Using Mobile Technology to Construct a Network Medical Health Care System

In this study, a multisensory physiological measurement system was built with wireless transmission technology, using a DSPIC30F4011 as the master control center and equipped with physiological signal acquisition modules such as an electrocardiogram module, blood pressure module, blood oxygen concentration module, and respiratory rate module. The physiological data were transmitted wirelessly to Android-based mobile applications via the TCP/IP or Bluetooth serial ports of Wi-Fi. The Android applications displayed the acquired physiological signals in real time and performed a preliminary abnormity diagnosis based on the measured physiological data and built-in index diagnostic data provided by doctors, such as blood oxygen concentration, systolic pressure, and diastolic pressure. In addition, in this study, the R waves, which are the highest peaks of the PQRST waves in electrocardiograms, were analyzed and detected using the heart rate variability (HRV) for time-frequency analysis and calculation of the RR interval time series. In this study, the heart rate data in the same age group were collected, and the optimal value of the standard deviation of normal to normal (SDNN) during the time domain of a normal heartbeat was found using the particle swarm optimization (PSO) algorithm and set as the risk level of the SDNN time domain analysis. A spectral analysis on the activity of the autonomic nervous system (ANS) was performed and the preliminary analysis results were displayed on the Android handheld devices for comprehensive physiological data analysis and HRV time-frequency analysis. Healthcare needs are distributed at all levels, therefore user-friendly software interfaces have been written to meet the healthcare needs at all ages.


Introduction
In recent years, technology has developed rapidly. The robotics promoted by artificial intelligence (AI) research and the development and Germany's Industry 4.0 driven by IoT have made technology gradually become the center of global economic growth. In such a highly competitive industrial situation, because enterprises are under pressures such as the demand for product innovation, the efficiency of solving 2 Literature Review

Internet of Thing
The concept of IoT originated from the book The Road Ahead by Bill Gates in 1995. The technology was not taken seriously at the time, however, because it was still limited by the development of equipment and wireless networks. The information mentioned by Bill Gates in The Road Ahead has been verified repeatedly. The book deeply analyzes the integration of information technology and business strategy in future business and elaborates how to have predictive enterprises [3,4].
The Internet connects many subsystems, such as sensor systems, radio frequency identification (RFID) systems [5], wireless communication systems, and global positioning systems, to build an internetwork with the functions of sensing, communication, and monitoring. Machine to machine (M2M) communication is established to transmit the sensed physical quantity to the management or monitoring system in the form of data and texts for further analysis and processing [6,7]. As defined by the European Telecommunications Standards Institute (ETSI) [8], according to different working procedures and contents, IoT can be further subdivided into the layers: the perception layer, the network layer, and the application layer. The IoT architecture is shown in Fig. 2. In addition, the gateway connecting the network layer is composed of a Bluetooth module, a Wi-Fi module, and a passive RFID label. A communications chip is the channel connecting the perception layer and the network layer and is used to transmit the information obtained by the sensing devices to the network layer for the next stage of IoT [9]. Through the above two technologies of sensing and networking, human beings can enjoy applications related to any object of interest around them at any time and in any place by any network access method. IoT has been applied in fields such as smart first aid kits for medical care [10,11], management and environmental monitoring for power systems [12], smart learning services for educational resources [13], and smart green buildings for construction projects [14]. Its application range covers food, clothing, residences, transportation, education, and entertainment, as shown in Fig. 3. The application layer also includes an application support layer, which mainly acts as a data and information analysis and processing platform to connect various networks and provide application services. The application support layer can be divided into three types: a cloud computing platform, a service support platform, and an information disclosure platform [15][16][17].

Amazon EC2
Amazon Elastic Compute Cloud (Amazon EC2), with its architecture based on Amazon Web Services (AWS), is a cloud computing platform created by Amazon that provides numerous remote web services.  The Amazon S3 architecture is also on this platform. The architecture of the Amazon EC2 and AWS service integration is shown in Fig. 4.
AWS provides various computing and web services that allow users to configure a virtual host server, set a firewall, configure network access, cut, and get exclusive routing IP addresses, and users can flexibly expand and increase (for example, network traffic, hard disk capacity, and CPU processing speed) according to their own needs at any time [18][19][20].

Medical Electronics
Aging and rapid increases in medical expenses have driven the growth of medical electronics. According to the estimation by Evaluate Pharma, the medical device market grew to about US$ 780 billion in 2016, mainly because of the increase in demand for chronic disease detection and monitoring. In 2016, the United States Congress announced approval of the 21st Century Cure Act, which is a precision medical policy. Driven by the development of medical big data analysis platforms and telemedicine, as well as portable mobile medical development, it has grown at a compound annual rate of 6.2% and is expected to grow to US$995 billion by 2020. The details are shown in Fig. 5.

