key: cord-0823080-kitkncwx authors: Fang, Bin; Sun, Fuchun; Quan, Zhou; Liu, Huaping; Shan, Jianhua title: Smart Bracelet System for Temperature Monitoring and Movement Tracking Analysis date: 2021-07-20 journal: J Healthc Eng DOI: 10.1155/2021/8347261 sha: 677efe2bcc88ed6e75b1abcf5c3a36f8cb2c4aa3 doc_id: 823080 cord_uid: kitkncwx In several epidemic diseases, one of the main symptoms exhibited by people is abnormal body temperature. Therefore, monitoring body temperature is crucial for preventing the spread of infectious diseases and facilitating timely responses. This study presents a wearable bracelet that can be used as a temperature monitoring and trajectory analysis system. The temperature sensor in the bracelet can effectively monitor the body temperature of the wearer in the target scene, and the data transmission between the bracelet and the data collector can effectively detect the movement range and trajectory of the wearer. Through these, the whole set can also detect the direct and indirect contact of the wearer in any period of time, which is very helpful for the prevention and control of infectious diseases and the isolation of potentially infected persons. COVID-19 is a new infectious disease that has caused enormous economic losses around the world because of its long incubation period and strong infection capabilities. Effectively isolating the virus and ensuring the safety of human lives have become a top priority. For such extremely infectious diseases, screening potential infectors according to their symptoms would be of great significance for the prevention and control of epidemics. Similar to COVID-19, a main symptom of many infectious diseases, such as influenza, SARS, MERS, plague, and cholera, is fever [1] , and this study showed that half of the COVID-19 patients had a body temperature above 37°C on admission and 78.5% of the study population had a body temperature above 37°C during the onset of the disease [2] . erefore, a system for monitoring body temperatures of individuals would be very useful in the prevention and control of infectious diseases. Many researchers have proposed various methods for detecting human body temperatures. Al-Hossain et al. proposed a contactless syndromic surveillance platform for influenza-like illnesses in hospital waiting areas called FluSense [3] . is system uses thermal imaging technology to detect body temperatures of individuals in hospitals. Fajrin et al. proposed a telemedicine system that collects body temperature data using DS18B20 sensors and then processes the data using an Atmega328 microcontroller before sending the results to mobile phones via Bluetooth [4] . Yang and Zheng proposed the use of wireless low-power thermal sensors for basal body temperature detection based on frequency-modulated telemetry devices. eir devices adopt a temperature compensation mechanism and achieve good results for the measurement of human body temperature [5] . Although these methods have been applied in various scenarios, they still suffer from some significant limitations. For example, Liu et al. demonstrated the limitations of temperature detection methods that use forehead surface temperature [6] . Temperature monitoring in communities and schools around the world is largely performed using point temperature measurement solutions, such as temperature guns, infrared thermometers, and thermal imagers. However, such methods have numerous problems, including crowding, tedious temperature measurement, and statistical limitations; this implies that there are still significant blind spots from the perspective of epidemic control. With the development of sensor and communication technologies in recent years, wearable devices have begun to play an important role in our daily lives. Many researchers have used data collected by various wearable sensors to perform different types of research. Gu et al. used sensor data and Wi-Fi technology to continuously monitor sleep patterns [7] . Vhaduri et al. proposed a method that uses data collected by smartphones and sensors to opportunistically fill gaps in user location traces [8] . As one of the most widely used wearable devices, smart bracelets have become an important part of the daily life of many people. Such bracelets can be used to monitor sleep patterns and heart rates and also to calculate running speeds [9] . Many researchers have performed studies based on the data collected by smart bracelets. Nguyen et al. performed daily wrist activity recognition using smart bracelet data [10] . Heart rate data collected from smart bracelets have been used for emotional analysis [11] and health status monitoring [12] . Smart bracelets also play an important role in gait analysis of the elderly [13] and urination analysis of patients with certain diseases [14] . In this study, we propose a community temperature monitoring system using smart bracelets which can accurately measure body temperatures of wearers. is system requires several data collectors to be installed on campuses or communities, which can serve as base stations to locate wearers. Temperature data of wearers can be rapidly collected, and their locations can be determined by regular reporting of collected data. Using this system, monitoring personnel can rapidly screen people with infectious diseases so that efficient measures for disease prevention and control can be taken. is system can collect large amounts of body temperature and location data every day and only requires users to wear bracelets; thus, it can avoid the aggregation problem while detecting body temperature, which is of great significance for disease prevention and control. Other methods can only screen direct contacts, whereas our system can effectively screen all direct and indirect contacts of potentially infected individuals. COVID-19 is still rapidly spreading worldwide. To control this epidemic effectively, many research institutions and companies have developed their own systems