NFC bracelet service recommendation system based on user behavior analysis

Abstract

The invention discloses an NFC bracelet service recommendation system based on user behavior analysis, and the system comprises a data collection module which is used for collecting health data of a wearer; the storage module is used for storing historical health data; the processing module is used for calculating and recommending a daily recommendation target suitable for the wearer according to the historical health data and the current health data; and the comparison module is used for comparing the daily recommendation target recommended by the intelligent equipment with the currently collected and counted corresponding health data and result, judging whether the actual data of the wearer on the day is up to the standard and judging the completion ratio and difference, and if the judgment result is that the actual data is not up to the standard, reminding the wearer in time through the reminding module. According to the invention, the bracelet carried by a person is utilized, the bracelet is combined with the sensing module and the intelligent module, the emotion data of the user is obtained through analysis according to the behavior data, the actual data of the wearer can be monitored, whether the actual data reach the standard or not is judged, and the wearer is reminded in time when the actual data do not reach the standard.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of a recommendation system, in particular to an NFC bracelet service recommendation system based on user behavior analysis. BACKGROUND

[0002] The bracelet is a kind of jewelry that has been worn since ancient times and is very popular. Nowadays, the bracelet has a single function and is mainly used for decoration.

[0003] With the rise of fitness, with the development of mobile phone APPs, fitness enthusiasts often need to wear multiple wearable accessories, such as pedometers, calorie consumption calculators, heart rate monitors, sleep monitors and the like. Carrying too many things is not only not conducive to exercise, but also not conducive to aesthetics, and it is also very troublesome to operate. The existing bracelet only plays a decorative role.

[0004] With the development of high-tech, smart bracelets have entered the field of vision of people. Many people who like to wear traditional bracelets are attracted by smart bracelets, but it is difficult to make a choice. Therefore, the combination of smart bracelets and traditional bracelets has great market value. SUMMARY

[0005] The application aims to provide an NFC bracelet service recommendation system based on user behavior analysis to solve the problems in the background art.

[0006] To achieve the above-mentioned purpose, the application provides the following technical scheme: The NFC bracelet service recommendation system based on user behavior analysis comprises: A data collection module is used to collect the health data of the wearer, and the health data includes heart rate changes, exercise amount, sleep duration and acceleration. A storage module is used to store historical health data. A processing module is used to calculate and recommend daily recommended targets suitable for the wearer according to the historical health data and current health data. A comparison module is used to compare the daily recommended targets recommended by the smart device with the corresponding health data and results collected and counted at present, to determine whether the actual data of the wearer on the day meets the standard and the completion rate and the difference. If the judgment result of the health data index is not up to standard, the reminding module is used to timely remind.

[0007] Further, the health data further includes sound changes, and the system further comprises an emotion module used to analyze the sound changes, obtain the emotional data of the wearer, send the emotional data to a background server connected with the bracelet for storage, and the wearer obtains the emotional data from the connected background server and displays the emotional data on the display module.

[0008] Furthermore, it also includes the following steps: The system counts and determines whether the frequency of reminders given to the wearer within the system's recommended and preset duration exceeds the previous frequency threshold. If the result is yes, the reminder density is increased. The system counts and determines whether the frequency of reminders given by the wearer within a preset time exceeds the post-frequency threshold. If the result is yes, a warning reminder is sent to the wearer's pre-associated terminal. Wherein, the subsequent frequency threshold is greater than the preceding frequency threshold.

[0009] Furthermore, when the processing module detects that the wearer's actual sleep duration is lower than the corresponding recommended sleep duration, it issues a nap reminder and suggestion, including a suggested nap duration. The suggested nap duration is calculated based on the difference between the recommended sleep duration and the actual sleep duration according to a preset ratio.

[0010] Furthermore, the emotional data is divided into two categories: positive emotions and negative emotions.

