HEALTH STATE MONITORING SYSTEM DESIGN A Project Presented [PDF]

periods of time. The sensors for vital data acquisition are non-invasive and do not require a professional health care p

0 downloads 3 Views 2MB Size

Recommend Stories


Review: Health Monitoring System
Pretending to not be afraid is as good as actually not being afraid. David Letterman

Monitoring a Livelihoods Project
Where there is ruin, there is hope for a treasure. Rumi

Project Monitoring
If you want to go quickly, go alone. If you want to go far, go together. African proverb

Project Monitoring
What you seek is seeking you. Rumi

Design Of Underground Coal Mine Monitoring System
Come let us be friends for once. Let us make life easy on us. Let us be loved ones and lovers. The earth

Design and Development of Patient Monitoring System
Don't ruin a good today by thinking about a bad yesterday. Let it go. Anonymous

Design of Simple Landslide Monitoring System
You can never cross the ocean unless you have the courage to lose sight of the shore. Andrè Gide

Soldier Monitoring and Health Indication System
The beauty of a living thing is not the atoms that go into it, but the way those atoms are put together.

autonomous intravenous infusion system & health monitoring
Don't fear change. The surprise is the only way to new discoveries. Be playful! Gordana Biernat

Mobile Phone Sensing System for Health Monitoring
Just as there is no loss of basic energy in the universe, so no thought or action is without its effects,

Idea Transcript


HEALTH STATE MONITORING SYSTEM DESIGN

A Project

Presented to the faculty of the Department of Electrical and Electronic Engineering California State University, Sacramento

Submitted in partial satisfaction of the requirements for the degree of

MASTER OF SCIENCE

in

Electrical and Electronic Engineering

by Viorel Rotar FALL 2012

HEALTH STATE MONITORING SYSTEM DESIGN

A Project

by

Viorel Rotar

Approved by: __________________________________, Committee Chair Dr. Warren D. Smith

__________________________________, Second Reader Koullis Pitsillides

____________________________ Date

ii

Student: Viorel Rotar

I certify that this student has met the requirements for format contained in the University format manual, and that this Project is suitable for shelving in the Library and credit is to be awarded for the Project.

__________________________, Graduate Coordinator, __________________ Preetham B. Kumar Date

Department of Electrical and Electronic Engineering

iii

Abstract of HEALTH STATE MONITORING SYSTEM DESIGN

by Viorel Rotar

The health of hospitalized patients often deteriorates, because the available medical staff is unaware of the deterioration. The condition of normally healthy people under stressful situations can deteriorate, because they are not aware that they have reached their physical limits. Infants can die in their “Sleep” because of Sudden Infant Death Syndrome (SIDS). Death could be prevented by waking the child, but the parents are unaware that the infant experiences lack of oxygen. People with epilepsy can die because a seizure was not detected in a timely manner. In severe epileptic episodes, death from airway constriction can occur in less than 3-5 minutes unless help is immediately provided. Therefore, a Health State Monitoring System (HSMS), able to detect and alarm when a health abnormality develops, is desirable. The HSMS should be broadly useful from being used by professional health-care providers to personal in home use. It should be easy to use by first responders, military personnel, firefighters, athletes, and parents monitoring infants.

iv

The HSMS consists of a wearable device with sensor and communication circuitry, a personal computer (PC) based graphical user interface (GUI), and a basestation interface between the wireless sensor and the PC. The wearable device incorporates sensors that can provide a quick general assessment of a person’s cardiorespiratory status, temperature, and level of physical activity. The cardio-respiratory status is assessed by a pulse oximeter, which gives the percent oxygen saturation in the blood. Heart rate is derived from the pulsatile waveform of the pulse oximeter which corresponds to the cardiac cycle. Skin temperature is measured with a digital temperature sensor. A micro Secure Digital (microSD) card is used to store raw data for extended periods of time. The sensors for vital data acquisition are non-invasive and do not require a professional health care provider for attachment to the body. The system is configurable to accommodate different applications. For continuous patient monitoring, the system can be configured to wirelessly transmit collected data to a computer. These data then can be sent to a physician, who then reviews the data and intervenes when necessary. For applications where a computer is out of wireless communications range, like when monitoring the effect of stressful situation on health of first responders, the data are saved on the microSD card for later review. For applications where just an alarm is necessary to indicate a breathing or heart rate abnormality, such as monitoring of infants for the detection of SIDS, for example, the system can be configured to activate the light and/or sound alarm.

v

In this project, a small wearable wireless sensor was designed and built that incorporates multiple sensors, and all the sensors operated properly in a laboratory setting. The sensors that are less susceptible to motion artifacts, such as temperature, were easier to implement. The pulse oximeter performed well under different light conditions, but data were disrupted or incorrect as a result of motion artifacts. Development of software algorithms to reduce the effects of motion artifacts is desirable. The pulse oximeter signal was obtained easily when a finger was placed over the sensor, but the sensor was not tested at other locations on the body. The wireless communication was stable, with a range suitable for hospital and home use. The necessity for different component values for the transceiver, if there is a need to change the frequency band to comply with local radio-communication rules, is a drawback.

Approved by: __________________________________, Committee Chair Dr. Warren D. Smith

____________________________ Date

vi

ACKNOWLEDGEMENTS It is a pleasure to acknowledge the assistance received during the development of the Health State Monitoring System from Dr. Warren D. Smith, who is committee chair for the project, and Koullis Pitsillides, a biomedical engineer from Endosomatic Systems.

vii

TABLE OF CONTENTS Page Acknowledgements ........................................................................................................... vii List of Figures .................................................................................................................... xi Chapter 1.

INTRODUCTION ........................................................................................................1

2.

BACKGROUND ..........................................................................................................5 2.1.

2.2.

2.3.

Temperature Monitoring ....................................................................................5 2.1.1.

Temperature Monitoring Importance ................................................. 5

2.1.2.

Temperature Measurement Principle ................................................. 6

Activity Monitoring ...........................................................................................8 2.2.1.

Activity Measurement Importance ..................................................... 8

2.2.2.

Activity Measurement Principle ......................................................... 9

Cardio-Respiratory Assessment .......................................................................10 2.3.1.

Pulse Oximeter Importance .............................................................. 10

2.3.2.

Principle of Pulse Oximeter Measurements ..................................... 11

2.3.3.

Optical Properties of the Tissue ....................................................... 12

2.3.4.

Transmission and Reflectance Modes of Pulse Oximeter Measurements .................................................................................. 18

2.3.5. 3.

Interference to the Pulse Oximeter Signal ........................................ 19

HEALTH STATE MONITOR CIRCUITRY ............................................................21 viii

3.1.

Health State Monitor Circuitry Overview........................................................21

3.2.

Sensor Circuitry ...............................................................................................23

3.3.

3.4. 4.

3.2.1.

Temperature Monitoring Circuitry ................................................... 24

3.2.2.

Activity Monitoring Circuitry .......................................................... 25

3.2.3.

Pulse Oximeter Circuitry .................................................................. 26

Supporting Circuitry ........................................................................................31 3.3.1.

Power Supply Circuitry .................................................................... 31

3.3.2.

Microcontroller (MCU) Circuitry..................................................... 32

3.3.3.

Radio Communication Circuitry ...................................................... 35

3.3.4.

MicroSD Card Circuitry ................................................................... 36

3.3.5.

Alarm Circuitry ................................................................................ 37

Base-Station Circuitry......................................................................................38

WIRELESS SENSOR SOFTWARE ..........................................................................39 4.1.

Microcontroller Main Routine .........................................................................39

4.2.

Temperature Acquisition .................................................................................42

4.3.

Accelerometer Acquisition ..............................................................................43

4.4.

Pulse Oximeter Acquisition .............................................................................44 4.4.1.

Pulse Oximeter Software Overview ................................................. 44

4.4.2.

Controlling the LEDs and Taking Data Samples ............................. 46

4.4.3.

Red and Infrared (IR) Signal Processing .......................................... 49

4.4.4.

Calculating the Heartbeats ................................................................ 52 ix

4.4.5.

5.

Calculating the Percent Oxygen Saturation in Blood ....................... 55

4.1.

MicroSD Card Software ..................................................................................58

4.2.

Wireless Communication .................................................................................61

GRAPHICAL USER INTERFACE (GUI) ................................................................63 5.1.

Initializing a Wireless Sensor ..........................................................................63

5.2.

Controlling a Wireless Sensor .........................................................................63

5.3.

Displaying Sensor Data....................................................................................66

5.4.

Saving Sensor Data ..........................................................................................66

6.

BASE STATION INTERFACE SOFTWARE ..........................................................69

7.

TESTING RESULTS .................................................................................................70

8.

SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS .............................73

References ..........................................................................................................................77

x

LIST OF FIGURES Figures

Page

1.

Figure 2.1. Circadian temperature rhythm. ......................................................... 5

2.

Figure 2.2. Temperature measurement variation with location. ......................... 7

3.

Figure 2.3. Data from a daily routine activity monitor. ...................................... 9

4.

Figure 2.4. Laser light penetration. .................................................................... 13

5.

Figure 2.5. Tissue light absorption. .................................................................. 14

6.

Figure 2.6. Absorption of oxygenated and deoxygenated haemoglobin versus wavelength...................................................................................... 15

7.

Figure 2.7. Ratio of red to infrared amplitudes versus blood oxygenation....... 16

8.

Figure 2.8. Dependence of light absorption on tissue. ...................................... 17

9.

Figure 2.9. Dependence of blood cell alignment on flow. ................................ 17

10.

Figure 2.10. Systole and diastole correlation to signal. .................................... 18

11.

Figure 2.11. Transmission mode (left) and reflectance mode (right) ............... 19

12.

Figure 2.12. Normal and undesirable pulse oximeter signals. .......................... 20

13.

Figure 3.1. Wireless sensor schematic. ............................................................. 22

14.

Figure 3.2. Base-station schematic ................................................................... 23

15.

Figure 3.3. Temperature sensor circuitry .......................................................... 24

16.

Figure 3.4. Accelerometer circuitry .................................................................. 25

17.

Figure 3.5. Wireless sensor electronic assembly .............................................. 26 xi

18.

Figure 3.6. Red and IR light source. ................................................................. 27

19.

Figure 3.7. Photodetector and amplification circuitry ...................................... 28

20.

Figure 3.8. Relative intensity versus wavelength. ............................................ 29

21.

Figure 3.9. Photodiode relative spectral sensitivity versus wavelength ........... 30

22.

Figure 3.10. Power supply circuit ..................................................................... 32

23.