for monitoring human body temperature and tracking human movement. For example, Singapore launched a system called TraceTogether [15] . e working principle of TraceTogether is to exchange short-range Bluetooth signals between mobile phones (devices are required to always have their Bluetooth functions activated for this) to detect other users of TraceTogether within short distances. ese close contact records are then stored on each user's mobile phone; collected data are encrypted. e data are not uploaded through a network; only if users are diagnosed with an infection will the government ask them to provide data collected on their phone so they can track other individuals who have been in close contact with them. e Ministry of Health can then notify those individuals using TraceTogether so that they can monitor their own health and check whether they are exhibiting any symptoms of the infection; this allows them to receive proper guidance and care. Apple and several US federal agencies have launched an app and a website, as a screening tool, called Apple COVID-19 [16] . is app is mainly used to provide guidance to users regarding COVID-19 by the US Center for Disease Control. By using this app, users can understand the risks within their scope of influence so that they can respond in a timely manner. Additionally, general COVID-19 information and proper protection measures are also available on this app. In June 2020, an app called Corona-Warn, supported by the German federal government and German companies SAP and Telekom, was launched [17] . is application uses Bluetooth technology to send signals within a few meters. Users can learn if there are infected individuals nearby through push notifications. For example, if person A is infected, their device sends a warning message to all nearby mobile phones with this app installed through push notifications, thereby warning users of this app. Although these systems and apps can be very useful for epidemic prevention and control, they still pose several limitations. For example, such systems rely entirely on mobile phones, which implies that individuals without smartphones cannot be factored into the analysis. Additionally, all the systems described above use Bluetooth technology, which requires users to keep their Bluetooth feature turned on, which can be a threat to personal privacy. Due to these issues, many users have concerns regarding personal privacy protection when using these systems. ere are many different control methods available for managing temperature monitoring systems. Figure 1 summarizes the issues of various methods. Our system was developed as an attempt to resolve these shortcomings. e proposed system mainly consists of wearable smart bracelets and data collectors. Data collectors are installed at fixed locations. e proposed system can be used in government agencies, schools, and aggregated office spaces. e proposed workflow for a detection and tracking model is presented in Figure 2 . First, body temperatures of smart bracelet wearers are measured. Next, data collectors collect wirelessly transmitted data from the smart bracelets. Finally, data are uploaded to a cloud platform, on which abnormal temperatures and activity ranges are analyzed. is system only uses data collected by the smart bracelets to avoid private data leakage associated with mobile phones. Each bracelet has a unique ID and a corresponding wearer ID. e list of IDs can only be obtained by a system administrator. Additionally, the application scope of the proposed system is limited to designated office or community spaces to avoid leakage of user location information. erefore, even if a smart bracelet is lost, personal information of the original wearer cannot be obtained. Figure 3 , our smart bracelet consists of nine main modules: a liquid crystal display (LCD) module, power management module, charging module, 2.4 G communication module, main control system module, temperature measurement module, data storage module, touch management module, and clock management module. ese modules can provide not only the temperature measurement functionality but also basic smart bracelet functions to wearers. Using radio frequency (RF) technology, a smart bracelet can transmit data to data collection and processing devices. RF technology is a noncontact automatic identification technology. Its basic principle is to use the transmission characteristics of RF signals and electromagnetic coupling for automatic identification of target objects. erefore, when a smart bracelet enters the wireless connection range of a data processing device, it can connect to the data processing device wirelessly and automatically. Additionally, a data processing device can identify smart bracelets of multiple wearers simultaneously. In the application scenario considered in this study, the bracelets are connected via RF technology to data collection devices. Bluetooth can also be used for data transmission in other application scenarios. When a smart bracelet is worn correctly, the temperature measurement module will be in contact with the wearer's body, allowing it to obtain his/her body temperature. en, the acquired data will be sent to the main control chip (main control system module), which transmits the data via the wireless transmission module. Finally, the data are transferred to a data collector for further analysis. Each smart bracelet also has a touch screen (LCD module), which shows the users their body temperature data in real time; furthermore, users can interact with the system through the touch management module. e power system consists of power management and charging modules for optimal power consumption. In addition, the sustainable working time of smart bracelet is one week. e data storage module is used to record daily temperature data and other routine data from users. e clock management module is used to provide users with basic time information. For better detection and acquisition, we