Electronic prototyping for cardiovascular health monitoring and postural analysis

MASTER'S THESIS

Electronic prototyping for cardiovascular

health monitoring and postural analysis

Dries Agten

2013

Master of Science (120 credits)

Space Engineering - Space Master

Luleå University of Technology

Thesis for the Degree of Master of Science in Space Techniques and Instrumentation & Master of Science with Major in Space Technology

Erasmus Mundus Master of Space Science and Technology

-Electronic prototyping for cardiovascular

health monitoring and postural analysis

DRIES AGTEN

September 2013

This thesis was carried out in the framework of

the Erasmus Mundus Master of Space Science and Technology with the trajectory

Julius-Maximilians-University Würzburg (Germany) Luleå University of Technology (Sweden) Université Paul Sabatier Toulouse III (France)

Supervisor: Prof. A. Blaber

Aerospace Physiology Laboratory

Department of Biomedical Physiology & Kinesiology Simon Fraser University

Examiners: Prof. P. von Ballmoos

Université Paul Sabatier Toulouse III Dr. V. Barabash

Luleå University of Technology

c

Dries Agten 2013

All rights reserved. However, in accordance with the Copyright Act of Canada, this work may be reproduced, without authorization, under the conditions for ‘Fair

Dealing’. Therefore, limited reproduction of this work for the purposes of private study, research, criticism, review and news reporting is likely to be in accordance

Abstract

The microgravity in space affects the human body in many different ways, leading to postflight problems such as orthostatic intolerance and postural instability. Continuous monitoring is therefore becoming increasingly important, especially when related to the cardiovascular and vestibular system. Weight scales are ideal candidates for this type of monitoring due to their low cost and ease of use. In this thesis, the electronic design and experimental verification of a weight scale capable of cardiovascular health monitoring and postural analysis is presented. Ballistocardiography (BCG) circuits use strain gauges, amplifiers and filters to measure the forces exerted by accelerating blood onto the body, while an electrocardiography (ECG) circuit records the electrical activity of the heart with an innovative two-electrode set-up integrated with the scale. These circuits can be used to assess cardiovascular health. A two-electrode electromyography (EMG) circuit to monitor the electrical activity of the lower-body muscles may also be used in this system. A center of pressure (COP) circuit is described, based on a combination of analog-to-digital and digital-to-analog conversion. The EMG and COP circuits can be used in postural analysis. Designs for each of these circuits are presented, together with the design of a dedicated power supply circuit. The BCG, ECG and EMG circuits were assembled and tested, while the COP and power supply circuits are preliminary designs that will be assembled in the future. Tests with the BCG, ECG and EMG circuits provided signals corresponding to previous measurements with standard equipment. When combined with COP and power supply circuits, the systems developed in this thesis can be used for cardiovascular health monitoring and postural analysis, which has applications in both space physiology and clinical environments.

Keywords: BCG, COP, ECG, EMG, Space physiology, Weight scale.

Acknowledgements

Although there is only one name on the cover, this thesis is not the work of a single person. I would therefore like to take this opportunity to acknowledge those who contributed knowingly and unknowingly to my thesis. First of all, my gratitude goes towards everyone in the Aerospace Physiology Laboratory of the Simon Fraser University. Towards Andrew, for supervising me during six months, and towards Carole, for setting this incredible experience into motion, but also towards Alex, Kouhyar, Da and Anna for valuable guidance, advice and standing on a scale for minutes on end. From the BPK Department of the Simon Fraser University I would like to thank Joe and King for their help with both electronics and general handiwork, Susie and Laura for handling the paperwork and, finally, everyone else I met for making my excursion into the wonderful world of physiology and kinesiology something to remember.