Figure 3.11. Microcontroller circuit.................................................................. 35

24.

Figure 3.12 RF circuit ....................................................................................... 36

25.

Figure 3.13. MicroSD card circuit .................................................................... 37

26.

Figure 3.14. Sound and light alarm circuit ....................................................... 38

27.

Figure 4.1. Wireless sensor initialization .......................................................... 39

28.

Figure 4.2. Wireless sensor microcontroller ..................................................... 40

29.

Figure 4.3. IR (bottom) and red (top) LEDs on/off sequence ........................... 47

30.

Figure 4.4 First stage amplifier settling time .................................................... 47

31.

Figure 4.5. First (top) and second (bottom) stage amplifier outputs ................ 48

32.

Figure 4.6. Displaying unfiltered red and filtered IR signals ........................... 51

33.

Figure 4.7. Pulse detecting signal (on T channel) ............................................. 52

34.

Figure 4.8. IR heart pulse (top) and pulse detect (bottom) ............................... 54

35.

Figure 4.9. The ac and dc components of the IR signal. ................................... 56

36.

Figure 4.10. Percent oxygen saturation versus R/IR ratio. ............................... 57

37.

Figure 4.11. Pulse oximeter function ................................................................ 59

38.

Figure 4.12. MicroSD card function ................................................................. 60 xii

39.

Figure 5.1. GUI timed acquisition control panel .............................................. 64

40.

Figure 5.2. Wireless sensors grouped for timed acquisition ............................. 65

41.

Figure 5.3. Wireless sensors grouped for timed acquisition for different On/Sleep times ............................................................................... 65

42.

Figure 5.4. Display monitor. ............................................................................. 67

43.

Figure 5.5. Plotting saved data in Excel. .......................................................... 68

44.

Figure 7.1. Accelerometer data (three axis) ...................................................... 72

xiii

1 CHAPTER 1 1. INTRODUCTION Development of continuous remote health monitoring is necessary, because people live longer and “baby boomers” are aging, threatening to overwhelm health-care providers [1]. Implementation of continuous health monitoring contributes to an increase in health-care quality and a decrease in health-care cost. Prevention of the onset of chronic disease can dramatically reduce hospitalization time and increase the quality of life, with the associated reduction of health care costs. A person’s health can drastically worsen between scheduled physician visits. But, with continuous health state monitoring, office visits can be reduced, because a physician reviews collected data more frequently. Continuous remote health monitoring will enable aging people and people with chronic illnesses to live in their homes and be independent. A continuous health monitoring system will aid people with epilepsy. The standardized mortality rate (SMR) is 1.6-9.3 times higher for people with epilepsy than in the healthy population [2]. In severe epileptic episodes, death from airway constriction can occur in less than 3-5 minutes unless help is immediately provided. Detecting changes in heart rate and oxygen concentration in the blood would indicate that the person is in crisis or that a crisis is imminent and help is needed. Continuous monitoring aids in detecting changes in health condition in postoperative hospitalized patients. A post-operational patient monitoring system for detecting deterioration in post-operative hospitalized patients, where medical staff is

2 available but is unaware of the deterioration, has proven very useful [3]. The monitoring system consisted of pulse oximeters worn by the patients that notified a nurse via wireless pager when abnormalities were detected. The system helped doctors to intervene before a critical situations developed. According to the results of the study, rescue events decreased from 3.4 to 1.2 per 1,000 patient discharges and intensive care unit transfers from 5.6 to 2.9 per 1,000 patient days. Pulse-oximeter implementation saved the 36-bed intensive care unit 135 days per year. In the United States, about 2,500 infants die each year because of SIDS [4]. Apparently, the brain triggers the babies to wake from sleep and cry when a lack of oxygen is detected. This response changes their heartbeat and breathing patterns to make up for the lowered oxygen and excess carbon dioxide. If this mechanism is not developed, the infants die in their sleep. A continuous monitoring system, capable of measuring heart rate and oxygen concentration in blood, could prevent many infant deaths. Continuous health monitoring could be used for healthy people who are in stressful situations, such as firefighters, combat personnel, and athletes. The development and initial bench testing of a continuous monitoring system using a pulse oximeter and an accelerometer has been done for such applications [5]. The system was used for monitoring first responders and critically injured persons. The oxygen concentration and heart rate gives the health state of a soldier, and the accelerometer shows the soldier’s activity and body orientation.

3 An internet-based health monitoring system has been developed and tested for monitoring brain-injured children [6]. The test results showed that even though the system needs improvements, 78% of patients and families were interested in further use of the system. Continuous health monitoring will play a big role in the health-care industry and, according to Kalorama, the market will increase from $ 5.7 billion in 2009, with an annual growth of around 26 % through 2014 [7]. General Electric and Intel joined to create a health-care company to develop telehealth systems [8]. The goal of this project is to develop a remote continuous noninvasive Health State Monitoring System (HSMS), capable of indirect measurement of the percent oxygen saturation in the blood, heart rate, physical activity, and temperature. The system would consist of a wearable device with a unique identification number (ID), containing sensors and communicating circuitry, called further the “wireless sensor,” a personal computer (PC) based graphical user interface (GUI), and a base-station interface between the wireless sensor and the PC. The wireless sensor, with a unique ID, consists of a pulse oximeter, activity and temperature sensors, wireless circuitry for communication with the base-station, a programmable microcontroller, local data storage, and local sound/light alarms. The GUI allows the user to remotely access an individual ID addressable wireless sensor for controlling the wireless sensor and for retrieving the sensor data in real time. The control involves setting the sensor’s data acquisition rate, enabling/disabling the sensors, enabling/disabling data saving on the wireless sensor microSD card, and setting

4 the source and levels for triggering sound and/or light alarms. The base-station is an interface between the PC and the wireless sensor that allows the wireless sensor to send data to the PC and the PC to control the wireless sensor. This report is structured as follows: Chapter 2 provides a background description of the sensors for collecting vital data used in the project. The description includes the importance of the sensor measurement, application areas, and the potential for expanding the application areas. Chapter 3 describes the circuit that was built for this project. The circuit description is divided into two main sections, the sensor circuitry and supporting circuitry. The sensor circuitry section describes the circuits and hardware of the temperature, activity, and pulse oximeter sensors. The core components of each sensor are described. The supporting circuitry section describes the circuit and the hardware of the power supply, wireless communication, microcontroller features, microSD storage, the sound and light alarm, and the microcontroller that controls the hardware. Chapters 4, 5, and 6 describe the software and provide software flowcharts and code snippets for the HSMS. There are three independent units: the wireless sensor software is described in Chapter 4, the GUI software is described in Chapter 5, and the base-station software is described in Chapter 6. Test results for the HSMS are provided in Chapter 7. Chapter 8 gives a summary, conclusions, and recommendations for improving the system.

5 CHAPTER 2 2. BACKGROUND 2.1. Temperature Monitoring 2.1.1. Temperature Monitoring Importance Temperature measurement commonly is used to determine if a person has a fever and, if so, the severity of the fever. However, body temperature also is affected by stressful situations, a person’s age and activity, time of the day and season, pregnancy, hormonal contraceptives, etc. The temperature of the human body has a twenty-four hour pattern (see Figure 2.1), called a circadian rhythm, and a season pattern, called a circannual rhythm [9].

Figure 2.1. Circadian temperature rhythm. The figure show 24-hour temperature variation for various location [9].

6 Continuous temperature measurement can provide additional information about a person’s state of health. For example, the disruption in timing of acrophase (the time at which the peak of a rhythm occurs) is associated with insomnia and several chronic diseases, including cancer and HIV. Continuous temperature measurement not only can be used for detecting medical abnormalities but also for evaluating the effectiveness of medications. 2.1.2. Temperature Measurement Principle The importance of clinical temperature measurement is traced back to the times of Hippocrates [10]. At that time, hands were used to estimate the heat or cold of a human body. The first physician to use a thermometer at his patients' bedside is considered to be Hermann Boerhaave in the 16th century. Sir Thomas Allbutt invented in 1867 the first practical medical thermometer used for taking the temperature. It was portable, 6 inches in length, and able to record a patient's temperature in 5  min. Many devices were developed for temperature measurement. One of the most common devices for measuring temperature is the glass thermometer which uses the thermal expansion properties of a liquid (spirits, mercury). Modern thermometers are based on semiconductor temperature sensors and are capable of taking a measurement in 5 to 10 s. Temperature sensors are available from different manufacturers. A sensor needs to be selected based on size, accuracy, type of output (analog or digital), power consumption, sensor price, and, for digital sensors, the choice of I2C or SPI communication protocol.

7 For this project, a low-cost temperature sensor, the TMP121 from Texas Instruments (TI), Dallas, Texas, with ± 1.5 °C accuracy and with the 0.0625 °C resolution was selected. For continuous monitoring, body tracking temperature variation is more important than accuracy. Body temperature varies based on gender and on different locations on the body (Figure 2.2) and depends on the season, time of the day, and the person’s fitness [9]. The body temperature measurement is done every 30 s (default setting), because body temperature changes relatively slowly.

Figure 2.2. Temperature measurement variation with location. The figure shows the temperature measurement variation in dependence of gender and measurement location [9].

8 2.2. Activity Monitoring 2.2.1. Activity Measurement Importance Health is in direct relationship with a person’s lifestyle. Many health problems could be prevented by sufficient physical activity and by sufficient rest. Staying “in shape” requires exercise [11]. But, over-exercising harms health. For example, “intense exercise can cause scarring and fibrosis in the heart” [12]; vigorous activity can “increase the risk of sudden cardiac death and acute myocardial infarction” [13]. Insufficient as well as excessive [14] sleep can cause cardiovascular diseases [15], diabetes, mental problems [16], [17], and accidents [18]. Therefore, by monitoring activity, in correlation with other physiological measurements, a safe exercise intensity and rest time program can be set. For example, cardiac infarction can occur during jogging. By correlating the occurrence of heart abnormalities with activity intensity, a safe level of activity intensity can be determined. Exceeding the set activity level would trigger an alarm, warning the jogger to decrease the activity. By analyzing the rest time (sleep, watching TV etc.) of a person, a correction to the lifestyle can be suggested. Figure 2.3 shows an example of data from a daily routine activity monitoring [19]. A decrease in activity during the day or increase in activity during the night, in comparison with the previously collected data for a specific person, can indicate that the person experiences some discomfort. For example, a person suffering from pain may reduce daily activities. At the same time, because of the pain, the person might not sleep well, resulting in an increase in activity during the rest period.