imposed some regulations on smart bracelet wearing: (1) It is suggested to wear it on the inside of wrist to improve the measurement accuracy. (2) Please avoid testing within 30 minutes after exercise or bathing. (3) Please avoid wearing too tight or too loose, so as not to affect the test results. (4) is product cannot replace the doctor's diagnosis, only for screening. If there is any abnormality, please see a doctor in time. Collector. In our system, data collectors are connected to smart bracelets via RF technology, which can realize readout operations in the smart bracelets. In the application scenario considered in this study, the wireless connection range is limited to approximately 15 m to ensure data acquisition stability. Figure 4 presents the components of our data collectors. A data collector contains receiving and transmitting antennas outside the shell. ese antennas are used to read signals from the smart bracelets and send results of temperature data analysis to the smart bracelets. As shown in Figure 5 , a data collector can also be used to determine the number of users in a particular area. Additionally, according to differences in signal strengths, the relative positions of the users and data collectors can be roughly estimated. A data collector consists of a temperature sensor that detects external environmental temperature; this is used to process temperature data collected by smart bracelets. e temperature sensor and data processing device are connected to the main component through a circuit and a temperature detector component extends outward. e temperature sensing component is exposed to the external environment so that it can transmit the detected ambient temperature to the data processing device. e temperature sensor is a thermistor; its resistance value changes with changes in temperature. When the temperature increases, its internal resistance decreases; thus, the detected current will increase. By detecting the magnitude of the current, the Lockdown of the city [18] , infectious disease screening Personnel access restrictions, only residents of the community are allowed to enter, human intervention [20] Control personnel by floor Human contact tracing By technical means It is difficult to carry out effective prevention and control without early warning and preparedness. Big data analysis, mobile base station positioning, app data collection, personal reporting. There is no efficient meth method. Mobile app tracking, such as Singapore TraceTogether app [15] , Apple COVID-19 app [16] , and Germany Corona-Warn app [17] . Problems Unconventional means. great economic losses [19] . To some extent, it restricts the freedom of the masses. It is difficult to coordinate between different communities. The economic loss is large. More reasonable and practical technology is needed. There are concerns about disclosing the privacy of citizens. temperature of the surrounding environment can be determined. e data processing device compiles the ambient temperature into an electronic signal that can be transmitted through the Internet of ings card circuit. e data uploaded by the data processing device consists of the compiled ambient temperature and body temperature data, which are transmitted to the background management equipment. Next, temperature compensation calculations are performed in the background management equipment, allowing background management personnel to determine the temperature information of users within the system detection range. For better detection and acquisition, we imposed some regulations on data collector locations: To achieve effective monitoring of human body temperature and save system resources, we set different upload intervals for data collectors in different locations. For example, at a school gate, considering the mobility of the target crowds, the upload interval was 60 s. In classrooms and dormitories, the upload interval was 480 s. Temperature measurements on the wrist fluctuate significantly with external temperature. erefore, a temperature compensation formula was defined as follows: where T B is the body temperature, T E is the environmental temperature, T M is the original measured temperature, and f is the mapping function. In daily temperature measurements, a variety of factors may result in temperature measurement errors. To achieve suitable temperature measurement performance, we performed various operations. First, we collected a portion of the original smart bracelet temperature measurement data, environmental temperature data, and mercury thermometer temperature measurement data. We then calculated the means and variances of the wrist temperature data (measured by smart bracelets) and environmental temperature data; we used the data to generate body temperature data with a normal distribution. Finally, we considered the temperatures measured by a mercury thermometer as accurate values and used the generated data to fit the accurate values using the least-squares method. Based on the randomness and noise in the generated data, the generalization ability of the fitting equation was improved to a certain extent using the generated normal data to fit actual body temperature measurements. where g is the generation function for the normal distribution; T ' E and T ' M are the original temperature data; and a and b are the fitting coefficients, which are equal to 0.32 and 0.82, respectively (Algorithm 1). In addition, we verify the whole system in extreme weather. When the temperature is higher than 50°C and the sunlight is too strong, the temperature rise of the hardware surface will have a negative impact on the measurement accuracy. e bracelet can receive the wireless connection request from the data collector (base station) within the connection range. After the verification of the wireless connection request, the bracelet returns a consent connection response to trigger the base station to establish a wireless connection with the bracelet