This thesis would not have been written if I had not been admitted into the SpaceMaster program. I would therefore like to express my gratitude towards the people who have made my SpaceMaster adventures possible. First, my thanks go to Dr. Victoria Barabash and the consortium, for selecting me for this wonderful program and helping me out during difficult administrative episodes. Second, I would like to thank the program coordinators at the Uni-versité Paul Sabatier Toulouse III, Prof. Christophe Peymirat and Prof. Peter von Ballmoos, for providing valuable guidance during my stay in France. Also those who provided adminis-trative support should not be forgotten. Thank you, Ursula Shahmary, Heidi Schließuss, Anette Snällfot-Brändström, Maria Winnebäck, Maude Perier-Camby and Hélène Perea. Finally, I want to acknowledge the financial support of the Erasmus Mundus programme of the European Union. On a personal level, my gratitude first goes towards the new friends I made here in Vancouver. ‘The blue house’ and all its wacky inhabitants will always be in my heart. Additionally, I would like to thank all the amazing people I met during my travels in Germany, Sweden and France. Of course an enormous ‘thank you’ goes out to all the Round 7 SpaceMasters. The past two years have been absolutely incredible and I am looking forward to the many exciting years to come. Finally, I would like to thank my family, my parents and my sister, for being there for me even though I am always far away. They made me into who I am today and they provided the basis for everything I have done so far, so I am extremely grateful for their support. I cannot promise that I will stay closer to home from now on, but rest assured that ‘home’ will always be the same place.

Vancouver, August 2013 Dries Agten

Contents

1.2 Continuous Health Monitoring . . . 2

1.3 Physiological Signals . . . 2 1.3.1 Ballistocardiography . . . 3 1.3.2 Electrocardiography . . . 4 1.3.3 Electromyography . . . 4 1.3.4 Center of Pressure . . . 5 1.4 Goal . . . 6 2 BCG Circuit 7 2.1 Literature Study . . . 7 2.2 Scale Selection . . . 7 2.3 Circuit Design . . . 8

3 ECG & EMG Circuit 10 3.1 Literature Study . . . 10 3.2 Circuit Design . . . 11 3.2.1 ECG Circuit . . . 11 3.2.2 EMG Circuit . . . 12 4 COP Circuit 13 4.1 Circuit Requirements . . . 13 4.2 Circuit Design . . . 13

5 Power Supply Circuit 16 5.1 Circuit Requirements . . . 16 5.2 Circuit Design . . . 16 5.2.1 Boost Circuit . . . 16 5.2.2 Inverter Circuit . . . 17 5.2.3 Isolation Circuit . . . 18 6 Experimental Verification 19 6.1 BCG Circuit . . . 19

6.2 ECG & EMG Circuits . . . 19

6.3 Calibration . . . 22

6.3.1 Set-up & Protocol . . . 22

6.3.2 ECG Signals . . . 22

6.3.3 Center of Pressure . . . 24

7 Conclusions & Outlook 26 7.1 Conclusions . . . 26

7.2 Outlook . . . 26

List of Figures

1.1 Typical BCG recording . . . 3

1.2 Typical ECG recording . . . 4

1.3 Typical EMG recording . . . 5

1.4 Sketch of a platform-based set-up to calculate COP . . . 6

2.1 Tanita BC-551 weight scale . . . 8

2.2 BCG circuit schematic . . . 9

3.1 ECG circuit schematic . . . 11

4.1 COP circuit schematic . . . 14

5.1 Boost circuit schematic . . . 17

5.2 Inverter circuit schematic . . . 18

5.3 Isolation circuit schematic . . . 18

6.1 BCG recording from two strain gauges and ECG recording . . . 20

6.2 ECG recording during quiet standing . . . 21

6.3 EMG recording . . . 21

6.4 Protocol for calibration test . . . 22

6.5 Quiet standing during calibration test . . . 23

6.6 Custom and standard ECG recordings . . . 23

6.7 COP from weight scale . . . 24

6.8 Comparison between COP from weight scale and force plate . . . 25

List of Abbreviations

ADC analog-to-digital converter

BCG ballistocardiography/ballistocardiogram BIA bioelectrical impedance analysis

BNC Bayonet Neill-Concelman COP center of pressure

DAC digital-to-analog converter

ECG electrocardiography/electrocardiogram EMG electromyography/electromyogram ESR equivalent series resistance

IC integrated circuit LCD liquid-crystal display op-amp operational amplifier

SCG seismocardiography/seismocardiogram SMS space motion sickness

Chapter 1

Introduction

This first chapter establishes the required background and the main concepts of this work. The effects of the space environment on the human body and continuous health monitoring are briefly discussed in the first two sections. Section 1.3 defines all relevant physiological signals, while the final section presents the goal of the thesis.

1.1

The Human Body in Space

During spaceflights astronauts live and work in an extreme environment that is characterized by a set of unique physical features, including a reduced gravitational force (microgravity), a modified dark/light cycle (the time between sunrises is about 90 min instead of 24 h) and ionizing radiation (caused by galactic cosmic rays or solar particle events) [1]. In space, the human body has to adapt physiologically and psychologically to these new characteristics. One of the most important adaptations is related to the lack of gravitational force. Microgravity has an effect on several physiological systems, such as the vestibular/sensorimotor system (balance and movement coordination), the cardiovascular system (blood circulation) and the musculoskeletal system (muscles and bones) [1].