Electrocardiograms (ECG)
By placing electrodes on specific parts of the body and using the non-combination of electrodes, a voltage record from 12 different angles can be produced, which is called the ECG lead. A 12-lead electrocardiogram is generally used, which includes three standard leads, three augmented leads, and six chest leads. The standard leads and augmented leads represent the frontal view of the heart, and the chest leads represent the horizontal view of the heart [21,22].

Standard Leads
A standard lead, also known as a bipolar lead, contains two electrodes (a negative electrode and a positive electrode). The records are comprised of Lead I, Lead II, and Lead III, and the potential differences between each two limbs are recorded, as shown in Eqs. (1)- (4). It is also known as Einthoven's Triangle, and the body's bipolar leads are shown in Figs. 6a-6c [23,24].

Augmented Leads
Augmented leads continue to use the electrodes of the standard leads (left hand, right hand and left foot) but use different combinations to produce potential differences. An augmented lead is known as a unipolar lead, as only one electrode (left hand, right hand, and left foot) is used to record voltages [25,26]. The human   7).

Blood Oxygen Concentration
Red and infrared light emitting diodes and a light sensing transistor are put above and below a finger, respectively. As shown in Fig. 8a, the light emitting diode is above the finger and the photosensitive transistor is below, to receive the intensity of the light passing through the finger. The red and infrared light emitting diodes are packaged in the same sensor and put below a finger, as shown in Fig. 8b, and a photosensitive transistor is used to sense the intensity of the red and infrared light reflected from the blood vessel.

Particle Swarm Optimization (PSO)
Particle swarm optimization (PSO) is a well-known swarm intelligence algorithm used to solve and optimize problems. PSO was proposed by J. Kennedy and R. C. Eberhart who, inspired by the foraging behavior of birds, based their study on bird flocks, fish schools, and human social systems to prove that the information exchange among individuals in a group facilitates the evolution of an entire group, which is the source of the core concept of the PSO algorithm [27]. The PSO flow chart is shown in Fig. 9.' The velocities and locations of all particles are updated using Eqs. (8) and (9).  Assume that a group comprised of M particles travels at a certain speed in a D-dimensional space; the state of particle i at time t will be set as below: ; v max , are the minimum and maximum velocities, respectively; Particle best location: v t id : Velocity of particle i at time t x t id : Location of particle i at time t ω: Inertia weight c 1 , c 2 : Learning factor r 1 , r 2 : Random number p id : The best location of particle i so far p gd : The best locations of all particles so far In the updated particle Eq. (10), the inertia weight and learning factor must be set. In this study, the inertia weight ω was set by referring to Wang et al. [28] After repeated experiments, the weight setting method linearly decreasing from 0.9 to 0.4 was proposed in 2000. The linear decrease is shown in Eq. (10).  [29,30]. The results show that the PSO algorithm can accurately forecast the performance and effectively improve the accuracy of optimization [31,32].

Research Method and Analysis
The Fig. 10 shows the block diagram of the system completed in this study.

Electrocardiogram Module
In this study, the main equipment included a physiological signal acquisition module, a wireless transmission module, data display equipment, back-end data storage, and operational analysis, as shown in Fig. 11. The development board used in this study was HUIOT-Basic, a product researched and developed by HUAYU, as shown in Fig. 12. The electrocardiogram module used in this study is shown in the Fig. 13.
Three leads were used as the inductions in this study, respectively labelled Lead I, Lead II, and Lead III, mainly because the small module reduced the electrode patches and time at measurement. In addition, the high-accuracy and high-reliability computation provided electrocardiogram data in real time. The block diagram of the electrocardiogram module circuit flow is shown in Fig. 14.

Blood Pressure Module
The blood pressure module used in this study, shown in Fig. 15, measured blood pressure by the resonance method, which is the most commonly-used noninvasive blood pressure measurement method. The advantages of this method were its simplicity and convenience. The block diagram of the blood pressure module circuit flow is shown in Fig. 16.

Respiratory Rate Module
The respiratory rate module used in this study operated based on the variation of the impedance pneumography, as shown in Fig. 17. For air can be considered as a direct current insulator; the human lungs are like variable capacitors, and the capacitance increases when inhaling and decreases when exhaling. The capacitance changes with inhalation and exhalation, and the tissue outside the chest is considered as a resistance. The block diagram of the respiratory rate module circuit flow is shown in Fig. 18.

Blood Oxygen Concentration Module
The blood oxygen concentration module used in this study, as shown in Fig. 19, performed measurements by the penetrative method. The block diagram of the blood oxygen concentration module circuit flow is shown in Fig. 20.