itself. rough wireless connection, the detection data is sent to the base station to make the base station acquire the detection data, and send the detection data, the connection information between the base station and itself, and the disconnection information to the server, thereby triggering the server to obtain its own activity information at the predetermined address. Among them, the detection data includes the detection temperature, detection time, and its own unique identification; the connection information includes the starting connection time and the unique identification; the disconnection information includes the connection disconnection time and the unique identification. rough these ways, the system can track the crowd activities and find the close contacts quickly and accurately. For the body temperature data and other information collected from wearers, we constructed a data protection system that ensures that only authorized personnel can access personal data, which protects user privacy: (1) Management level: to ensure that the business model of the proposed system functions properly, a confidentiality management organization is established; it consists of full-time personnel to control confidentiality management tools. e confidentiality system and tools are integrated into all levels of the business process to reduce the number of unnecessary links that could lead to leakage. If a leak occurs, it can be located rapidly by considering the responsible party as the target point and the corresponding business process as the line to the target point. (2) Hardware level: the proposed system is capable of monitoring, communicating, office automation, safety detection in electromagnetic environments, and securing the confidentiality of computer networks using human, physical, and technical preventive measures. (3) Technical level: an asymmetric encryption algorithm is used to encrypt data, and cryptography is used to protect data security. Daily maintenance of the system strictly follows the process operation schedule to ensure that information is not leaked to/tampered with by unauthorized users or entities during the processes of storage, usage, and transmission. Additionally, the proposed system can prevent authorized users from tampering with information or the system itself, which ensures the security of data. By using the proposed system, we can collect large amounts of body temperature data and other information daily. On average, hundreds of thousands of pieces of data are generated every day. is volume of data is significant for mining potential relationships using big data tools. In this study, we consider a system deployed in a primary school as an example and all participants were about 2000. Each smart bracelet corresponds to a student, teacher, or school worker and each data collector corresponds to a location. We have shown locations and corresponding IDs in Table 1 for better clarity. As shown in Figure 6 , we consider data from June 1 to June 27. Significant reduction in the amount of uploaded smart bracelet data is seen on weekends and vocational days (Chinese traditional festival called the Dragon Boat Festival); this corresponds to typically observed trend on weekends and other holidays. Figure 7 , the temperature data calculated by our system based on environmental temperature and body surface temperature data are closer to the values measured with a mercury thermometer than those obtained by other methods; this implies that our system has relatively higher accuracy of body temperature measurement. is work proposed that changes in oral/rectal and axillary temperatures correlated significantly (p < 0.0001) [21] . Considering that oral/rectal temperature measurement may be more likely to cause virus infection and in order to demonstrate the effectiveness of our measurement method, we consider armpit temperature values measured by a mercury thermometer as reference values and compare the values obtained by our method to those obtained by other methods. We obtain the body temperature of these methods by taking three different temperature measurement methods for the same patient in the hospital. In addition, there are some ways to make sure the data reliability. (1) e bracelets are worn all the time when patients are in hospital. (2) When the hospital uses its own equipment to measure the user's temperature, use the bracelet host with the same number as the bracelet to measure the user's armpit temperature, and make a record. (3) e number of measurements should be as many as possible, with an average of no less than 5 times a day. As shown in Table 2 , the mean absolute error (MAE) and mean-squared error (MSE) for our measurement method are smaller than those of some other measurement methods. Because the example system is deployed in a school, only few teachers or staff will upload data during weekends. For detailed analysis, we considered a sample of smart bracelet data. Figure 8 presents the curves of body temperature and surface temperature for a particular user during a working day. e proposed system can be used to monitor a user's temperature using these data. When a user's body temperature is abnormal, the system will query the smart bracelet number and locate the user based on the data collector that uploaded the corresponding data. (1) Get the temperature measured by the bracelet T M (2) Obtain the ambient temperature T E measured with the sensor in the data collector (3) Repeat experiments to obtain multiple measured temperature values and the corresponding real temperature values (4) Based on the randomness and noise in the generated data, the generalization ability of the fitting equation was improved to a certain extent using the generated normal data to fit actual body temperature measurements (5) For T E and T M , the least-square method is used to solve the coefficient fitting the real temperature T B (6) When the coefficient is determined, the algorithm is used to test and verify in the real environment ALGORITHM 1: Pseudocode. 