[0011] Furthermore, the backend server generates an emotion curve from the emotion data, with time on the horizontal axis and emotion on the vertical axis. Based on the wearer's emotion data, the backend server provides feedback to the wearer on emotion control suggestions.

[0012] Furthermore, the daily recommended goals for smart devices include scientifically recommended health target values ​​based on the average health indicators of the wearer and people of the same age and gender obtained from big data; among them, health indicators include exercise volume and sleep duration. The health data collected and statistically analyzed by the data collection module is used to determine whether the average value has been reached. If the result is not, the reminder module will issue a corresponding reminder and provide a comparison report to the backend server connected to the bracelet, so that the wearer can better understand the gap between themselves and the general population, so as to intervene or adjust in a timely manner.

[0013] Compared with the prior art, the beneficial effects of the present invention are: This invention utilizes a personal, worn bracelet, combined with sensors and a smart module, to collect behavioral data on an individual's physiology, movement, and voice. Based on this behavioral data, it derives the user's emotional data. Since changes in emotional data often reflect changes in a person's physiological state, this invention determines the user's emotions based on this behavioral data and reports the analyzed emotional data to a backend server for storage. Users can access and view their emotional changes at any time. This method not only facilitates data collection but also allows for the collection of a large volume and wide range of data, resulting in more accurate emotional data analysis. This allows the bracelet to not only serve as an ornamental accessory but also to be integrated with a smart module for health monitoring.

[0014] This invention comprehensively evaluates the wearer's historical health data and current health data collected by the data collection module to determine daily health goals tailored to that individual. The wearer's current health data is also a crucial reference during the evaluation, allowing for timely adjustments to daily health goals should the wearer's health change recently. Simultaneously, the bracelet monitors the wearer's actual data and determines whether the goals are met. If goals are not met, the wearer receives timely reminders. If goals are consistently not met, the frequency of reminders is increased, and even family and friends are notified to reduce avoidable physical damage or health risks caused by unhealthy habits, enabling timely intervention. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the NFC bracelet service recommendation system based on user behavior analysis according to the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Example 1

[0017] Please see Figure 1 The present invention provides a technical solution: NFC bracelet service recommendation systems based on user behavior or health analysis include: The data collection module is used to collect the wearer's health data, including heart rate changes, exercise volume, sleep duration, and acceleration. The storage module is used to store historical health data; The processing module calculates and recommends daily goals suitable for the wearer based on historical and current health data. The comparison module is used to compare the daily recommended goals recommended by the smart device with the currently collected and statistically analyzed health data and results to determine whether the wearer's actual data for the day meets the goals and the completion rate and difference. If the health data indicators are judged to be unsatisfactory, a timely reminder will be issued through the reminder module.

[0018] It should be noted that the current health data collected by the data collection module becomes historical health data after it is overwritten by the latest health data. The data collection module can collect relevant health data through built-in sensors or download various data uploaded by users via the network.

[0019] For example, a preliminary health indicator can be obtained based on the wearer's physical condition and historical health data. Then, adjustments can be made based on current health data. For instance, if a swimmer or hiker shows frequent heart health abnormalities recently, the recommended exercise volume for that wearer needs to be significantly reduced, and the recommended sleep time increased accordingly to help them recover their health as quickly as possible. Another example is obtaining the average values ​​of various health parameters for peers through big data analysis. The recommended indicators are then adjusted upwards or downwards based on the average values ​​of several highly correlated health parameters. For instance, exercise recommendations involve several health parameters such as heart rate. If the heart rate deviates significantly from the average, the exercise volume is reduced accordingly. Simultaneously, different weight values ​​are assigned to each health parameter based on its correlation with the recommended indicator. Weighting all relevant health parameters according to their respective weights yields the adjusted value for the recommended indicator. Adjusting the preliminary health indicator based on this adjusted value results in a personalized daily health goal for the wearer. Of course, other existing methods can also be used to obtain daily health goals, and the above approach is not the only one.

[0020] Specifically, the emotional data is divided into two categories: positive emotions and negative emotions.