9

Figure 2.3. Data from a daily routine activity monitor. The figure shows three-axes accelerometer data in dependence of the persons activity. The standing, sitting, walking, lying and running activities are detected.

Activity monitoring can be used for fall detection [20], balance control evaluation, and determining walking [21]. Falling is one of the major causes of injuries for elderly, and even if diligent efforts are made to prevent elderly, falls are often inevitable [20]. Fall detection can alert that a person has fallen and needs help.

2.2.2. Activity Measurement Principle For the project, a three-axis accelerometer is used for continuous activity monitoring. The accelerometer is capable of detecting six directions of movement and

10 can accurately detect gravity. Therefore, activities such as running, standing, and lying can be determined. The accelerometer can be set to generate an interrupt if the activity decreases below a set threshold or when activity exceeds a set threshold for a user set time duration. The variability in daily activity of a person can be used to estimate a person’s health. 2.3. Cardio-Respiratory Assessment 2.3.1. Pulse Oximeter Importance A pulse oximeter measures the percent oxygen saturation in the blood. Because the pulse oximeter quickly gives a general assessment of a patient's cardio-respiratory status, and the measurement is non-invasive, the pulse oximeter has become a standard monitoring device before, during, and after operations [22]. The device also is used in intensive care, recovery rooms, and in emergencies [23]. Existing pulse oximeters, due to their high power consumption, are not intended for long term monitoring and are usually attached to a finger, an earlobe, or the forehead. Using a pulse oximeter for long term monitoring allows determining the heart rate variability (HRV). The importance of HRV is that it can be used to predict potential complications in pregnancies, fetal distress, and in neonatal critical care [24]. The HRV can be used for assessing the severity of congestive heart failure [25], stroke, Parkinson’s disease and depression for people with chronic disorders [26].

11 2.3.2. Principle of Pulse Oximeter Measurements Blood oxygen concentration, pulse, and pressure show the function of the heart and lungs. Pulse oximeter devices use (at least) two sources of light of different wavelengths and use the fact that oxygen saturated blood and unsaturated blood absorb the light differently. The oxygen in the body is carried by the protein hemoglobin (Hb) in the red blood cells. The oxygen is bound to hemoglobin in the lungs and then transported through the entire organism for respiratory processes. Oxygen saturation (SO2) is defined as the ratio of the concentration of oxygenated hemoglobin (HbO2) to the total concentration of hemoglobin. This is,

SO2

CHbO2 CHbO2

CHb

,

(0.1)

where C is concentration and Hb is deoxygenated hemoglobin. Using two light sources with such wavelengths that one is absorbed more by the oxygenated hemoglobin and another more by the unoxygenated hemoglobin allows determining the blood oxygen saturation. The relevant parameter for estimating the blood oxygen concentration is the light intensity, which is affected by absorption, and is measured by a photosensor. The signal detected by the photodetector consists of ac and dc parts. The ac part corresponds to the light absorption by the blood cells, and the dc part corresponds to the light absorption by the tissue. The dc part corresponds to the light

12 absorption by the tissue because the consistency of the tissue (water, melanin) changes very slowly. Because the light absorption by blood cells varies as a result of blood pulsation, it results in the ac part of the detected signal. The ratio of ratios formula that is needed for calculating the blood oxygen saturation is

I max Red ) I min Red , I ln( max IRed ) I min IRed ln(

R

(0.2)

where Imin Red and Imin IRed are the minimum values of the ac components of the red and infrared light sources, respectively, and Imax Red and Imax IRed are the maximum values of the ac components of the red and infrared light sources, respectively. The ratio of logarithmic normalized intensities is used to calculate the oxygen concentration using an empirical calculated calibration curve.

2.3.3. Optical Properties of the Tissue The tissue is made up of cells of different functions and sizes; therefore, the tissue is an inhomogeneous medium with different and randomly distributed absorbers and scatterers. When the tissue is irradiated by light, photo-thermal and photochemical reactions occur, including fluorescence, optical reflectance, and transmission processes. The absorption of light through the tissue is wavelength dependent; therefore, the penetration depth of light in tissue is different for different wavelengths (see Figure 2.4)

13 [4]. The peak depth penetration of 1.4 mm is at 1064 nm. Light penetration depth is defined as the depth where the incident irradiance of the light is decreased to 37 % of the original intensity.

KrF XeCl Dye Argon Diode Nd:YAG Tm:YAG Ho:YAG Er:YAG CO2 248 308 465 514.5 830 1064 2010 2100 2940 10600 λ (nm) 1 µm

5µm

20 µm 40 µm

150 µm

250 µm

330 µm 300-400 µm

1300 µm 1400 µm

Figure 2.4. Laser light penetration. The figure shows light penetration depth in tissue the dependence of laser light wavelength (adapted from [27]).

Figure 2.5 shows plots of absorption coefficients versus wavelength for protein, melanin, collagen, water, deoxygenated hemoglobin, and oxygenated hemoglobin - the main components of the tissue. At wavelengths lower than the 805 nm isosbestic point,

14

Figure 2.5. Tissue light absorption. The figure shows the light absorption coefficients for protein, melanin, collagen, water, deoxygenated hemoglobin, and oxygenated hemoglobin versus wavelength [28].

Hb is the strongest absorbent, and at wavelengths higher than 805 nm, HbO2 is the strongest absorbent. In the range from 600 nm to 1000 nm, the deoxygenated and oxygenated hemoglobin are the main light absorbents, and this range is called the tissue window. The range of the tissue window and the absorption coefficients of Hb and HbO2 in this window dictate the wavelengths used for a pulse oximeter. In this tissue window, 660 nm, created by a red light emitting diode (LED), and 940 nm, created by an infrared LED, wavelengths are chosen. The absorptions of the red and infrared signals are not equal, as seen in Figure 2.6. Therefore, different signal amplitudes are expected for red and infrared LEDs having equal light intensities, and the changes in signal amplitude

15

Figure 2.6. Absorption of oxygenated and deoxygenated haemoglobin versus wavelength. The figure shows the relative light absorbtion levels of oxygenated and deoxygenated blood versus wavelength [29]. The light absorption levels for the red and infrared are marked. The isobestic point (light absorbtion coefficients of oxygenated and deoxygenated blood are equal) is at 805 nm and cannot be used for pulse oximetry.

received from red and infrared LEDs are different for different levels of oxygen concentration in blood. For saturations below 85 %, the signal from the red LED has the largest ac part and for oxygen saturations higher than 85 %, the infrared (IR) signal has the largest ac part. At 85 % oxygen saturation, the ratios of the red to the IR ac components of the signals equals one, and at 100 %, the signals’ ac ratio is 0.43 [6].

16 Figure 2.7 [30] shows the ratio of the red to infrared signal amplitudes for different percentage oxygen saturation in blood.

Figure 2.7. Ratio of red to infrared amplitudes versus blood oxygenation. The figure shows that the amplitude of the red signal decreases (absorption increases) and the amplitude of the infrared signal increases (absorption decreases) with the increase of oxygenation [30].

Equation (0.1) shows the oxygen saturation determined by a pulse oximeter. The measurement is actually called the functional oxygen concentration, because there are other types of hemoglobin in blood, such as carboxyhemoglobin (COHb) and methemoglobin (MetHb). The dependence of light absorption on tissue is shown in Figure 2.8 [28]. The light absorption by the tissue, venous blood, and nonpulsating arterial blood in the path of the light constitutes the constant portion of the signal.

17

Figure 2.8. Dependence of light absorption on tissue. The figure shows that the ac component of light absorption is created by the pulsating arterial blood and the dc component is created by nonpulsating tissue [28].

Figure 2.9 [29] shows that the ac portion is affected by the changes in the orientation of the blood versus blood flow rate. During systole (maximum), the blood cells, which have the form of a biconcave disk, are aligned perpendicular to the blood flow, increasing light absorption, and during diastole (minimum), the blood cells are aligned parallel to the flow (see Figure 2.10 [31]).

Figure 2.9. Dependence of blood cell alignment on flow. The figure shows changes in blood cell alignment during and in between heartbeats [29].

18

Figure 2.10. Systole and diastole correlation to signal. The figure shows blood cell alignment at systole and diastole and correlation to the detected signal [27].

2.3.4. Transmission and Reflectance Modes of Pulse Oximeter Measurements Transmission and reflectance type of pulse oximeters use the same technology. The difference is in the photodetector and light sources relative placement. In transmission mode, the light source and the photodetector are placed on opposite sides of the investigated object. This method is more common and is used in hospitals where a sensor-clip is attached to the person’s finger or ear lobe.

19 In reflective mode, the light sources and the photodetector are placed on the same surface. This configuration allows making smaller patch type sensors, which also have less restriction on body placement. Figure 2.11 shows the transmission mode and reflectance mode principle.

Figure 2.11. Transmission mode (left) and reflectance mode (right). The figure shows the position of the photodetector relative to the light sources for transmission and reflective types of pulse oximeters [32].

2.3.5. Interference to the Pulse Oximeter Signal The pulse oximeter signal is degraded by the low perfusion, noise artifacts, and motion artifacts. Figure 2.12 [30] shows the shape of a normal pulse oximeter signal (top) and degraded pulse oximeter signals. Noise artifacts are created by electrical sources. High frequency noise can be reduced by low pass filtering. To reduce low frequency noise, the signal can be averaged over multiple pulses.

20

Figure 2.12. Normal and undesirable pulse oximeter signals. The figure shows the waveforms obtained from a pulse oximeter. The shape of a normal signal is the desired signal. Low perfusion is caused by health conditions (cardiac arrhythmias, heart failure, peripheral vascular disease, hypotension) [33]. Noise artifacts are caused by electrical equipment. Motion artifacts are caused by the movments of the person using the pulse oximeter.

21 CHAPTER 3 3. HEALTH STATE MONITOR CIRCUITRY 3.1. Health State Monitor Circuitry Overview The HSMS circuitry consists of the wireless sensor circuitry and base-station circuitry. An overview of the wireless sensor circuitry is shown in Figure 3.1 and consists of 1.

Sensor circuitry

2.

Temperature sensor circuitry (TEMP)

3.

Accelerometer circuitry (ACCEL)

4.

Pulse oximeter circuitry that consist of photodetector (AMP) and light generation and control (IR_RED) circuitries

5.

Supporting circuitry

6.

Power supply circuitry (PS) for controlling the power supply voltage

7.

Microcontroller (MCU) circuitry that controls all the circuits

8.

Wireless communication circuitry (RF)

9.

Local data storage circuitry (microSD)

10.

Light and sound (ALARM) alarm circuitry.

