If the user fails to weigh in on time, the LED displays "Please weigh" and a reminder tone plays.

[0040] Weight data is transmitted to the mobile phone via Bluetooth, and the transmission time is about 1 second. If it is not connected, the data is temporarily stored in 1KBEEPROM (can store 7 days of data, format: {date: "2025-03-26", time: "08:01", W(t): 70.02}).

[0041] 2. Implementation of data acquisition module: The data acquisition module collects physiological data P through electronic scales and wearable devicesd and behavior data B d , the electronic scale measures the weight W(t) = 70.02 kg, and the BMI(t) = 70.02 / 1.65 is calculated according to the height H(t) = 1.65 m input by the user in the App (updated once a month) 2 = 25.72 kg / m 2 .

[0042] The wearable device is a smart bracelet with an optical heart rate sensor, a three-axis accelerometer, and an infrared sleep monitoring module. The battery capacity is 135 mAh, and data is transmitted through Bluetooth 5.0. The bracelet records heart rate, sleep duration (6.2 hours / night), and steps (5000 steps / day) every 5 minutes, and synchronizes to the App at 23:00 every day. The data format is {timestamp: "2025-03-26 08:00:00", heart rate: 75, steps: 5000, sleep: 6.2}.

[0043] Behavior data B d includes dietary calorie intake (2500 kcal, e.g. breakfast 500 kcal + lunch 1000 kcal + dinner 800 kcal + snacks 200 kcal), exercise frequency (1 time / week, 30 minutes of running), and screen usage time (2.5 hours / day), which are manually input by the user through the App. The App provides a food calorie database (1000 foods, such as "whole wheat bread 250 kcal per 100g") and a photo recognition function (scan the barcode to automatically fill in the calories).

[0044] All data is stored in time series form on the Ali ECS server, and the cloud cleanses abnormal values (e.g. W(t) suddenly changes to 80 kg, which is excluded by ±5 kg) and normalizes (steps 0-20000 are mapped to 0-1, 5000 steps = 0.25).

[0045] 3. Implementation of dynamic growth and development correction module: growth curve and time series prediction technology adjust obesity risk threshold T adj According to the WHO adolescent growth and development standard, the standard BMI for a 14-year-old male is T base = 21.5 kg / m 2 , and the standard height H(t) = 1.68 m. The current height H(t) = 1.65 m, and last month H(t-1) = 1.64 m, the normalized high growth rate ΔG(t) = (1.65-1.64) / (1.68-1.67) = 1.0 (dimensionless) is calculated.

[0046] A long short-term memory network was used to predict the BMI trend P(t) over the next 6-12 months. The input was the W(t) sequence (69.8, 69.9, 70.02 kg) from the past 30 days, normalized to (-1.1). The model parameters were 1 neuron in the input layer, 32 neurons in the hidden layer (tanh activation), and 1 neuron in the output layer (linear activation). The learning rate was 0.001, and the training was repeated for 50 epochs. The loss function was the mean squared error (MSE). The training data was 30-day weight records of 1,000 adolescents aged 14-16 (approximately 1 GB). The model ran on a cloud-based NVIDIA GTX 1660 GPU (6 GB of video memory, TensorFlow 2.10). The predicted result P(t) was 1.3 kg / m 2 .

[0047] The calculation formula is T adj =T base +α·ΔG(t)+β·P(t), parameter α=2kg / m 2 (reflecting the impact of growth rate), β = 0.8 (dimensionless, adjusting the prediction weight) was optimized through 100-person testing and T was calculated. adj =21.5+2·1.0+0.81·3=24.54kg / m 2 .

[0048] 4. Implementation of the Behavior-Physiology Linkage Analysis Module The behavior-physiology linkage analysis module calculates the conditional probability P(P) through the causal reasoning model. d |B d ) to identify obesity risk drivers.

[0049] Physiological data P d Including weight change rate ΔW(t) = (70.02-69.9) / 1 = 0.12 kg / day, behavioral data B d Including calorie intake of 2500kcal and 6 hours of sleep.