4.2. Crowd Distribution Analysis. By using the proposed system, we can determine the number of people in a certain classroom or entrance. Figure 9 presents a schematic of the system statistics for the number of people in a room. Because the collection range of each data collector is one room or entrance, the number of people in a particular room can be One of the entrances One of the data collectors installed in the room No contact Figure 11 : Contact tracing example. In our system, users can be traced to specific locations. Because data collectors only collect smart bracelet data within a small range, users who have been in contact with other infected individuals can be identified. Figure 10 presents the workflow of the movement and contact tracing system. First, we determine which data collectors were used by the target smart bracelet to upload data. We can then query the locations of these data collectors to determine the movement of the user. Finally, by querying the IDs of the bracelets that a base station has connected to within a certain time range and querying the corresponding personnel information, we can identify anyone who has had direct/indirect contact with an infected user. For a more detailed explanation, we consider bracelet No. 00010221 as an example. On June 9, this smart bracelet uploaded its data to the following data collectors: No. 1537 , No. 1487 , No. 1574 , No. 1456 , No. 1365 , No. 1540 , No. 1564 , and No. 1405 . From Table 1 , we can determine that the wearer of this bracelet visited the following locations in order: class 2, grade 2; dean's office; science classroom; class 1, grade 3; class 2, grade 6; teacher's office; class 2, grade 4; and class 1, grade 2. If we want to identify this user's direct or indirect contacts, we only need to query the IDs of the bracelets that uploaded data to these data collectors on the same day. In Figure 11 , each line represents a user's tracking record. Assuming that the red line represents an infected patient, direct or indirect contacts can be identified by determining the intersection points of the tracking records of this user and other users. If a connection between two users does not exist, it means that they did not come in contact with each other. In this study, we presented a group temperature measurement and contact tracing system using wearable smart bracelets. Compared to other methods/systems, our proposed system has lower deployment cost, is easier to operate, and can better protect user privacy. Compared to some systems that rely on Bluetooth technology, which can only screen individuals who have had direct contact with a person, our system can screen all potential contacts more comprehensively. e proposed system can collect user temperature data and location data in real time through smart bracelets. Because the bracelets do not store personal information, the proposed system avoids leakage of personal information in case the bracelet is lost. We used temperature sensors in the bracelets to measure relative body temperature and temperature sensors in data collectors to measure relative environmental temperature. We then combined these measurements to evaluate true body temperature, which significantly improves the measurement accuracy. is system also has some problems to be solved; for example, the disinfection of the surface of the bracelet may increase the workload, and it is only suitable for small-scale community use at present. e future research direction is to expand the application scope of the whole system and improve the universality of the whole system for different application scenarios. Infectious disease, fever, and the immune response Body temperature correlates with mortality in COVID-19 patients FluSense: a contactless syndromic 397 surveillance platform for influenza-like illness in hospital waiting areas Body temperature monitoring based on telemedicine Wireless low-power integrated basal-body-temperature detection systems using teeth antennas in the MedRadio band Limitations of forehead infrared body temperature detection for fever screening for severe acute respiratory syndrome Sleepy: wireless channel data driven sleep monitoring via commodity WiFi devices Opportunistic discovery of personal places using multi-source sensor data A survey of wearable devices and challenges Daily wrist activity classification using a smart band Wearable emotion recognition using heart rate data from a smart bracelet WiP] sentiment analysis electronic healthcare system based on heart rate monitoring smart bracelet Application of smart bracelet to monitor frailty-related gait parameters of older Chinese adults: a preliminary study Development of personalized urination recognition technology using smart bands A singapore government agency website Coronavirus disease (COVID 19) screening tool Corona-Warn app open source project Modelling the effect of lockdown on COVID-19 pandemic in 22 countries and cities Spatial spillover effects of mega-city lockdown due to COVID-19 outbreak Community interventions and the epidemic prevention potential How reliable is axillary temperature measurement? is work was supported by the National Natural Science Foundation of China (Grant no. 91848206) and Beijing Science and Technology Project (Grant no. Z191100008019008). is work was also supported by the Zhejiang Jinkai Company. In addition to COVID-19, this system has a certain role in the prevention of other infectious diseases that can cause abnormal body temperature, like plague, cholera, SARS, MERS, poliomyelitis, and some other infectious diseases. For the body temperature data and other information collected from wearers, the authors constructed a data protection system that ensures that only authorized personnel can access personal data, which protects user privacy. e authors declare that there are no conflicts of interest regarding the publication of this paper.