22

Figure 3.1. Wireless sensor schematic

An overview of the base-station circuitry is shown in Figure 3.2 and consists of 1.

Power supply circuitry (PS) for controlling the power supply voltage

23 2.

Microcontroller circuitry (MCU) that controls all the circuits

3.

Wireless communication circuitry (RF)

4.

Serial communication.

Figure 3.2. Base-station schematic

3.2. Sensor Circuitry This section describes the circuitry of the sensors that are used for determining health status data. It includes temperature, activity, and pulse oximeter circuitry. For measuring temperature and activity, dedicated integrated sensors are used.

24 3.2.1. Temperature Monitoring Circuitry For temperature measurement, a dedicated integrated sensor, the TMP121 is used. It has 12 bits of resolution plus one bit for indicating negative temperatures. The sensor has 0.0625 °C resolution and ± 1.5 °C accuracy in the −25 °C to +85 °C temperature range. In the anticipated temperature range, based on circadian rhythm (see Figure 3.3), the accuracy is -0.5, +0.9 °C. The temperature sensor is accessed via a serial peripheral interface (SPI), acquires a new temperature measurement every 0.5 s and needs 0.25 s to perform the conversion. The sensor has the following features: 1.

Power supply range: 2.7 V to 5.5 V

2.

Current Consumption a. Temperature acquisition - 50 µA b. Idle state - 20 µA c. SPI access - unspecified

Figure 3.3. Temperature sensor circuitry

25 3.2.2. Activity Monitoring Circuitry For activity measurement, an integrated three-axis 12-bit resolution accelerometer, the LIS331DLH from STMicroelectronics (Geneva, Switzerland) is used. Accelerometer circuitry is shown in Figure 3.4. The sensor has dynamically selectable full-scale ranges of ±2 g, ±4 g, and ±8 g, six direction detection, and a data rate of up to 1 kHz. From the available accelerometer options, of interest are low (lower than set threshold value) and high (higher than set threshold value) interrupts and, depending on the application, six direction detection.

Figure 3.4. Accelerometer circuitry

Low activity interrupts combined with the time period of low activity are useful for detecting lower than usual activity for a certain individual or for detecting no activity

26 at all, as when the individual is unconscious. High level activity interrupts are useful for detecting falls or discomfort. For example, higher than usual activity for a specific individual during the rest period can indicate that the person feels some discomfort and needs assistance.

3.2.3. Pulse Oximeter Circuitry The pulse oximeter circuitry was designed specifically for this project with low power consumption and small size (see Figure 3.5) as the primary criteria. The pulse

Figure 3.5. Wireless sensor electronic assembly

27 oximeter circuitry consists of red and IR light-emitters (Figure 3.6), and photodetector and amplifier circuitry (see Figure 3.7). The microcontroller controls the light-emitting circuitry. It powers sequentially each LED with such a voltage that the LED generates strong enough light for the photodetector to detect the light reflected by blood cells, but not too strong to exceed the amplifier U7 (see Figure 3.7) output range and the microcontroller’s analog-to-digital converter (ADC) input range.

Figure 3.6. Red and IR light source.

The red and IR LEDs’ anodes are connected to individual pins of the microcontroller, with which the microcontroller controls each LED. The cathodes of the LEDs are connected to the digital-to-analog converter (DAC) microcontroller pin, via which the microcontroller adjusts the intensity of the LEDs. The red and the IR LEDs require different voltage levels to generate equal light intensities. In addition, the photodetector sensitivity is different for red and infrared lights. Plots of light intensity

Figure 3.7. Photodetector and amplification circuitry

28

29 versus wavelength for the LEDs are shown in Figure 3.8 [34], and the plot of photodetector sensitivity versus wavelength is shown in Figure 3.9 [35]. The necessary voltages for the LEDs are calculated and are provided by the DAC. A specialized part, the SML12R3KIR941T LED from Ledtronics (Torrance, CA), that contains red and infrared LEDs in one package is used as the light emitting source.

Figure 3.8. Relative intensity versus wavelength. The figure shows the relative light intensity of the red LED (top), and infrared LED (bottom) of the light source (LIS331DLH) used in this project [34]

30

Figure 3.9. Photodiode relative spectral sensitivity versus wavelength. The figure shows the VBPW34S (Vishay Semiconductors, Malvern, PA) photodetector relative sensitivity versus wavelength [35].

If the wireless sensor is on, then every 10 ms, the microcontroller performs the following sequence: 1.

turns infrared LED on

2.

waits 500 µs

3.

ADC samples amplifier U7 (see Figure 3.6) output

4.

turns infrared LED off

5.

turns red LED on

6.

waits 500 µs

7.

ADC samples amplifier U7 (see Figure 3.6) output

8.

Turns red LED off.

31 The photodetector circuit consists of a photodiode and a two-stage amplifier. The first stage is an OPA381 transimpedance amplifier from TI. The second stage is an OPA333 op-amp, also from TI. The output from the first stage is sampled and is used for calculating the intensity of the LEDs. Another microcontroller DAC provides an adjustable voltage to the non-inverting input of the second stage amplifier for removing the dc component. The microcontroller samples the second stage amplifier output, which is the signal for calculating the heartbeats and the percent oxygen saturation in blood.

3.3. Supporting Circuitry This section describes wireless sensor supporting circuitry. It includes power supply, microcontroller, wireless communication, microSD card, and sound and light alarm circuitry. 3.3.1. Power Supply Circuitry The power supply circuit consists of an LM3670 dc-to-dc adjustable converter from TI. The circuit is shown in Figure 3.10. It provides 3.3 V or 1.8 V. In the active mode, the power supply voltage is set to 3.3 V. In “Sleep” mode, the output is 1.8 V and is set by connecting PS_REG to ground. In active mode, the microcontroller switches the pin to high input impedance.

32

Figure 3.10. Power supply circuit

3.3.2. Microcontroller (MCU) Circuitry An ATXMEGA32A4 microcontroller from Atmel (San Jose, CA) is used in this project. The microcontroller circuit is shown in Figure 3.11. It was chosen because of the following features: 1.

Five “Sleep” modes.

2.

16-bit Real Time Counter (RTC) with a separate oscillator, which runs at 38 kHz. From the microcontroller’s “Sleep” mode options, the power-save option is chosen. In this mode, the microcontroller has the lowest power consumption that has the RTC running. When the wireless device in active mode, the RTC wakes up the microcontroller from the power-save mode to start data acquisition every 10 ms. When the wireless device is in “Sleep” mode, the RTC wakes the microcontroller every 2 s.

33 3.

One twelve-channel, 12-bit, 2 Msps ADC. It is possible to configure the ADC to have up to four differential inputs with or without gain. The gain is software selectable from 1 x, 2 x, 4 x, 8 x, 16 x, 32 x or 64 x. For this project, the ADC is configured to have one differential input with gain and one single-ended input. The differential input is used for acquiring the signal from the photodetector’s second stage amplifier. The single-ended input is used to acquire the signal from the photodetector’s first stage, which is used for calculating the LEDs’ light intensities.

4.

Two-channel, 12-bit, 1 Msps DAC. One channel is used to adjust the LEDs intensities, and another for adjusting the second stage amplifier dc offset.

5.

Multiple options for ADC and DAC reference. According to the datasheet, the ADC can use the internal 1 V, internal Vcc/1.6, or external voltage reference on the AREFF pin as a reference voltage. The DAC has an internal 1 V, external AVcc, or AREFF reference voltage. In this project, for the DAC, AVcc is the reference, and for the ADC, Vcc/1.6, which is 2 V for a 3.3 V supply, is the reference.

6.

Five 16-bit Timer/Counters. Only one timer is used to handle all needed delays. For example, it handles the 500-µs delay needed for each LED. The wireless transceiver uses various delays for tune-up and data transmission. During the delays, the microcontroller is in the idle power saving mode. The timer interrupt wakes the microcontroller after each delay.

34 7.

Two SPIs. Only one SPI is used to handle the wireless transceiver, accelerometer, temperature sensor, and the microSD card communications.

8.

Multiple interrupt options for I/O pins. Only one interrupt is used to handle all the transceiver requests.

9.

Thirty four programmable input/output (I/O) pins. From 34 pins, 30 are connected. The rest have important functions but were difficult to access. This microcontroller has a 49-ball VFBGA package. Accessing these pins requires micro-vias that would increase the cost of the printed circuit board (PCB). Two connected I/Os also are not used. There are two pins for controlling the power supply voltage, but in this design, only one is used for switching between 1.8 and 3.3 V. The pin connected to the microSD card detect also is not used.

10.

Multiple dynamically selected clock options. To minimize the wireless sensor size, only the internal microcontroller clock sources are of the interest. The internal clock options are: 32-MHz run-time calibrated RC oscillator, 2-MHz runtime calibrated RC oscillator, 32.768-kHz calibrated RC oscillator, 32-kHz Ultra Low Power (ULP) oscillator with 1-kHz output, and a phase lock loop (PLL) with 1 to 31 x multiplication. The microcontroller always starts at 2 MHz after reset. For all microcontroller processes, an 8-MHz (2 MHz clock · 4PLL) clock is selected.

11.

Low power supply voltage. The microcontroller works in the 1.6-3.6 V range.

12.

Small 5 x 5 mm package size.

35

Figure 3.11. Microcontroller circuit

3.3.3. Radio Communication Circuitry For the wireless communications, the SI4432, a highly integrated single chip transceiver from Silicon Labs (Austin, Texas) is used. The schematic is shown Figure 3.12. This transceiver covers frequencies from 240–960 MHz in 156-Hz or 312-Hz steps, has high (–121 dBm) sensitivity, and has seven power transmission options with +20

36 dBm maximum power output for extended range. The wide frequency range can accommodate RF regulations in different countries. However, it requires different component value for different frequencies.

Figure 3.12 RF circuit

3.3.4. MicroSD Card Circuitry The microSD circuit just needs a microSD cardholder. The connections to the cardholder are shown in Figure 3.13. The microSD card supports SPI communication. Only R14 and R31 are placed in the microSD card circuitry. In a future version, the microcontroller universal synchronous asynchronous receiver transmitter (USART) in

37 SPI mode will be used for communication with the microSD card. The microcontroller supports DMA access to the USART. Using the microcontroller direct memory access (DMA) capability for writing to the microSD card will decrease power consumption.

Figure 3.13. MicroSD card circuit

3.3.5. Alarm Circuitry The alarm schematic is shown in Figure 3.14 and consists of an LED (D9) and a buzzer (SG2) with separate activating lines. Each is activated by the microcontroller when a monitored parameter is out of the range set by the user.