[0050] A Bayesian causal network was used, with nodes defined as "calorie intake," "sleep duration," and "ΔW(t)." Edges were defined as "calories→ΔW(t)" and "sleep→ΔW(t)." The training data consisted of 7-day records of 1,000 adolescents (approximately 5MB). The conditional probability table was calculated in the cloud using maximum likelihood estimation. The results were P(ΔW(t)>0.1|calories>2000)=0.78 (78% probability of weight gain) and P(ΔW(t)>0.1|sleep<7)=0.65 (65% probability of weight gain).

[0051] Combined with the group contrast learning optimization model, calculate the individual and group differences: D group= 0.12 - 0.05 = 0.07 kg / day (population mean 0.05), regularization parameter λ = 0.01 (1 / kg), loss value after optimization reduced to 0.15.

[0052] 5. Implementation of edge computing early warning module: the edge computing early warning module runs a lightweight risk assessment algorithm on the electronic scale microprocessor, calculates the real-time risk index R(t) = w1·BMI(t) + w2·ΔW(t) + w3·E(t), the input data is BMI(t) = 25.72 kg / m 2 , ΔW(t) = 0.12 kg / day, exercise index E(t) = 5000 / 20000 = 0.25 (step number standardization), w1 = 0.5 m 2 / kg, w2 = 10 day / kg, w3 = 0.1 (dimensionless) through 100 person linear regression normalization, the calculation process is R(t) = 0.5·25.72 + 10·0.12 + 0.1·0.25 = 14.085.

[0053] The firmware is developed using C language, the default threshold is 10 (based on the mean value of healthy population), when R(t) = 14.085 > 10, the LED flashes, and the loudspeaker plays "Please pay attention to weight change".

[0054] 6. Implementation of family-school linkage feedback module: the family-school linkage feedback module calculates the intervention priority I p = γ·R(t) + δ·F(t) according to the risk index R(t), input R(t) = 14.085, parent score 0.9 (indicating high feasibility of intervention), school score 0.5 (limited by course arrangement), calculate the implementation feasibility F(t) = (0.9 + 0.5) / 2 = 0.7, parameters γ = 0.7, δ = 0.3 are optimized through user feedback, calculate I p = 0.7·14.085 + 0.3·0.7 = 9.8595 + 0.21 = 10.0695.

[0055] The module supports multi-user permission management, the database is stored in the cloud, the push service is implemented through Firebase Cloud Messaging, the default threshold is 8, when I p = 10.0695 > 8, push to parent's mobile phone "Suggest reducing daily caloric intake by 200 kcal (for example, reduce 1 piece of bread), increase walking to 8000 steps (about 1 hour)", and synchronously push to the school mailbox "Suggest increasing 1 time of 30-minute physical education class per week".

[0056] 7. Application example: user is a 14-year-old male, 1.65 m tall, 70.02 kg, 5000 steps, 6 hours of sleep, 2500 kcal of caloric intake, electronic scale records W(t) = 70.02 kg at 8:00 every day, calculates BMI(t) = 25.72 kg / m 2 , bracelet records heart rate 75 beats / min, sleep 6.2 hours, 5000 steps, user inputs 2500 kcal of calories, 1 time / week of exercise, 2.5 hours of screen time through the App.

[0057] The dynamic correction module calculates ΔG(t) = 1.0, P(t) = 1.3 kg / m 2 , obtains T adj = 24.5 kg / m 2 , BMI(t) is over standard.

[0058] The behavior-physiology analysis obtains P(ΔW(t) > 0.1 | calories > 2000) = 0.78.

[0059] The edge computing module calculates R(t) = 14.085, which exceeds the threshold value 10, triggering LED flashing and voice prompts.

[0060] The feedback module calculates I p = 10.0695, pushes intervention suggestions to parents and schools.