[0021] Specifically, the health data also includes voice changes, and the system also includes an emotion module for analyzing voice changes, obtaining the wearer's emotion data, and sending the emotion data to a backend server connected to the bracelet for storage. The wearer obtains their emotion data from the connected backend server and displays it on the display module.

[0022] The bracelet also includes a communication unit, which includes an NFC module, a BLE module, and an interface module. The NFC and BLE modules are used to connect with smart devices such as smartphones and tablets, while the interface module is used to connect to the backend server network.

[0023] Collecting voice change data can reveal whether the wearer's voice is loud or deep. Different positive and negative emotions may manifest in different behaviors and physiological changes, so different data collection methods are needed, including: For positive emotions such as happiness, optimism, satisfaction, and enthusiasm, the behavior is "laughter," and the physiological changes can be reflected in changes in facial expressions. The data collection method can be to take a picture of the wearer's head using the image unit of the emotion module. For excited positive emotions, the behavior is "laughter, increased heart rate, and increased blood pressure," and the physiological changes can be reflected in changes in facial expressions, increased heart rate, and increased blood pressure. The data collection method can be to take a picture of the wearer's head and collect heart rate change data. For negative emotions such as sadness, grievance, and heartbreak, the behavioral manifestation is "solemn expression," and physiological changes can be reflected in facial expression changes. The data collection method can be taking a picture of the wearer's head. For negative emotions such as tension and fear, the behavioral manifestation is "solemn expression, increased heart rate, and increased blood pressure," and physiological changes can be reflected in facial expression changes, increased heart rate, and increased blood pressure. The data collection method can be taking a picture of the wearer's head and collecting heart rate change data. For negative emotions such as anger, the behavioral manifestation is "frowning, flushed face, raised voice, increased heart rate, and increased blood pressure," and physiological changes can be reflected in facial expression changes, increased heart rate, and increased blood pressure. The data collection method can be taking a picture of the wearer's head, collecting heart rate change data, and obtaining sound change data from the sound sensing unit of the emotion module.

[0024] The method of analyzing facial expression changes by photographing the wearer's face has many existing implementations, which will not be elaborated here. Analyzing the facial expression changes on the panel can determine whether the wearer is smiling, has a serious expression, or is frowning, thus indicating whether the wearer is experiencing a positive or negative emotion. Similarly, the heart rate data collected by the data collection module can determine whether the wearer's heart rate is fast, slow, or normal, thus indicating whether the wearer is experiencing a positive or negative emotion. Furthermore, the collected voice change data can determine whether the wearer's voice is deep or high-pitched, thus indicating whether the wearer is experiencing a positive or negative emotion.

[0025] The backend server generates an emotion curve from the emotion data, with time on the horizontal axis and emotion on the vertical axis. Based on the wearer's emotion data, the backend server provides feedback to the wearer on emotion control suggestions.

[0026] Furthermore, emotions can be scaled, for example, positive emotions can be rated up to 3 points and negative emotions up to -3 points. This will create a curve of emotional data that fluctuates over time. Wearers can access this emotional curve through smart devices, allowing them to intuitively understand their emotional changes on the bracelet itself or on a smart device (such as a mobile phone or tablet) paired with it. They can then decide whether to make adjustments based on these changes. In addition, wearers can adjust the timeline unit during the viewing process, such as by day, week, or month, to easily understand their emotional changes in the near, medium, and long term.