38

Figure 3.14. Sound and light alarm circuit

3.4. Base-Station Circuitry Base-station circuitry consists of the following: 1.

Power supply circuitry (PS) for controlling the power supply voltage

2.

Microcontroller circuitry (MCU), that controls all the circuits

3.

Wireless communication circuitry (RF)

4.

Serial communication.

The wireless sensor contains all the base-station sub-circuits, including the serial communication, which it does not use. Therefore, a wireless sensor is used as a basestation. It requires different software and a serial-to-universal serial bus (USB) converter. In the project, a serial-to-USB converter is used that also provides 3.3 V to power the circuit, eliminating the need for a separate power supply. The base-station communicates with the PC at 115200 bps.

39

CHAPTER 4 4. WIRELESS SENSOR SOFTWARE This chapter describes the software of the wireless sensor microcontroller. The software is written in C language, and the CVAVR compiler from HP InfoTech (Bucharest, Romania) is used. 4.1. Microcontroller Main Routine After powering up, the wireless sensor configures the ports and the real time clock (RTC) for 2 s, and then goes into power save mode. The flow chart for wireless sensor initialization is shown in Figure 4.1. All processes start after the RTC interrupt, which occur every 2 s if the wireless sensor is in the “Sleep” mode and every 10 ms if it is in active mode.

Start Initialization Sleep

Figure 4.1. Wireless sensor initialization The processes that the microcontroller performs after each RTC interrupt are shown in the flow chart Figure 4.2. In the “Sleep” mode, after each RTC interrupt, the

40 REAL TIME CLOCK INTERRUPT

YES

NO

IS T INTERVAL >1S?

IS ON MODE? NO

YES GET TEMPERATURE

CHECK IF RX DATA

YES IS RF DATA? IS ACCELEROMETER ON?

UPDATE SETTINGS

YES GET ACCELEROMETER NO

NO

YES IS OXIMETER ON?

GET OXIMETER

NO

IS ALARM STATUS CHANGED? NO

YES

SET ALARM ON\OFF MODE

IS RX RECEIVED? YES IS SD ON?

NO

YES

SD FUNCTION

UPDATE SETTINGS NO YES IS TX BUFFER FULL? NO

NO IS RX TIME?

NO RX FOR >18S?

NO SET SLEEP 10US

YES SER SLEEP 2S

YES TRANSMIT\RECEIVE DATA

SLEEP

Figure 4.2. Wireless sensor microcontroller

41 microcontroller checks for an incoming RF signal. If no signal is received from the basestation, the wireless sensor goes to “Sleep” for 2 s to repeat the procedure again. If a base-station transmission was received, the wireless sensor updates its settings based on the received information and then checks again the sleep/on setting and proceeds accordingly. If the wireless sensor is in the “On” mode, then it acquires the temperature data if more than 30 s elapsed since the last temperature acquisition. Then, the microcontroller checks the accelerometer. If the accelerometer is set for continuous data acquisition, then the accelerometer data are acquired based on the sampling rate set for the accelerometer. If the accelerometer is set to generate an out-of-range interrupt, then the microcontroller checks the interrupt, and if an interrupt is registered, then accelerometer data are acquired. Then the microcontroller acquires the pulse oximeter data if the oximeter is activated. The flow chart for pulse oximeter data acquisition is described in section 4.4.4. After that, the microcontroller checks the light and sound alarms. If any of the parameters set to be monitored is out of range, the alarm set to be activated is activated. If either of the alarms was previously activated, and a command to deactivate is received from the base-station, then the microcontroller disables the alarm. Next, the microcontroller checks if the acquired data are to be saved on the microSD card. The microSD card routine and flow chart are described in section 4.5. After data acquisition, the microcontroller checks if more than 1 s passed since the last command from the base-station was received, and if so, then regardless of the

42 amount of data in the buffer, the data are sent to the base-station, along with the code querying the base-station for a command. If the last RF transmission was less than 1 s ago, then the microcontroller checks if the 64-byte buffer is full, and if it is full, then the data are transmitted to the base-station. If a code querying the base-station for a command was sent, then the wireless sensor enables the wireless receiver and waits for 2 ms for a command from the basestation. If a command was received from the base-station, then the wireless sensor updates its settings and goes into power-save mode for the remaining time to complete 10 ms. If no command was received from the base-station for more than 20 s, then the wireless sensor assumes that the base-station is not in the range and goes into “Sleep” mode.

4.2. Temperature Acquisition The temperature sensor is accessed via a three-wire interface and does not accept any incoming data. It is the only sensor that requires more than 12 bits to represent the data and does not fit the 12 bits data plus four bits for sensor identification scheme used for the wireless protocol, which is described in section 4.6. The temperature data are provided in two’s compliment format and consist of twelve bits plus one sign bit. To fit the 12+4 format, the sensor data could be trimmed to represent only positive temperatures or to reduce the resolution.

43 The wireless sensor is intended to be placed on human skin, and negative temperatures are not expected. Therefore, only positive temperatures are measured for this project. The code snippet for temperature acquisition is void get_TMP(void) { S16 temp_tmp; TMP_cnt_SR+=RTC_PERf; if (TMP_cnt_SR>=3; sensor_data[data_amount++]=((unsigned char *) &temp_tmp)[1]|ID_TMP; sensor_data[data_amount++]=((unsigned char *) &temp_tmp)[0]; TMP_OFF; TMP_cnt_SR=0 . }

4.3. Accelerometer Acquisition The 12-bit three-axis accelerometer is accessed via SPI communication. It is programmed for the ±2 g range. The accelerometer can be set to generate an interrupt when a value lower and/or higher than preset threshold value is measured. The following options are available from the GUI: 1.

enabling/disabling the accelerometer

2.

continuous operation or interrupt-based monitoring

44 3.

in continuous operation, setting the data acquisition sampling rate (maximum 100 Hz)

4.

in interrupt-based monitoring, setting the free fall detection, minimum, and maximum activity out-of-range detection.

The data acquisition sampling rate refers to the frequency the accelerometer is accessed; the accelerometer itself is set to a 100-Hz sampling rate.

4.4. Pulse Oximeter Acquisition 4.4.1. Pulse Oximeter Software Overview If the pulse oximeter is enabled, then regardless of the pulse oximeter sample rate setting, the data are sampled at 100 Hz. The oximeter sample rate that is set by the user refers to the rate the sampled data are sent to the PC. The display of detected red and IR signals is useful for positioning the wireless sensor on a patient. For continuous monitoring of the percent oxygen saturation in blood and heart rate, the sampling rate is set to a lower rate. Physiologically relevant changes in the percent oxygen saturation in blood and heart rate do not change abruptly; therefore, a lower sampling rate is desired to increase data storage and lower power consumption. The software for the pulse oximeter routine has the following hardware access and control: 1.

Turning on and off and adjusting the intensity of the red and IR LEDs.

45 2.

Differential DAC with gain to sample the output of the second stage photodetector amplification circuitry.

3.

DAC voltage output to the second stage non-inverting amplifier input for controlling the second stage amplifier output offset.

4.

DAC to the output of the first stage amplifier. These data are used for adjusting the intensity of the IR and red LEDs. Briefly, the pulse oximeter software turns on the IR LED and applies a voltage

offset to the second stage amplifier, waits for the processes to settle, takes measurements of the first and second amplifier outputs, turns the IR LED off and removes the offset. After that, the microcontroller checks if the output from the second amplifier is in the DAC measurable range. If data are not in range, the microcontroller uses the data from the first amplifier stage to calculate the new LED intensity for the next measurement. Then it does the same procedure with the red LED. If the red, IR, or both red and IR data are out of range, then new LED intensities are calculated. This procedure eliminates creating a software loop for adjusting the light intensity. It takes up to 10 to 20 s to get the signal in range when the wireless sensor is placed on a patient or when the surrounding conditions change for a prolonged time. But, in this way, attempts to adjust the LED intensities to accommodate out of range signals caused by short duration motion artifacts are minimized. The signal from the second stage amplifier is filtered and is used to calculate the heart rate and percent oxygen saturation in the blood. After that, based on the user setting, the filtered wave data or calculated heartbeats and percent oxygen

46 saturation are sent to the transceiver to be transmitted to the PC or saved on the wireless sensor microSD card.

4.4.2. Controlling the LEDs and Taking Data Samples Every 10 ms, if the oximeter is active, the pulse oximeter routine starts with connecting the IR LED’s anode to a high voltage, by reconfiguring the corresponding microcontroller control pin from input to a pulled up output. The output voltage from the DAC is connected to the common LED anode. The decrease in the DAC voltage causes an increase in LED intensity. The voltage from the second DAC channel is applied to the second stage non-inverting amplifier input for adjusting the output offset. After that, the microcontroller is put in “Sleep” mode for 500 µs. The on/off red and IR LED sequences are shown in Figure 4.3. The voltage rise on the output of the first stage in response to the LED light is shown in Figure 4.4. A timer-interrupt wakes up the microcontroller after 500 µs. After this time, all the processes settle. The LED light and the amplifier output voltages are settled. The microcontroller samples the outputs of both amplifier stages. The second stage amplifier is sampled by the ADC configured as a differential input with gain. For the differential input configuration, the microcontroller has 1x, 2x, 4x, 8x, 16x, 32x, or 64x options for software selectable gain. In the project, the gain is adjusted based on the maximum wave swing, with a maximum gain limit of the 8x.

47

Figure 4.3. IR (bottom) and red (top) LEDs on/off sequence

Figure 4.4 First stage amplifier settling time

48 The code snippet for controlling the red LED and sampling both amplifiers is dacb_write(1,Roffset); //apply v offset to the non-inverting input, second stage dacb_write (0,Rintensity);//apply v to LED’s cathode R_LED_ON; //apply high voltage to red LED anode SET_t_out_us (500); //set time out idle (); //set microcontroller in idle ad_R_data= adca_read1 (0);//sample second stage amplifier in_level_R= adca_read (3); //sample first stage amplifier R_LED_OFF; //turn red LED off

The ac component of the useful signal from the first stage has a swing of 3-10 ADC units. This swing does not affect the measurement for detecting the dc component, which could be up to 300 times bigger. For example if the measured signal is 3000 ADC units, with a useful signal swing of 3-10 ADC units, 2990 ADC units can be safely removed without affecting the useful signal.