In microgravity the vestibular system no longer receives information on the direction of the head or body with respect to the vertical, but is limited to information about linear and angular body accelerations [1]. This change leads to the development of space motion sickness (SMS), similar to motion sickness on Earth [2]. Another major consequence of being in space is that astronauts suffer from postural instability during a postflight period of about a week. Even their gait (movement pattern) is influenced [2].

Microgravity manifests its effect on the cardiovascular system through a redistribution of body fluids (leading to less fluid in the trunk and legs) and a decrease in both blood and heart volume [1]. These phenomena do not significantly influence an astronaut during his/her time spent in microgravity, but the consequences are clear upon returning to Earth. For example, astronauts frequently suffer from orthostatic intolerance (the inability to tolerate an upright posture without experiencing symptoms related to fainting) [3] and a reduced effectiveness of the baroreflex (an important mechanism that maintains blood pressure by adjusting heart rate and vascular resistance) [4].

Finally, spaceflight has an effect on the musculoskeletal system through muscle atrophy (decrease in muscle volume and strength) and a decrease in bone mass (caused by bone demineralization). Astronauts are not significantly affected during spaceflights, but problems do arise when they return to the normal gravity of Earth. As a result of muscle atrophy, many astronauts have difficulties maintaining an upright position or moving effectively, while the decreased bone mass leads to an increased fracture risk [5].

2 CHAPTER 1. INTRODUCTION

1.2

Continuous Health Monitoring

The cardiovascular system is arguably the most important of the physiological systems discussed in the previous section, for both astronauts and patients in healthcare. Cardiovascular diseases are currently one of the main causes of death worldwide (they are responsible for 47% of all deaths in Europe [6]). It is clear that continuous cardiovascular health monitoring, including at home, is imperative to aid the elderly, patients with chronic diseases and recovering patients [7]. Data obtained via such a systematic monitoring system can then be used for space physiology research.

Both ‘wearable’ and ‘environmental’ sensors have been investigated as candidates for continu-ous (cardiovascular) health monitoring at home. One example of a wearable sensor is a ring to monitor blood oxygen saturation [8]. Environmental sensors, on the other hand, have been implemented in, for example, toilet seats (for blood pressure measurements) [9] and beds (to eval-uate sleep quality) [10]. A standard weight scale is now considered as an intermediate between wearable and environmental sensors. It is portable and hence not part of the environment, but it also cannot be worn [11]. Weight scales are generally low-cost devices that are easy to use and do not require a trained professional to perform measurements [12, 13]. They are also part of almost every household and are frequently used (even daily). All these factors make weight scales an ideal platform for continuous (cardiovascular) health monitoring [14].

While a scale can naturally be used to check the weight of the person standing on the scale (which can be sufficient for doctors to identify health risks [15]), the addition of other sensors can provide for the assessment of a wide range of important cardiovascular parameters. Two examples are the addition of electrocardiography/electrocardiogram (ECG) functionality (monitoring the elec-trical activity of the heart, see subsection 1.3.2) [14,16–18] and a measurement of the ballistocar-diography/ballistocardiogram (BCG) signal (forces imparted by the accelerating blood onto the body, see subsection 1.3.1) [15–19]. Some of the added sensors are not limited to cardiovascular functionality. For example, Inan et al. [15] measured electromyography/electromyogram (EMG) signals (the electrical activity of skeletal muscles, see subsection 1.3.3) from the legs and the center of pressure (COP) (see subsection 1.3.4) has also been measured to perform postural analysis [20, 21].

Recently, several efforts have been undertaken to combine some of these measurements in order to obtain multiple cardiovascular parameters at the same time. One example is the integration of ECG with BCG measurements [16, 17], and even EMG [14]. Scales combined with sensors capable of measuring cardiovascular signals and signals related to postural analysis would be equally useful for patients in in-home care and for astronauts during pre- and postflight tests. Such systems could even be used to assess the effects on the cardiovascular and vestibular system while performing a zero-g flight or living on the Moon [22].