[0027] The scores for positive and negative emotions can be determined based on the wearer's degree of "smiling," "serious expression," or "frowning," as well as heart rate and volume. For example, if a wearer's facial expression is "smiling" and their heart rate is increasing, they are considered to be in a positive emotion. A fixed score for "smiling" is pre-set (preferably, smiling can be pre-divided into several levels, each with a different score, thus determining the score for "smiling" under the wearer's current expression). At the same time, points are added based on the wearer's heart rate, resulting in a total score (maximum 3 points). This gives the wearer's positive emotion score for a certain period of time. If the wearer is "smiling" but their heart rate is not increasing, only the score for "smiling" is given, thus obtaining the positive emotion score. Similarly, negative emotions can be determined using a similar method. Fixed scores for "serious expression" and "frowning" can be pre-set, while different bonuses can be given based on heart rate and volume. The faster the heartbeat, the higher the bonus (i.e., the greater the negative emotion), and the louder the voice, the higher the bonus (i.e., the greater the negative emotion), thus obtaining the total score for negative emotions (maximum 3 points).

[0028] After receiving the wearer's emotional data, the backend server can combine big data analysis to provide feedback on emotion control. For example, if the wearer has recently experienced increased negative emotions, especially at work, the server will suggest taking a trip to relax, communicating with friends to improve the situation, or seeking guidance or help from colleagues or superiors. Because the backend server stores a large amount of the wearer's emotional data, it can analyze the wearer's negative emotions based on this massive amount of data, determining when or where these negative emotions are most pronounced. For instance, if negative emotions are more frequent during work hours and at the workplace, the server can combine this data with other wearers' methods for managing negative emotions at work to provide emotional control suggestions. Ideally, the server should identify the control methods used by other wearers in similar industries and jobs when experiencing similar negative emotions and recommend these methods to the current wearer. This allows the current wearer to better understand and learn from others experiencing similar situations, tailoring their approach to their own work and adjusting their emotional control accordingly.

[0029] Specifically, it also includes the following steps: The system counts and determines whether the frequency of reminders given to the wearer within the system's recommended and preset duration exceeds the previous frequency threshold (10 reminders within a month). If the result is yes, it indicates that the wearer has been failing to meet the standard for a long time, which may be harmful to their health, so the reminder frequency is increased. The system counts and determines whether the frequency of reminders given to the wearer within a preset time period exceeds a later frequency threshold (20 reminders within a month), where the later frequency threshold is greater than the earlier frequency threshold. If the result is yes, it indicates that the wearer has seriously failed to meet the standards or has a high probability of developing a disease, and an alert is sent to the wearer's pre-associated terminal; supervision and intervention are then carried out by external parties such as relatives, superiors, or service personnel.

[0030] Specifically, when the processing module detects that the wearer's actual sleep duration is lower than the corresponding recommended sleep duration, it issues a nap reminder and suggestion, including a suggested nap duration. The suggested nap duration is calculated based on the difference between the recommended sleep duration and the actual sleep duration according to a preset ratio.

[0031] Specifically, the daily recommended goals for smart devices include scientifically recommended health target values ​​based on the average health indicators of the wearer and people of the same age and gender obtained from big data. Among these health indicators are exercise volume and sleep duration. If a wearer is in a special position, such as a delivery rider or working at heights, poor sleep quality may affect safety and may make it unsuitable for the job. The device will then immediately remind the wearer's family members and workplace to adjust their schedules or make necessary adjustments.

[0032] For example, by weighting various indicators and adjusting based on the average sleep duration of peers, a suitable recommended sleep duration for the day can be obtained, such as requiring 8 hours of sleep. If the bracelet detects that the wearer has only slept 6 hours the next day, it will issue a sleep reminder. In this case, since 2 hours of sleep have been missed, a 0.5-hour or 1-hour nap can be recommended according to a predetermined ratio to avoid affecting health.

[0033] The health data collected and statistically analyzed by the data collection module is used to determine whether the average value has been reached. If the result is not, the reminder module will issue a corresponding reminder and provide a comparison report to the backend server connected to the bracelet, so that the wearer can better understand the gap between themselves and the general population, so as to intervene or adjust in a timely manner.