[0061] The content not described in detail in the description belongs to the prior art known to those skilled in the art, although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features, any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A teenager obesity risk warning system, characterized by: It includes an electronic scale, a data acquisition module, a dynamic growth and development correction module, a behavior-physiology linkage analysis module, an edge computing early warning module, and a home-school linkage feedback module, among which: The electronic scale is used to regularly weigh and collect weight data of adolescents; The data acquisition module collects physiological data P through electronic scales and wearable devices d and behavioral data B d , the wearable device includes a smart bracelet or a smart watch; The dynamic growth and development correction module dynamically adjusts the obesity risk threshold T based on growth curve and time series prediction technology. adj ; The behavior-physiology linkage analysis module calculates the conditional probability P(P d |B d ), identifying key drivers of obesity risk; The edge computing warning module monitors and calculates the risk index R(t) in real time on the electronic scale to trigger an early warning; The home-school linkage feedback module calculates the intervention priority I according to the risk index R(t) p And push it to parents and schools.

2. The adolescent obesity risk warning system according to claim 1, characterized in that: The electronic scale integrates a weighing sensor, a microprocessor, and a wireless communication module, supports a periodic weighing function, and users can automatically record weight data W(t) at preset time intervals. The data is uploaded to a mobile terminal or a cloud database in real time via Bluetooth or Wi-Fi.

3. The adolescent obesity risk warning system according to claim 1, characterized in that: The physiological data Pd collected by the data acquisition module include weight W(t) and BMI(t) measured by electronic scales, heart rate, sleep duration and number of steps collected by wearable devices, and behavioral data B d Including dietary calorie intake, exercise frequency and screen usage time, all data is stored in time series form and updated regularly.

4. The adolescent obesity risk warning system according to claim 1, characterized in that: The dynamic growth and development correction module uses the following formula to calculate the adjusted obesity risk threshold T adj : T adj =T base +α·ΔG(t)+β·P(t) Among them, T base is the standard BMI threshold based on age and gender, ΔG(t) is the standardized height growth rate, P(t) is the BMI change trend predicted in the next 6-12 months based on the regular weighing data W(t) of the electronic scale, α and β are correction coefficients, and P(t) is calculated using the short-term memory network model.

5. The adolescent obesity risk warning system according to claim 1, characterized in that: The behavior-physiology linkage analysis module calculates the behavior data B through the Bayesian causal network d With physiological data P d The conditional probability P(P d |B d ): Among them, P d Represents physiological data, B d represents behavioral data, P(P d |B d )、P(B d |P d )、P(P d ) and P(B d ) is a probability value used to identify the causal relationship between behavior and weight change.

6. The adolescent obesity risk warning system according to claim 1, characterized in that: The edge computing warning module runs a lightweight risk assessment algorithm on the microprocessor of the electronic scale to calculate the real-time risk index R(t): R(t)=w1·BMI(t)+w2·ΔW(t)+w3·E(t) Among them, BMI(t) is the current BMI, ΔW(t) is the weight change rate recorded by regular weighing on an electronic scale, E(t) is the exercise index, w1, w2, and w3 are prediction weights, and R(t) is the dimensionless risk index. When R(t) exceeds the preset threshold, a local warning is triggered.

7. The adolescent obesity risk warning system according to claim 1, characterized in that: The home-school linkage feedback module supports multi-user authority management and calculates the intervention priority I by the following formula p : I p =γ·R(t)+δ·F(t) Where R(t) is the risk index, F(t) is the implementation feasibility score of family and school feedback, γ and δ are dimensionless weighting coefficients, and I p is the dimensionless priority score.

8. The adolescent obesity risk warning system according to claim 4, characterized in that: The long short-term memory network model is based on the periodic weighing data W(t) of the electronic scale for prediction, and the formula is: P(t+1)=σ(W f ·[h(t),x(t)]+b f ) Among them, P(t+1) is the predicted BMI value of the next time step, h(t) is the hidden state of the previous time step, x(t) is the normalized weight data of the current input, and W f and b f are the weight matrix and bias respectively, and σ is the dimensionless activation function.

9. The adolescent obesity risk warning system according to claim 5, characterized in that: The behavior-physiology linkage analysis module combines group contrastive learning to optimize the causal model through the following loss function: Among them, y i is the true causal label, is the predicted probability based on the electronic scale weight data W(t), D group is the difference between individual and group weight distribution, λ is the regularization parameter, and L is the dimensionless loss value.

10. The system according to claim 1, wherein: The electronic scale has a low-power design, built-in voice prompt function and LED display, and supports users to set up regular weighing plans through buttons or mobile applications.

Images