1.3

Physiological Signals

Several physiological signals related to cardiovascular health monitoring or postural analysis can be measured using a weight scale in combination with specialized sensors. This section discusses the relevant signals for this thesis by examining their physiological origin, listing typical measurement techniques and presenting an example that can be used as a reference signal.

1.3. PHYSIOLOGICAL SIGNALS 3

1.3.1 Ballistocardiography

The BCG signal is generated from the blood being pumped around the body and is thus a cardiovascular signal. The human heart consists of four chambers, two upper chambers called atria and two lower chambers called ventricles. During an individual heartbeat (cardiac cycle), both atria contract first, filling both ventricles with blood, followed by a contraction of the ventricles. This ventricular contraction pumps blood into the lungs (right ventricle) and into the rest of the body (left ventricle). When the left ventricle contracts, blood is initially propelled (accelerated) upward into the aorta; however, at the aortic arch the aorta curves downward and the blood follows a downward path [23]. These upward and downward accelerations apply corresponding forces. According to Newton’s third law [24], every action is opposed by a reaction of equal magnitude and opposite direction. An upward force on the blood thus results in a downward force on the body and vice-versa. A BCG measurement records these reaction forces or their resulting displacement of the body [25].

BCG recordings can be obtained with many different devices, including suspended beds (the reaction forces displace the body and thus the bed) [25–27], force plates [28] and, the focus of this thesis, weight scales (the reaction forces influence the weight measurement) [15–19, 29, 30]. A typical BCG recording during a single heartbeat, obtained from a bed-based system, is shown in figure 1.1. H J I _K L M N

Figure 1.1: Typical BCG recording (bed-based). Modified from [30]

The characteristic features of the BCG signal reflect the different phases of the cardiac cycle [31]. Seven waves are visible, labelled H through N [27]. The first four waves (H, I, J and K) are related to the systolic phase of the cardiac cycle, corresponding to ventricular contraction and ejection of blood. The H wave is the first upward deflection following the R wave of the ECG signal if both signals are recorded simultaneously (also see subsection 1.3.2) [32]. This wave peaks at or near the end of the ventricular contraction and the beginning of ejection of blood into the aorta. I is the downward deflection following the H wave and corresponds to the maximal upward acceleration of blood in the aorta. Because the reaction force is recorded, it is visible as a downward deflection. J is the largest upward deflection immediately following the I wave and is related to the acceleration of blood in the descending aorta and the deceleration of blood in the ascending aorta as the blood impacts the aortic wall at the aortic arch [28, 31]. K is the downward deflection following the J wave and occurs near the end of systole [27]. The three following waves (L, M and N) correspond to the diastolic phase of the cardiac cycle, meaning relaxation of the ventricles and refilling with blood. L and N are two smaller upward deflections following the K wave, while the downward deflection in between is called M [27].

The useful bandwidth of the BCG signal is between 0.5 Hz and 10 Hz [13, 33], with expected amplitudes on the order of 1 N. However, the actual amplitude depends on many factors (weight, cardiac health, etc.) [25, 26].

4 CHAPTER 1. INTRODUCTION

1.3.2 Electrocardiography

The ECG signal is another cardiovascular signal and is related to the electrical activity of the heart. The cardiac cycle, consisting of systolic and diastolic phases, is controlled by a complex series of coordinated electrical events. The cycle starts when potential changes (action poten-tials) are created in the sinoatrial node, caused by spontaneous depolarization and repolarization of the membranes surrounding the heart muscle fibres of the node. Depolarization corresponds to contraction of the muscle fibres, while repolarization corresponds to relaxation. The action potentials originating from the sinoatrial node stimulate the two atria to contract almost sim-ultaneously. The signal then travels to the ventricles and causes them to contract from the apex to the base, expelling the blood and ending the cardiac cycle [23]. The potential changes caused by contraction and relaxation during the cardiac cycle are not measured directly when performing an ECG measurement. Instead, the potential changes give rise to ionic currents in the conducting fluid surrounding the muscle fibres and these ionic currents are recorded as electrical potential changes in electrodes attached to the skin [34].

Various electrode set-ups can be used to record an ECG signal. The most common set-up involves three electrodes placed on the left arm, right arm and left leg, with recordings performed between different combinations of electrodes. Figure 1.2 shows an example of an ECG signal during a single heartbeat, recorded in the Lead II configuration (between right arm and left leg, using the left leg as the positive electrode) [34].