Experimental Simulation and Analysis
This study included a physiological data sensing module, a cloud server, and a smartphone application. The physiological sensing module used HUIOT-Basic centered on dsPIC30F4011 as the master control and   was equipped with four different physiological data sensing modules, including a heart rate module, a blood pressure module, a respiratory rate module, and a blood oxygen concentration module, which were used to acquire human data in a wired way and then achieve wireless physiological data transmission through Wi-Fi and Bluetooth. The transmission control protocol architecture is shown in Fig. 21.
In this study, an Android Studio handheld device was mainly used to display the physiological data for measurement and analysis. The HUIOT-Basic circuit board of HUAYU was used as the master control device and was equipped with a physiological sensing module for physiological data acquisition. In order to clearly observe changes in the R convex waves of the electrocardiogram, Lead II was used for the electrocardiogram measurement to compute the changes in the R-R interval over a five-minute measurement period and to conduct the time frequency analysis of HRV theory. Fig. 22 shows the actual heart rate measurement. The experiment proved that the pressure amplitude in the tourniquet was the maximum when the pressure in the tourniquet was equal to the mean pressure. The Fig. 23 shows the actual blood pressure measurement.  In this study, a blood oxygen finger sensor was worn on the finger of the subject to acquire the blood oxygen concentration. There is the red and infrared light with different wavelengths in the blood oxygen finger sensor. The blood oxygen concentration module was mainly used to further compute the accuracy of the measured values and get accurate blood oxygen concentration values. The Fig. 24 shows the actual blood oxygen concentration measurement.

Heart Rate Analysis of HRV Theory
In this experiment, the HRV was obtained based on the heart rate measured by the electrocardiogram, and programs were used to obtain the HRV of the measured R waves of the electrocardiogram within the measured time series. The HRV flow chart is shown in Fig. 25. The R-R interval of the heartbeat actually observed by a mobile phone is shown in Fig. 26.

HRV Value Time Domain Analysis
The HRV standard deviation was mainly computed as part of the electrocardiogram time domain analysis. In this study, the main parameters of the index analysis were the SDNN analysis values, and the index values were analyzed and computed by using an algorithm, using the reference Eqs. shown in (11)- (14). The algorithm calculation was performed on the heart rate variability acquired by the electrocardiogram during a short period of time using the smartphone application [33].  Total heart beats computed by the program: THB ðTotal heart beatsÞ ¼ N (11) Time series of mean R-R interval: Mean Standard deviation of NN intervals: SDNN In this study, 200 items of physiological data from each age group were obtained using the Physio Net database, as shown in Fig. 27, in which the green bar represents the range of human values. The SDNN relation equation and PSO algorithm were used to find the optimal SDNN value of the physiological data in HRV, which changes with age [34,35]. The X-axis represents age and the Y-axis represents the range of SDNN analysis data. The optimal value of the physiological data was used to develop the index diagram of the physiological data risk levels, as shown in Fig. 28.

HRV Value Frequency Domain Analysis
The HRV frequency domain analysis was mainly computed using discrete Fourier transform. The HRV frequency domain conversion is shown in Eqs. (13) and (15). The ECG signal is γ(t), and its autocorrelation function is γ(t):  Its spectral power is W(f):

Comparison of Various Measurements
In this study, the blood pressure sensing module and heart rate sensing module in the physiological sensing module platform were compared with the sphygmomanometer (model BC30) developed by Beurer, a German company, as shown in Fig. 29. The mean absolute percentage error (MAPE) was used to determine the error rate of the system, as shown in Eq. (16).
A: Data measured by commercially available instrument B: Data measured by the module of this study C: Data measured by commercially available instruments  The Fig. 30 shows the comparison between this research module and the commercially available blood pressure meter to measure heart rate. The Fig. 31 shows the average absolute error rate of the heart rate. The Fig. 32 shows the comparison between this research module and the commercially available blood pressure meter to measure systolic blood pressure. The Fig. 33 shows the mean absolute error rate of systolic blood pressure. The Fig. 34 shows the comparison between this research module and the commercially available blood pressure meter to measure diastolic blood pressure. The Fig. 35 shows the mean absolute error rate of diastolic blood pressure [36].

Conclusion
According to the statistical analysis, the total number of sudden cardiac deaths is increasing every year. On average, three people have a heart attack every minute and die within an hour, and the success rate of rescue is less than 10%. In fact, it is because most young people think that they are in good health and do not immediately seek medical treatment if they feel uncomfortable, however such bad habits are the main cause of sudden death.
In the networked physiological measurement system of this study, an Android mobile phone application was used as the main physiological value display device, human physiological signals were acquired by a data acquisition module, the values were wirelessly transmitted to the handheld application via Wi-Fi and a Bluetooth module for display, and the handheld device application preliminarily diagnosed the values to advise the users. The handheld application analyzed the risk level of the indicative physiological value determination by doctors and used HRV for in-depth time-frequency analysis of the heart rate values. It could make many adjustments according to the users' physiological conditions, use the PSO algorithm to seek the optimal value of the SDNN time domain analysis according to age, develop a range of risk levels, and store physiological values using the Amazon EC2 platform to provide a reference for the changes in measured values. The proposed system was built to prevent humans from experiencing excessive stress due to overworking and to prevent users from developing cardiovascular and cerebrovascular diseases. The unexpected karoshi is usually caused by the carelessly slow accumulation of body loads.