[0034] Often, wearers are unclear about their daily exercise goals, sleep duration targets, and other indicators to better suit their age and health condition. In such cases, they can refer to the average levels of their peers. For example, they can use big data to calculate the average exercise volume of people of the same age and check whether their own exercise volume meets the standards and the percentage achieved; whether their sleep quality meets the standards and average levels; whether they need to take a nap during the day, and for how long, to meet the body's normal physiological needs; and whether their peers' exercise and sleep indicators are in a healthy state, what potential health risks exist, and what unhealthy lifestyle habits and their harms exist. Example 2

[0035] By collecting acceleration data, it is possible to detect behaviors such as walking, jogging, climbing stairs, and descending stairs based on acceleration. Acceleration data can be collected using an accelerometer.

[0036] The directions of the three axes of the accelerometer are defined as follows: when placed horizontally on a table, the positive direction of the X-axis is horizontal to the right, the positive direction of the Y-axis is horizontal upward, and the positive direction of the Z-axis is perpendicular to the horizontal plane and upward. The coordinate system formed by the three axes conforms to the definition of the right-hand rectangular coordinate system.

[0037] Four behaviors—walking on a flat surface, jogging, climbing stairs, and descending stairs—were analyzed. The arm swing posture and amplitude differed among these behaviors, resulting in significant variations in the generated acceleration signals. All acceleration data are expressed in units of gravitational acceleration (g).

[0038] During data acquisition, irrelevant movements of the arm may cause data interference. Therefore, the data signal needs to be processed during the data preprocessing stage to reduce noise. Using a median filtering algorithm to process the original triaxial signal, setting the value of each point in the data sequence to the median of all points within a certain neighborhood window of that point, can effectively filter out spikes in the data.

[0039] The acquired signal features include time-domain features and frequency-domain features. This embodiment extracts features from the triaxial acceleration data.

[0040] Frequency domain techniques are used to capture the periodic characteristics of sensor signals, and commonly used frequency domain techniques employ Fourier transform. For data lengths of... n vector { x 1 , x 2 , ..., x n The Fourier transform process is achieved through formula (1):

[0041] The frequency ranges of jogging and walking differ significantly, resulting in substantial differences in signal energy. The signal energy is obtained by averaging the sum of squares of the frequency domain coefficients over all sample windows using formula (2):

[0042] Sample entropy was used as a data feature to better distinguish between horizontal walking and climbing stairs. This resulted in a total of 33 features.

[0043] The sample data is normalized to the interval [0, 1] using formula (3), where X ( i , j ) represents the elements in the input sample matrix. X min（j） and X max（j） The first j The minimum and maximum values ​​of the column elements.

[0044]

[0045] A backpropagation (BP) neural network typically consists of three layers: an input layer, hidden layers, and an output layer. Assume the number of feature samples is... n , n ∈ N The number of target categories is m Then the number of input layer nodes of the network can be determined as follows: n Number of output layer nodes m If the input sample x Belongs to the j If the class is specified, then the output vector Y = [ y 1 , y 2 , ..., y m [Middle] j Each component yj The value of is 1, and the values ​​of the other components are all 0.

[0046] The number of nodes in the hidden layer is calculated using formula (4) as a reference. N h This represents the number of hidden layer nodes. N in The number of nodes in the input layer. N out and N class These represent the number of output layer nodes and the number of target classes, respectively. The number of hidden layer nodes obtained using this empirical formula has been shown to keep the number of training iterations near a minimum without making the network too large, making it a preferable option.

[0047]

[0048] The activation functions used include linear functions, sigmoid functions, and tanh functions. The sigmoid and tanh functions map data to the ranges [0, 1] and [-1, 1], respectively. Using a sigmoid activation function in the output layer restricts the network output to a smaller range, while using a linear function allows for arbitrary output values. In this embodiment, the hidden layers use the tanh function as the activation function, while the output layer uses a linear function.

[0049] Based on the standard backpropagation algorithm, a momentum factor is added, and a portion of the previous weight adjustment is superimposed on the new weight adjustment calculated based on the current error, serving as the actual weight adjustment. This method can adjust the weight correction and reduce oscillations. The weight adjustment after adding the momentum factor is calculated using formula (5), where... α The momentum coefficient, η For the learning rate, ▽ E ( n () represents the training error.