P QRS T

Figure 1.2: Typical ECG recording (Lead II configuration). Example from [34]

Three characteristic features are visible on this figure: the P wave corresponds to the initial depolarization (contraction or systole) of the atria; the QRS complex is the second important feature and consists of a small downward deflection (Q), a narrow upward peak (R) and another small downward deflection (S); and is linked to both the repolarization (relaxation or diastole) of the atria and the depolarization of the ventricles; the T wave is the result of the repolarization of the ventricles. In between these features the heart is either fully depolarized (contracted) or fully repolarized (relaxed), which means that there are no potential changes across the membranes and no ECG signal [23].

A typical ECG signal has an amplitude of approximately 1 mV, while the bandwidth needed for diagnosis is 0.1 Hz - 100 Hz (allowing an accurate reproduction of the signal). For monitoring purposes (checking if the heart is beating) the bandwidth can be reduced to 40 Hz [34]. There is an overlap between the ECG bandwidth and the typical bandwidth of the EMG signal (10 Hz -5 kHz, see subsection 1.3.3), so electrical activity from muscles lying between the ECG electrodes can cause interference. Filtering the ECG signal or attaching the electrodes to the chest to avoid arm and leg muscle activity are common approaches to decrease the noise [34].

1.3.3 Electromyography

The EMG signal is linked to the electrical activity of both nerves and muscles, but the focus in this work is on muscle activity. EMG signals from leg muscles are related to postural analysis. The physiological processes giving rise to the EMG signal are very similar to the processes

1.3. PHYSIOLOGICAL SIGNALS 5

controlling the cardiac cycle (see subsection 1.3.2). Depolarization and repolarization of the membranes surrounding the muscle fibres lead to action potentials and the respective contraction and relaxation of the muscles.

The potential changes once again lead to ionic currents that can be measured by attaching electrodes to the skin over a muscle. Surface electrodes are most suitable for recording the overall activity of a (superficial) muscle for research purposes. In such a situation, the EMG signal is caused by all muscles between the two electrodes [34]. A typical EMG recording obtained from surface electrodes is presented in figure 1.3.

0.1 mV

10 ms

Figure 1.3: Typical EMG recording (surface electrodes). Example from [34]

It is very difficult to distinguish the individual spikes caused by motor units (the functional unit of muscle made up of a group of individual fibres [35]). Instead, the amplitude of the EMG signal is the sum of all generated action potentials. Since both positive and negative signals are possible (depending on the direction of the action potentials), the signals sometimes add and sometimes cancel. The resulting EMG signal looks very much like a random noise waveform, but the signal energy (the integral of the squared signal) is a function of the amount of muscle activity [34].

EMG signal amplitudes range from less than 1 µV to about 10 mV, depending on the size of the muscle. The typically measured bandwidth of the signal is 10 Hz - 5 kHz [34].

1.3.4 Center of Pressure

The COP is one of the most important parameters related to postural analysis. It is defined as the point of application of the ground reaction force. This force is the force exerted by the ground on the body in reaction to the force exerted by the body on the ground (Newton’s third law) [24, 36].

The COP is not directly measured, but rather calculated from the available information using different methods. The preferred method in this work requires four forces (F₁, F₂, F₃ and F₄), obtained from separate force sensors at the four corners of a platform [20]. Figure 1.4 on the next page illustrates the configuration of the force sensors and defines the relevant variables.

Chapter 2

BCG Circuit

In the second chapter of this thesis, the focus is on a cardiovascular signal: the BCG signal. Section 2.1 briefly discusses previous research on circuitry for BCG recordings. The second section focuses on the selection of an appropriate weight scale, while the final section concentrates on the design of the custom BCG circuit.

2.1

Literature Study

The first person to publish a BCG circuit design for an off-the-shelf weight scale was Williams in 1990 [19]. While the goal was to weigh human subjects with a very high accuracy, it was also noticed that the acceleration of the blood moving in the arteries showed up as a change in weight each time the heart of the subject pumped [19]. The circuit design consisted of a set of strain gauges (resistors that change resistance when pressure or force is applied) in a Wheatstone bridge configuration, a differential input amplifier and some filter stages. The total gain and bandwidth were not disclosed.

Whereas Williams mainly considered the BCG signal as undesired noise, González-Landaeta et al. [38] intentionally designed a BCG acquisition circuit a few years later. Their design included a Wheatstone bridge, an ac-coupled differential amplifier with a gain of 1000, a first-order low-pass filter with a cut-off frequency of 10 Hz and a second amplifier with a gain of 75. This set-up led to a total gain of 98 dB and a bandwidth of 0.1 Hz - 10 Hz. Consistent BCG signals were obtained for 12 volunteers with this set-up.