Figure 4.5. First (top) and second (bottom) stage amplifier outputs

49 The code snippet for calculating the light intensity for the red LED is if (in_level_R>top) { if (Rintensity0; n--) { x[n] = x[n-1]; y[n] = y[n-1]; } //Calculate the new output x[0] = NewSample; y[0] = ACoef[0]; y[0] *= x[0]; for(n=1; n s_max_ir) s_max_ir = ad_IR_data; if (ad_IR_data < s_min_ir) s_min_ir = ad_IR_data;. For determining the number of heartbeats per min., the IR signal is passed through a highpass first-order IIR digital filter and then rectified. The code for the highpass filter and rectifier is hpf_data = (hpf9 (ad_IR_data)); if (hpf_data < 0) hpf_data = 0; //rectifier int hpf9 (int NewSample) { static signed long int y[]={2024,2024};; static signed int x[]={2024,2024};; const int ACoef[] = { 25391,-25391 }; const long int BCoef[] = { 32768, -18014 }; long int A,B; x[1] = x[0];

52 y[1] = y[0]; //Calculate the new output x[0] = NewSample; y[0] = ACoef[0]; y[0] *= x[0]; A=ACoef[1]; A*=x[1]; B= BCoef[1]; B *= y[1]; y[0]+= A - B; y[0] /= BCoef[0]; return (y[0]); }. For visualizing the shape of the highpass filter, the signal was temporarily redirected to the temperature channel, and the waveform is shown in Figure 4.7.

Figure 4.7. Pulse detecting signal (on T channel)

4.4.4. Calculating the Heartbeats The resulting signal, obtained after highpass filtering and shown in Figure 4.7, is very clean, and it is easy to calculate the heartbeats. The code snippet is

53 time++;//10ms period if (hpf_data>threshold && detect_beat==0) { detect_beat=1; beats++; } else if(hpf_data==0 && detect_beat==1) { detect_beat=0; beat_min.=6000/time; // (1/ (t*10ms)*1000ms/s*60sec/min time=0; }. For the signal depicted in Figure 4.7, the heartbeat calculation is correct for a threshold level of zero. For the signal depicted in Figure 4.6, the threshold should be higher to eliminate the signal “bumps” that appears in between heartbeats. To suppress signal artifacts created by motion, a more complicated algorithm for detecting noise and calculating the heartbeats has been implemented. The algorithm analyzes the signal to detect heartbeats, noise, and out-of-range signal. An out-of-range signal is a signal that exceeds the minimum or maximum allowable signal duration. The maximum allowable period of a signal is calculated from the heartbeat frequency in (0.3) and is

TMAX =

1000 ms = 2000 ms . 0.5 Hz

(0.5)

If no pulse is detected and TMAX is exceeded, then it is possible that the wireless sensor is removed from the patient or the patient is in a crisis. Therefore, exceeding TMAX is always a good reason for setting off the alarm.

54 The minimum signal period is calculated from the heartbeat frequency in (0.4) and is

TMIN =

1000 ms = 222 ms . 4.5 Hz

(0.6)

When looking for a pulse, the time up to TMIN can be used for determining the threshold of the noise in the signal. In this project, the threshold for heartbeat detection is the maximum value of the pulse detect signal during the 200 ms after a heartbeat is detected. Figure 4.8, depicts a heartbeat corresponding to an 80 beat/min. heart rate. The pulse detect signal maximum occurs 100 ms after a pulse is detected. If a maximum occurs after 200 ms, then the pulse is considered to be contaminated with noise and is discarded. Each valid detected pulse is passed through a second-order averaging filter, and after detecting 10 heartbeats, the heartbeat rate is averaged.

350 300

IR heart pulse and pulse detect

250 200 150 100 50 0 40 80 120 160 200 240 280 320 360 400 440 480 520 560 600 640 680 720 760 800 840 880 920 960 1000 1040

0

Figure 4.8. IR heart pulse (top) and pulse detect (bottom)

ms

55 4.4.5. Calculating the Percent Oxygen Saturation in Blood All the calculations are done by the microcontroller. Calculating the ratio of logarithmic normalized intensities as in (2.1) could be a difficult task for a microcontroller. The software for oxygen saturation calculation is based on the relationship of oxygen saturation to the ratio of red to IR [29]. Thus, instead of calculating the ratio of logarithmic normalized intensities, the ratio of normalized red and IR values obtained by the pulse oximeter are used. That is,

R

VRED , VIR

(0.7)

VR

ACRED DCRED

(0.8)

VIR

ACIR . DCIR

(0.9)

where VR and VIR are, respectively,

and

Figure 4.9, [29] shows the ac and dc components of the signal. Then the ratio formula is

R

ACRED / DCRED . ACIR / DCIR

(0.10)

56

Figure 4.9. The ac and dc components of the IR signal. The figure shows that the dc component for the red and IR signals are not equal [29]. This difference in the dc component of the signal is caused mainly by the different light intensities of the LEDs and by the change in photodiode sensitivity with wavelength.

From Figure 4.9, the dc value is the minimum value of the ac component. As the minimum and maximum values of each signal are collected, the minimum and maximum values can be substituted in the ratio formula,

R

ACRED / DCRED ACIR / DCIR

R

( R max R min) / R min ( IR max IR min) / IR min

( R max R min) IR min . ( IR max IR min) R min

(0.11)

(0.12)

The signal of the second stage amplifier output, shown in Figure 4.5, decreases with the increase of the light intensity and is inverted in comparison with the signal shown in Figure 4.9. The ac portion of the signal remains the same, while the dc portion is

57 calculated as the difference between the high level voltages corresponding to no-light level and collected maximum values. Therefore, for calculating the dc portion of the signal, the voltage of the second stage amplifier is measured when the LEDs are off. Then the formula becomes

R

( R max R min) (Vdark IR max) . ( IR max IR min) (Vdark R max)

(0.13)

The oxygen saturation is converted from R using the curve shown in Figure 4.10 [29].

Figure 4.10. Percent oxygen saturation versus R/IR ratio. The figure shows the empirical calibration curve [29]. The 70% to 100% oxygen saturation range, which is of the interest in this project, can be approximated by a linear equation.

58 In practice, the

SPO2

a bR

(0.14)

clinical empirical formula for the oxygen saturation is used. The a and b are coefficients are determined when the pulse oximeter is calibrated and could be determined by performing the linear fit of the R values of multiple samples using the least squares method. Because oxygen saturation below 70 % is not of interest, then the a and b coefficients are 110 and 25 respectively [36]. The software snippet for calculating the oxygen saturation is arr = ((long)lpf_dataIR+offset)/(lpf_dataR+offset);// IR/R R = (((long) (s_max_r-s_min_r))*arr)/ (s_max_ir-s_min_ir); arr = 110 -25*R; Sp_O2_mf[rr_array_counter] = OxiMovingAverage(arr); Sp_O2 = median_filter (Sp_O2_mf); .

The software flow diagram for pulse oximeter acquisition is shown in Figure 4.11.

4.1. MicroSD Card Software The option to save on the wireless sensor microSD card can be activated by the GUI and is used for continuous monitoring when the base-station and the PC are out of RF range. The flow chart of the microSD card function is shown in Figure 4.12.

59 START SUPPLY OFFSET VOLTAGE TO THE 2ND STAGE AMPLIFIER TURN INFRARED LED ON WAIT 500 ms ADC 1ST AND 2ND STAGES ACQUISITION FOR IR WAVE TURN INFRARED LED OFF SUPPLY OFFSET VOLTAGE TURN RED LED ON WAIT 500ΜS ADC 1ST AND 2ND STAGES ACQUISITION FOR R WAVE TURN RED LED ON

YES

NO

IS THE R AND IR WAVE IN RANGE?

ADJUST R AND IR LEDS INTENSITY LPF(IR AND R DATA) HPF(LPF IR DATA) PEAK DETECTING

NO

IS NOISE DETECTED?

YES RESTART PULSE COUNTING

HEART RATE CALULATION SPO2 CALCULATION RETURN

Figure 4.11. Pulse oximeter function

60

START

NO

IS 512BYTE BUFFER FULL?

YES

IS DIRECTORY CREATED?

NO

CREATE DIRECTORY

YES

YES

IS FILE OPENED? NO

YES

CREATE/OPEN FILE

IS FILE SIZE < SET SIZE?

NO

CLOSE FILE

IS DIRECTORY SIZE < SET SIZE?

YES

WRITE FILE NO CLOSE DIRECTORY RETURN

Figure 4.12. MicroSD card function

The microSD function creates files of fixed length and directories that contain a fixed number of files. The maximum possible number of directories on the microSD card is 256, and each directory contains 256 files with the exception for the last directory. Created directories contain numeric names from 0 to 255. Created files contain in the file name the “LOG_” followed by a three-digit number, followed by “.TXT”. If 256 directories have been created, then in the last directory, the microSD function writes

61 beyond 256 files. The name of this and following files in that directory is “LOG”, followed by a four-digit number. The data written to the card are in binary format and need a binary-to-ASCII converter, which has been written in Java on a PC. Each saved file includes a header that is compatible with “LabChart 7”, data acquisition and analysis software from ADInstruments. The microSD card internally writes the data in chunks of 512 bytes, which is the microSD card sector size, regardless of whether the data size passed to microSD card is smaller. To decrease power consumption, data are kept in the microcontroller in a 512byte buffer. When the buffer is full, the microSD card is accessed, and the data are written to the microSD card.

4.2. Wireless Communication The transceiver is programmed to transmit/receive at 200 kbps at 915 MHz. It has a 64-byte transmit buffer. In active mode, the data are transmitted every time the buffer is full. If, based on the wireless sensor settings, the buffer fills up in more than 1 s, then the wireless senor transmits the buffer contents once a second. The wireless sensor queries the base-station for a command once a second. The first byte in the transmit data string indicates if the wireless sensor expects communication from the base-station. If there is no communication with the base-station for 18 s, then the wireless sensor assumes that

62 the base-station is not in RF range and switches to “Sleep” mode. Transmitted data consist of 12 bits of data from a particular sensor plus a four-bit sensor identifier. In “Sleep” mode, the transceiver, configured as a receiver, is enabled for 4 ms every 2 s and waits for incoming communication. If the wireless sensor is not addressed, then it remains in “Sleep” mode. If the sensor is addressed, then it executes the command it receives.