[0050]

[0051] Increase when the error is small. η In order to shorten learning time, when η When the value is too large to converge, and Time decrease η The learning rate is adjusted until convergence is achieved. In this embodiment, formula (6) is used to adjust the learning rate.

[0052]

[0053] By employing the momentum factor method, the backpropagation (BP) algorithm can find a better solution; by using an adaptive learning rate adjustment algorithm, the training time of the network can be effectively shortened. Combining the two yields the momentum-adaptive learning rate BP neural network, which can find a better solution while reducing learning time.

[0054] In this embodiment, feature extraction is performed based on the collected triaxial acceleration data, and combined with the momentum-adaptive learning rate BP neural network classification algorithm, four behaviors can be effectively identified: walking, jogging, climbing stairs, and descending stairs.

[0055] The parts of this invention not described herein are prior art.

[0056] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An NFC bracelet service recommendation system based on user behavior analysis, characterized in that, include: The data collection module is used to collect the wearer's health data, including heart rate changes, exercise volume, sleep duration, and acceleration. The storage module is used to store historical health data; The processing module calculates and recommends daily goals suitable for the wearer based on historical and current health data. The comparison module is used to compare the daily recommended goals recommended by the smart device with the currently collected and statistically analyzed health data and results to determine whether the wearer's actual data for the day meets the goals and the completion rate and difference. If the health data indicators are judged to be unsatisfactory, a timely reminder will be issued through the reminder module.

2. The NFC bracelet service recommendation system based on user behavior analysis as described in claim 1, characterized in that, The health data also includes voice changes. The system also includes an emotion module, which is used to analyze voice changes, obtain the wearer's emotion data, and send the emotion data to a backend server connected to the bracelet for storage. The wearer obtains their emotion data from the connected backend server and displays it on the display module.

3. The NFC bracelet service recommendation system based on user behavior analysis as described in claim 1, characterized in that, It also includes the following steps: The system counts and determines whether the frequency of reminders given to the wearer within the system's recommended and preset duration exceeds the previous frequency threshold. If the result is yes, the reminder density is increased. The system counts and determines whether the frequency of reminders given by the wearer within a preset time exceeds the post-frequency threshold. If the result is yes, a warning reminder is sent to the wearer's pre-associated terminal. Wherein, the subsequent frequency threshold is greater than the preceding frequency threshold.

4. The NFC bracelet service recommendation system based on user behavior analysis as described in claim 1, characterized in that, When the processing module detects that the wearer's actual sleep duration is lower than the corresponding recommended sleep duration, it issues a nap reminder and suggestion, including a suggested nap duration. The suggested nap duration is calculated according to a preset ratio based on the difference between the recommended sleep duration and the actual sleep duration.

5. The NFC bracelet service recommendation system based on user behavior analysis as described in claim 2, characterized in that, The emotional data is divided into two categories: positive emotions and negative emotions.

6. The NFC bracelet service recommendation system based on user behavior analysis as described in claim 2 or 5, characterized in that, The backend server generates an emotion curve from the emotion data, with time on the horizontal axis and emotion on the vertical axis. Based on the wearer's emotion data, the backend server provides feedback to the wearer on emotion control suggestions.

7. The NFC bracelet service recommendation system based on user behavior analysis as described in claim 2, characterized in that, The daily recommended goals for smart devices include scientifically recommended health target values ​​based on the average health indicators of the wearer and people of the same age and gender obtained from big data; among them, health indicators include exercise volume and sleep duration. The health data collected and statistically analyzed by the data collection module is used to determine whether the average value has been reached. If the result is not, the reminder module will issue a corresponding reminder and provide a comparison report to the backend server connected to the bracelet, so that the wearer can better understand the gap between themselves and the general population, so as to intervene or adjust in a timely manner.

Images