The most detailed description of a BCG acquisition circuit was given by Inan et al. [30, 39]. The set-up, largely based on the design by Williams [19], consisted of four strain gauges, configured as a Wheatstone bridge, and an analog circuit. The analog circuit included an instrumentation amplifier, a second-order low-pass filter and a non-inverting gain stage. ac-coupling was used to attenuate the dc-component of the weight. The resulting gain was equal to 81 dB, with a bandwidth of 0.1 Hz - 25 Hz. This system has been thoroughly and frequently tested since its first publication.

2.2

Scale Selection

The BCG acquisition circuits mentioned in the previous section were all interfaced with a stand-ard weight scale. The first step in the design of a custom BCG circuit was therefore the construc-tion or selecconstruc-tion of an appropriate scale. Developing a custom structure involved issues such as sensor alignment, platform stability and cost. Instead of performing this difficult and time-consuming task, an off-the-shelf weight scale was selected as the basis for this work. Following the lead of Inan et al. [30], a scale from Tanita was chosen, the BC-551 [40].

The BC-551 is a standard weight scale that also includes four electrodes for the measurement of bone mass, muscle mass and body water percentage through bioelectrical impedance ana-lysis (BIA). The scale has a liquid-crystal display (LCD), four buttons to personalize the

meas-8 CHAPTER 2. BCG CIRCUIT

urements, one on/off button and a switch to select the unit for the weight. The entire device is powered by two 3 V CR232 batteries. The BC-551 has a weight capacity of 150 kg and the smallest measurable weight increment is equal to 100 g [41]. Figure 2.1 shows a picture of the scale prior to modifications.

Figure 2.1: Tanita BC-551 weight scale

The sensors used to measure the weight of the person on the scale are strain gauges mounted on beams inside the supports. There are four supports and the four strain gauges are configured as a Wheatstone bridge. The four electrodes on the top surface of the scale (see figure 2.1) are made of an electrically conducting transparent coating. Each corner of the scale has one electrode to allow both current injection and measurement during BIA. The two lithium batteries of 3 V are connected in series, which doubles the voltage, but maintains the current capacity at about 210 mAh. All buttons are simple push buttons, but the unit selection switch is a single-pole triple-throw switch that allows the selection of three different units.

2.3

Circuit Design

During this thesis work, a custom amplification and filtering circuit was developed, based on the design by Inan et al. [39]. This design was selected as the basis for this work due to the details provided and the success achieved so far. The custom circuit for a single strain gauge consisted of the following stages:

1. Wheatstone bridge 2. Instrumentation amplifier 3. Low-pass filter

4. Non-inverting amplifier

Chapter 3

ECG & EMG Circuit

This chapter describes the acquisition circuits for the ECG and EMG signals. The first section presents a brief overview of previous ECG and EMG circuit designs. The custom circuits from this thesis work are then discussed in the second section.

3.1

Literature Study

Many researchers have recorded the ECG signal of a subject standing on a scale [13, 30]. In one study [18], three plantar (referring to the bottom of the feet) electrodes on a scale were used: two active electrodes (one under each foot) and one reference electrode. Each active electrode was first connected to a buffer to reduce the power line interference. The second stage, a first-order band-pass filter (0.5 Hz - 100 Hz), was followed by a differential amplifier with a very high gain (92 dB) and another first-order band-pass filter (0.5 Hz - 40 Hz). The bandwidth corresponded to a typical ECG monitoring bandwidth (see subsection 1.3.2). The high gain was required due to the very low ECG signal amplitude between the feet as compared to a normal ECG measurement from the chest. The recorded ECG signals were corrupted by EMG noise caused by muscle activation in the legs and feet during standing. As mentioned in subsection 1.3.2, the bandwidths of ECG and EMG signals overlap, making it difficult to separate the two contributions. Averaging and realignment techniques were applied to solve this issue [18]. Inan et al. [42] have developed an interesting ECG acquisition circuit based on a novel transim-pedance amplifier requiring only two electrodes. The use of two electrodes minimized the amount of electrodes on the patient. The differential ECG voltage at the skin created a current via the resistance of the skin-electrode interface (Ohm’s law). The current was then fed into the transimpedance stage of the circuit, which provided an amplified output voltage proportional to the input current. This output voltage was connected to an active current feedback stage that injected current into one of the input electrodes to stabilize the circuit. The gain was equal to 48 dB, while the bandwidth extended from 0.05 Hz to 200 Hz (a diagnostic bandwidth, see subsection 1.3.2). The circuit was tested together with a standard ECG circuit and provided very accurate measurements. The same system was later used by the same group in combination with hand-held electrodes and BCG measurements [14]. Also, when placed on each leg, the ECG electrodes could be simultaneously used for leg EMG.