63

CHAPTER 5 5. GRAPHICAL USER INTERFACE (GUI) The GUI has been written in Java. Via the GUI, a user can control, retrieve, display, and save the data from a specific wireless sensor on a PC. For controlling a wireless sensor, the sensor needs to be initialized. 5.1. Initializing a Wireless Sensor Initialization is the process of detecting the wireless sensor, assigning it a specific name, which can be the patient’s name, and saving that information in the GUI database. To do this, the user presses the scan button on the GUI. During this process, the GUI issues wake-up and scan commands. The wireless sensor, which is in “Sleep” mode, checks for the incoming command every 2 s. When the wireless sensor receives the wake-up command, it continues to wait for the next command, and when it receives the scan command, the wireless sensor sends its ID. 5.2. Controlling a Wireless Sensor A user can change, via the GUI, the wireless sensor mode to “On” or “Sleep”, activate/deactivate the accelerometer and the pulse oximeter and change the accelerometer, pulse oximeter, and temperature sampling rates. The sampling interval of any sensor can be change from 10 ms to 2550 ms, in 10 ms step. The pulse oximeter sampling interval affects the data transmitted from the wireless sensor. The on-board

64 pulse oximeter sampling interval, if it is activated, is always 10 ms, regardless of the GUI-specified pulse oximeter transmission sampling interval. The GUI allows grouping multiple wireless sensors for timed acquisition control. Figure 5.1 shows the GUI timed acquisition control panel. The number of wireless sensors grouped in timed acquisition mode depends on the time any wireless sensor can safely not be monitored.

Figure 5.1. GUI timed acquisition control panel

For example, three wireless sensors (see Figure 5.2) were grouped for timed acquisition. Assume that the “On” and “Sleep” modes were both set to 1 min., which actually is the minimum time that can be set. Then, the first wireless sensor (John) in the list will be on for 1 min., then, after 1 min., only the second sensor (Mia) is “On”, and so on. After the last sensor in the list, all implants are in “Sleep” mode for one min.

65

Figure 5.2. Wireless sensors grouped for timed acquisition

Therefore, each wireless sensor has been on for 1 min., and in “Sleep” mode for 3 min. It is possible to include in the list the same wireless sensor multiple times, thus allowing minimizing the “Sleep” time and maximizing the “On” time for a certain wireless sensor. For example, in the group of sensors in Figure 5.3, the wireless sensor Alexander is “On” for 1 min. and in “Sleep” for 1 min., while wireless sensor John is “On” for 1 min. and in “Sleep” for 3 min., for the same 1-min. “On” and 1-min. “Sleep” timed acquisition settings. This feature is useful when one sensor needs more frequent monitoring. Figure 5.3 shows an example of increasing the time “On” for a specific wireless sensor.

Figure 5.3. Wireless sensors grouped for timed acquisition for different On/Sleep times

66 5.3. Displaying Sensor Data The GUI has a monitor window that displays the data received from a wireless sensor. The following functions are available: manual and auto amplitude adjusting to fit in the display frame and displaying specific data on the entire GUI monitor frame. The display is useful for positioning the wireless sensor on a person and for real time monitoring. 5.4. Saving Sensor Data The GUI allows saving data received from the wireless sensor on the PC. To enable the function, the user clicks the “Data Logger is Off” button, which turns into “Data Logger is On”. When the data logger is activated, the text on the button (see Figure 5.4) changes color between red and blue every second. Also the data display background changes. The data are saved in ASCII format with a header compatible with “LabChart 7” for further viewing and analysis. Figure 5.5 shows saved data plotted in Excel, a spreadsheet program from Microsoft (Redmond, WA). A new file is created on a daily basis, and the saved file contains the wireless sensor name and the date.

67

Figure 5.4. Display monitor. The figure shows the GUI monitor panel with the Data Logger function on. The signals received from the wireless sensor and displayed on the GUI monitor are saved on the PC. The pulse detect signal was temporarily redirected to the temperature signal.

68 1900 1800 1700 1600 1500 1 52 103 154 205 256 307 358 409 460 511 562 613 664 715 766 817 868 919 970 1021 1072 1123 1174 1225 1276 1327 1378

1400

Figure 5.5. Plotting saved data in Excel. The figure depicts the plot of the received wave signal from the wireless receiver (red pulse oximeter channel)

69

CHAPTER 6 6. BASE STATION INTERFACE SOFTWARE The base-station is always in receiving mode, except when an activated wireless sensor requires communication. When the base-station’s wireless sensor receives data, it generates an interrupt. While waiting for an interrupt, the base-station’s microcontroller checks the USART for data from the PC. Received PC data are stored in the microcontroller and then transmitted to the wireless sensor when the sensor queries for changes in settings. An activated wireless sensor transmits data when the 64 byte buffer is full or if a transmission did not occur for 1 s. The transmission rate depends on the user-set sensor rate. The temperature is sampled once every 30 s. Continuous waveform transmission can be useful for wireless sensor placement. For a person at rest, heart rate and percent oxygen saturation in the blood also can be transmitted once every 30 s. If the accelerometer is set for detecting low and high activity levels, then the RF transmission rate is low, therefore lowering power consumption. Once a second, the activated wireless sensor queries the base-station for changes in settings. The query code is included in the header transmitted with the sensor data.

70

CHAPTER 7 7. TESTING RESULTS The GUI – base-station – wireless sensor communication link is operational, and the wireless sensor responds to GUI commands to change “Sleep” and “On” modes. The GUI monitor panel displays the data received from the wireless sensor. It includes the data from X, Y, and Z accelerometer axes, the red and IR pulse oximeter wave or number of heartbeats and percent oxygen saturation in blood, and the temperature. The GUI temperature monitor displayed 23.7 °C received temperature data, which corresponded to ambient temperature. The temperature received from the wireless sensor was verified against a TPI 310 digital thermometer from Test Products International (Beaverton, OR) and was correct to the first decimal place. The GUI monitor for each accelerometer axis correctly displayed the changes in the received waveform in response to sensor motion in the vertical direction and in the two horizontal directions. The wireless sensor was rotated around three axes. Each complete rotation caused the data to change over a ± 1 g range. The accelerometer was tested for a standing, waking, standing, and jumping sequence. The results, saved by the GUI, and plotted in Excel are shown in Figure 7.1. The accelerometer was positioned with the Z axis horizontal and in the direction of movement, the Y axis vertical, and the X axis horizontal and perpendicular to the direction of movement. The number of steps and number of jumps can be counted from the Z axis data,

71 The detected number of heart beats was checked against a Nellcor Pulse Oximeter from Covidien (Mansfield, MA). The heartbeat range measured by the Nellcor Pulse Oximeter was from 74 to 81 beats per min. The number of heartbeats measured by the wireless sensor was in the same range. When measured simultaneously with both devices, the difference in detected heartbeat number was from one to three heartbeats. The wireless sensor was tested on the workbench by placing a finger over the photosensor. While monitoring the pulse oximeter signals on the GUI monitor, it was observed that the pulse oximeter was adjusting to the external changes in light, but the signal was disrupted in response to hand movement, during which the heartbeat rate update occurred more slowly. The data for oxygen saturation in blood were tested for one point only. The data were in the same range as the data for the oxygen saturation obtained from the Nellcor Pulse Oximeter. The data logging on the microSD card was activated for a few minutes. Then the microSD card was removed and checked on a PC. As expected, the microSD contained a new folder “LOG 001”, which contained a file “LOG001.txt”. The binary file was converted to ASCII using the converter written for this purpose. The data from the wireless sensor were recorded. The microSD data logger was tested for longer periods of time with test data. The PC data logging was enabled for a few minutes, after which it was disabled. This process was repeated several times. Then, the file was opened. The data, as expected, were saved in ASCII format. The time the PC data logger was enabled was properly recorded. The record contained a header compatible with “LabChart 7”.

1 57 113 169 225 281 337 393 449 505 561 617 673 729 785 841 897 953 1009 1065 1121 1177 1233 1289 1345 1401 1457 1513

72

4000

3500

3000

2500

2000 x

1500 y

z

1000

500

0

Figure 7.1. Accelerometer data (three axis)

73

CHAPTER 8 8. SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS The HSMS was designed for continuous noninvasive monitoring of a person’s health state. The importance of continuous monitoring is that health problems are detected when they first occur, before adverse effects develop. Treatment at an early stage is most effective and less costly than treatment of an advanced health problems. The system incorporates multiple types of non-invasive sensors that give a general assessment of a person’s cardio-respiratory status, temperature, and level of physical activity and do not require a professional health care provider for implementation. The HSMS is configurable to accommodate a wide variety of applications. It could be used in emergency rooms and in the hospital for monitoring post-operative patients to improve patient safety outcomes and reduce health care cost caused by undetected patient health degradation. In the home, it will reduce unnecessary routine physician office visits, while increasing health-care quality, as collected data can be sent to a physician for a daily review. Continuous health monitoring allows elderly to feel more comfortable when living independently. The system could be used for monitoring people with specific health problems, such as people with epilepsy. Another in-home use of the system is for parents monitoring infants for preventing SIDS. For this or similar applications, where data collection is not necessary, the system can be configured to activate a sound and/or light alarm when an abnormality, is detected.

74 Wireless communication with a PC allows a person mobility, eliminating the inconvenience of being strapped to a stationary monitoring system. The wearable wireless sensor is small enough to be implemented as a patch. The data saved on a PC can be imported into different acquisition software programs for complex analyses. For applications where a computer is out of wireless communication range, such as monitoring the effect of a stressful situation on the health of first responders, the data are saved on a microSD card for later review. Activity monitoring in correlation with other vital data the wireless sensor provides can help in establishing a safe level of physical exercise. Activity monitoring also can be used for fall detection, balance control evaluation, and determining walking. The wearable wireless sensor samples temperature at 30-s intervals and provides temperature data that have a -0.5, +0.8 °C accuracy (in the 36-40 °C range) and 0.125 °C resolution. The activity is monitored by a three-axis 12-bit accelerometer with maximum sampling rate of 100 Hz. The activity sensor can be set to detect falls and out-of-range (below and above a set threshold) activity and can be set for continuous monitoring. The pulse oximeter sensor provides the number of heartbeats/min and the percent oxygen saturation in blood in the 30 to 270 beats/min. The sampling interval of any sensor can be change from 10 ms to 2550 ms, in 10 ms step. The data from the wireless sensor can be saved and viewed in real time on a PC or can be saved on a microSD if a computer is out of RF range. The wireless sensor can be set to issue a light and/or sound alarm if an abnormality (fall, activity out of range, low or no heartbeat) is detected.

75 Each function of the wireless sensor was tested and worked satisfactorily in the laboratory environment, but the system was not tested for continuous monitoring in a clinical environment. The wireless sensor has a small size (38 x 30 x10 mm) but is not enclosed in a case to be applied as a patch. The described benefits of implementing the system are based on the results of long-term monitoring studies in the literature. An important feature of the wireless sensor is that it contains multiple noninvasive sensors. Multiple sensors provide a more complex status of a person health and can help in identifying measurements errors. For example, motion artifacts can disrupt the pulse oximeter data, but the activity sensor will show movement. The temperature and activity sensors performed better than the pulse oximeter sensor. The temperature sensor provided accurate temperature (-0.5, +0.8 °C accuracy). From accelerometer data, it was easy to distinguish among standing, walking, and jumping. For future versions, the RF communication can be changed to Bluetooth wireless technology, because the new Bluetooth 4.0 is developed for low-power and up to 50 meters communication range. It will be easier to organize sensor networks controlled remotely from a PC or from mobile devices. For the wireless sensor, a function for updating the real time clock should be developed. It will be used by the microSD function for time-stamping saved data. The communication with the microSD card can be changed from the SPI port to the microcontroller USART port configured for SPI. This change will allow writing to the microSD card using the microcontroller DMA, which decreases the load on the microcontroller and saves power. The pulse oximeter sensor should be tested with more photodetectors (the circuit is designed to accommodate up to three

76 photodetectors but was tested with one only), which could improve light signal reception. For the light/sound alarms, a method for turning the alarm off needs to be developed for situations when the sensor is operating out of PC range. The LED for the light alarm could be changed to a through-hole type for better light visibility.

77 REFERENCES [1]

C. Blydenburgh, “Increase in Healthcare Demands Due to the Aging Baby Boomers,” Yahoo! Contributor Network. [Online]. Available: http://voices.yahoo.com/increase-healthcare-demands-due-aging-baby11832625.html. [Accessed: 15-Nov-2012].

[2]

D. Rodriguez, “Preventing Epilepsy Seizures - Epilepsy Center EverydayHealth.com.” [Online]. Available: http://www.everydayhealth.com/epilepsy/preventing-epilepsy-seizures.aspx. [Accessed: 25-Oct-2012].

[3]

A. H. Taenzer, J. B. Pyke, S. P. McGrath, and G. T. Blike, “Impact of Pulse Oximetry Surveillance on Rescue Events and Intensive Care Unit Transfers,” Anesthesiology, vol. 112, no. 2, pp. 282–287, Feb. 2010.

[4]

Centers for Disease Control and Prevention, “CDC - Sudden Infant Death Syndrome (SIDS) and Sudden, Unexpected Infant Death (SUID) - Reproductive Health.” [Online]. Available: http://www.cdc.gov/sids/. [Accessed: 27-Oct-2012].

[5]

Y. Mendelson, R. J. Duckworth, and G. Comtois, “A Wearable Reflectance Pulse Oximeter for Remote Physiological Monitoring,” in Engineering in Medicine and Biology Society, 2006. EMBS’06. 28th Annual International Conference of the IEEE, New York, NY, 2006, pp. 912–915.

[6]

A. Tura, L. Quareni, D. Longo, C. Condoluci, A. van Rijn, and G. Albertini, “Wireless Home Monitoring and Health Care Activity Management Through the

78 Internet in Patients with Chronic Diseases,” Med Inform Internet Med, vol. 30, no. 4, pp. 241–253, Dec. 2005. [7]

“Remote & Wireless Patient Monitoring Markets  : Kalorama Information.” [Online]. Available: http://www.kaloramainformation.com/Remote-WirelessPatient-2645944/. [Accessed: 27-Oct-2012].

[8]

K. Robertson, “Intel-GE Device Gets FDA Green Light - Silicon Valley / San Jose Business Journal.” [Online]. Available: http://www.bizjournals.com/sanjose/news/2011/03/04/intel-ge-device-gets-fdagreen-light.html. [Accessed: 25-Oct-2012].

[9]

G. Kelly, “Body Temperature Variability (Part 2): Masking Influences of Body Temperature Variability and Review of Body Temperature Variability in Disease,” Altern Med Rev, vol. 12, pp. 49–62, 2007.

[10] J. M. S. Pearce, “A Brief History of the Clinical Thermometer,” QJM, vol. 95, pp. 251–252, 2002. [11] “Answer to: What is the Relationship Between Physical Exercise and Health?,” Answers.com. [Online]. Available: http://wiki.answers.com/Q/What_is_the_relationship_between_physical_exercise _and_health. [Accessed: 15-Nov-2012]. [12] K. Crowe, “Heart Damage Risk from Excess Endurance Training - Health - CBC News.” [Online]. Available: http://www.cbc.ca/news/health/story/2012/06/01/marathon-endurance-heart.html. [Accessed: 15-Nov-2012].

79 [13] P. D. Thompson, B. A. Franklin, G. J. Balady, S. N. Blair, D. Corrado, N. A. M. Estes, J. E. Fulton, N. F. Gordon, W. L. Haskell, M. S. Link, B. J. Maron, M. A. Mittleman, A. Pelliccia, N. K. Wenger, S. N. Willich, and F. Costa, “Exercise and Acute Cardiovascular Events Placing the Risks Into Perspective: A Scientific Statement From the American Heart Association Council on Nutrition, Physical Activity, and Metabolism and the Council on Clinical Cardiology,” Circulation, vol. 115, pp. 2358–2368, 2007. [14] DAILY MAIL REPORTER, “The Wrong Amount of Sleep ‘Can Age Your Brain By Up To SEVEN YEARS’,” Mail Online. [Online]. Available: http://www.dailymail.co.uk/health/article-1382665/The-wrong-sleep-age-brainSEVEN-YEARS.html. [Accessed: 25-Nov-2012]. [15] BeWellBuzz, “Insufficient Sleep Increases Risk of Stroke and Heart Disease,” BeWellBuzz News. [Online]. Available: http://www.bewellbuzz.com/newsroom/news/insufficient-sleep-increases-riskstroke-heart-disease/. [Accessed: 25-Nov-2012]. [16] B. F. Macdonald-Smith, “Can Sleep Deprivation be the Cause of Mental Illness?,” Telegraph.co.uk, 02-Mar-2009. [17] D. Freeman and J. Freeman, “Can Lack of Sleep Drive You Mad?,” The Independent. [Online]. Available: http://www.independent.co.uk/life-style/healthand-families/features/can-lack-of-sleep-drive-you-mad-1705954.html. [Accessed: 25-Nov-2012].

80 [18] M. Hoeffel, “Lack of Sleep Caused 250,000 Vehicle Accidents Last Year,” Examiner.com. [Online]. Available: http://www.examiner.com/article/lack-ofsleep-caused-250-000-vehicle-accidents-last-year. [Accessed: 25-Nov-2012]. [19] M. Lee, J. Kim, K. Kim, I. Lee, S. H. Jee, S. K. Yoo, “Physical Activity Recognition Using a Single Tri-Axis Accelerometer,” Lecture Notes in Engineering and Computer Science, vol. 2178, pp. 14–17, 2009. [20] M. Sollitto, “Automatic Fall Detection: Call for Help Automatically when Seniors Fall - AgingCare.com,” AgingCare. [Online]. Available: http://www.agingcare.com/Articles/automatic-fall-detection-calls-for-helpautomatically-143104.htm. [Accessed: 25-Nov-2012]. [21] C.-C. Yang and Y.-L. Hsu, “A Review of Accelerometry-Based Wearable Motion Detectors for Physical Activity Monitoring,” Sensors, vol. 10, no. 8, pp. 7772– 7788, Aug. 2010. [22] J H Eichhorn, “Pulse Oximetry Monitoring and Late Postoperative Hypoxemia on the General Care Floor,” J Clin Monit Comput, vol. 14, pp. 49–55, 1998. [23] G. Ingram and N. Munro, “The Use (or Otherwise) of Pulse Oximetry in General Practice,” Br J Gen Pract, vol. 55, pp. 501–502, 2005. [24] Y. Gang and M. Malik, “Heart Rate Variability Analysis in General Medicine,” Indian Pacing Electrophysiol J, vol. 3, pp. 34–40, 2003. [25] J. Nolan, P. D. Batin, R. Andrews, S. J. Lindsay, P. Brooksby, M. Mullen, W. Baig, A. D. Flapan, A. Cowley, R. J. Prescott, J. M. M. Neilson, and K. A. A. Fox, “Prospective Study of Heart Rate Variability and Mortality in Chronic Heart

81 Failure Results of the United Kingdom Heart Failure Evaluation and Assessment of Risk Trial (UK-Heart),” Circulation, vol. 98, pp. 1510–1516, 1998. [26] D. M. Kallio, K. Suominen, A. M. Bianchi, T. Mäkikallio, T. Haapaniemi, S. Astafiev, K. A. Sotaniemi, V. V. Myllylä, and U. Tolonen, “Comparison of Heart Rate Variability Analysis Methods in Patients with Parkinson’s Disease,” Med. Biol. Eng. Comput., vol. 40, pp. 408–414, 2002. [27] L. Walsh, B. R. Greene, A. Burns, and C. N. Scanaill, “Ambient Assessment of Daily Activity and Gait Velocity,” in 2011 5th International Conference on Pervasive Computing Technologies for Healthcare (PervasiveHealth), 2011, pp. 418 –425. [28] R. G. Haahr, “Reflectance Pulse Oximetry Sensor for the Electronic Patch.” M.Sc. Thesis. Technical University of Denmark, Denmark, 2006. [29] J. T.B. Moyle, Pulse Oximetry, 2nd ed. BMJ Books,Tavistock Square,London, 2002. [30] A. Jubran, “Pulse Oximetry,” Crit Care, vol. 3, pp. R11–R17, 1999. [31] T. O’Dwyer, “Pulse Oximetry System Design.” Analog Devices. [32] J. G. Webster, Design of Pulse Oximeters. Taylor & Francis Group, New York, NY, 1997. [33] C. Valdez-Lowe, S. A. Ghareeb, N. T. Artinian, “Pulse Oximetry in Adults,” AJN, American Journal of Nursing, vol. 109, no. 6, pp. 52–59, Jun. 2009. [34] Ledtronics, “SML12R3KIR941T-TR Datasheet.” Ledtronics, Mar-2007.

82 [35] Vishay Semiconductors, “VBPW34S Datasheet.” Vishay Semiconductors, Aug2011. [36] J. E. Scharf, S. Athan, and D. Cain, “Pulse Oximetry Through Spectral Analysis,” in Biomedical Engineering Conference, 1993, Proceedings of the Twelfth Southern, 1993, pp. 227 –229.

Smile Life

When life gives you a hundred reasons to cry, show life that you have a thousand reasons to smile

Get in touch

© Copyright 2015 - 2024 PDFFOX.COM - All rights reserved.