While ECG and EMG signals are very similar electrically, very few researchers have investigated the measurement of EMG signals from a subject on a weight scale. A notable exception is Inan et al. [15]. The specific goal of the EMG measurement was flagging noisy BCG measurements. If a subject on the scale was moving excessively (as registered by the EMG recording), the measurement was discarded from the BCG analysis. The set-up included electrodes attached to the top of the feet of the subject and a custom EMG circuit. The gain was set to 60 dB and the bandwidth to 0.1 Hz - 200 Hz.

12 CHAPTER 3. ECG & EMG CIRCUIT

The first stage of the circuit, consisting of op-amp 1A, resistor R2 and capacitor C1, is also the first part of the input stage. Two electrodes are connected to the positive and negative inputs of op-amp A. The main function of this stage is to fine-tune the input stage roll-off, which means that it increases the attenuation of the signal above the low-pass cut-off frequency. The second stage is the second part of the input stage. This non-inverting amplifier with a gain of 16 is constructed from op-amp 1B and two resistors, R₃ and R₄.

An elaborate feedback system acts as the third stage, consisting of a T-network of resistors, a feedback capacitor C2 and a non-inverting integrator (based on op-amp 2A). The T-network (resistors R₅, R₅0, R₆, R₇ and R₇0) increases the transimpedance gain without introducing extra noise. Together with the surrounding capacitors, it also sets the high-pass cut-off frequency of the circuit (0.4 Hz) and the low-pass cut-off frequency of the integrator (111 Hz). The integrator removes the ac-component from the output of the second stage. The resulting dc-signal is converted into a current and fed back to one of the input electrodes. This negative feedback system drives the average low-frequency variations of the output voltage to zero [42].

The fourth stage of the circuit is a Sallen-Key second-order low-pass filter, set to a cut-off frequency of 41 Hz. The complete ECG circuit thus has a bandwidth of 0.4 Hz - 41 Hz and a gain of 57 dB. Simulations with MacSpice 3f5 confirmed the calculated values. The bandwidth is deliberately decreased to a typical ECG monitoring bandwidth (see subsection 1.3.2), because this bandwidth is sufficient to detect the R-wave of the ECG signal and correctly interpret the BCG signals.

The fifth and final stage is the isolation stage, which is required because of the contact between the electrodes and the feet of human subjects. The isolation barrier protects the subjects from dangerous currents entering the system at the output (where it is indirectly connected to the mains). The isolation amplifier used in this design was suggested in the work of Inan et al. [30,42] and provides an isolation barrier of 1.5 kV. It uses separate power supplies for its input and output stage, including bypass capacitors (C7 - C10) on both sides. Once again, due to the availability of 9 V batteries, the entire circuit is powered by two isolated ±9 V battery sets.

3.2.2 EMG Circuit

As acknowledged by Inan et al. [42] during the discussion of the ECG transimpedance amplifier, the same topology can readily be adapted for EMG acquisition. The EMG acquisition circuit can use the same electrodes and design as the ECG acquisition circuit, apart from minor changes to the T-network and the low-pass filter stage. The T-network is changed to increase the gain of the circuit, as required by the lower signal amplitude of the EMG signal compared to the ECG signal (see section 1.3). The changes in the low-pass filter stage increase the bandwidth for a better match with the bandwidth of the EMG signal (approximately 10 Hz - 5 kHz, see subsection 1.3.3). The total gain of the circuit is equal to 62 dB, while the bandwidth is set to 0.8 Hz - 100 Hz. MacSpice 3f5 simulations corroborated the calculation results. An isolation amplifier is once again required, using a separate power supply (±9 V battery set) for the output stage.

Chapter 4

COP Circuit

The fourth chapter of this thesis presents the preliminary design of the circuit responsible for the acquisition of a signal for postural analysis: the COP. The first section sketches the requirements of the circuit, while the second section focuses on the preliminary circuit design.

4.1

Circuit Requirements

The COP cannot be measured directly and is instead calculated from the forces and moments imparted by the person onto the scale. According to the discussion in subsection 1.3.4, forces recorded from the four corners of a platform are sufficient to calculate the COP. The COP circuit is thus a circuit that measures forces, very much like the BCG circuit. However, apart from measuring forces to calculate the COP, the circuit can also be used to measure the weight of the subject (the total force on the platform). The circuit requirements for complete functionality are listed below.

1. Record signals to calculate the COP of the subject. 2. Record signals to calculate the weight of the subject. 3. Allow the scale to be zeroed.

The third requirement is important for the correct interpretation of the weight and COP. Zeroing allows the scale to be calibrated (or reset) at any given time, thus cancelling all influences of previous measurements or objects on the scale.

4.2

Circuit Design

The preliminary design of the COP circuit was based on the BCG circuit and re-used two important elements. The Wheatstone bridge configuration for individual strain gauges and the instrumentation amplifier allowed simple measurements of the four forces required for COP calculations. However, significant design changes were also necessary, especially concerning the third requirement from section 4.1. In order to reset the scale, a memory unit for the analog output value of the Wheatstone bridge had to be implemented. This output value could then be used as a permanent reference value for the instrumentation amplifier.

After an extensive literature study, an analog-to-digital converter (ADC)/digital-to-analog con-verter (DAC) combination was selected to implement the memory unit. The ADC converts the analog output voltage of the Wheatstone bridge into a digital value and saves it in digital memory. Immediately after this conversion the digital value is converted back into an analog value that can be fed back to the rest of the circuit to zero the output. The preliminary design of the COP circuit is a combination of the BCG circuit from section 2.3 and an adapted reset circuit from Analog Devices [43].

4.2. CIRCUIT DESIGN 15

The Wheatstone bridge, the first stage of the circuit, has exactly the same design as in the BCG circuit. Two opposite nodes (A and C) are connected to the instrumentation amplifier, which forms the second stage of the circuit. It amplifies the difference between its two inputs with a gain set by the resistor Rg(equal to 100 in this circuit). The reference pin of the instrumentation amplifier is connected to ground.

The third stage is a unity gain differential amplifier, so the output of the stage is the difference between its two inputs. The non-inverting input is the output from the instrumentation amplifier, while the inverting input is the output from the ADC/DAC in the fourth stage. The gain of the amplifier is set by the four resistors R₁, R₂, R₃ and R₄. Because the resistors are all equal, the gain is equal to 1.

The output of the differential amplifier stage is connected to both the fourth stage and the fifth stage. The fourth stage is the stage responsible for the reset functionality. The stage input is connected to the analog input of the ADC/DAC. This component does not influence the other stages of the circuit, unless the reset switch is closed. Closing the switch starts a reset cycle, leading to the input being sampled and held infinitely at the analog output until the next reset cycle. Since the analog output is connected to the inverting input of the differential amplifier in the third stage, the output of the amplifier is driven to zero, resetting the circuit and cancelling all undesired signals. It is important to notice the power supply for this stage. Since the ADC/DAC is essentially a digital component, it requires a power supply voltage of 5 V. The standard power supply voltage of 9 V is therefore converted to 5 V with a regulator. The fifth and sixth stage are very similar to the two final stages of the BCG circuit. The fifth stage is connected to the output of the third stage and consists of a Sallen-Key low-pass filter with a cut-off frequency of 20 Hz. This cut-off frequency is based on the typical COP bandwidth (see subsection 1.3.4) and on the bandwidth of digital filters typically used for COP signals (see subsection 6.3.3). The final stage is a non-inverting amplifier with a gain of 110. The total gain of the COP circuit is therefore 81 dB. The low-pass frequency of the entire circuit is 20 Hz, but the high-pass frequency was not calculated.

Chapter 5

Power Supply Circuit

This chapter focuses on the preliminary design of a circuit capable of delivering power to each of the circuits presented in the previous chapters. Section 5.1 lists the circuit requirements, while section 5.2 presents the design of the different circuit elements.

5.1

Circuit Requirements

The requirements for the power supply circuit, listed below, are defined by the circuit designs for the acquisition circuits.

1. Select internal (battery) or external (mains) power input 2. Boost the power input to 9 V if required

3. Invert the power input to create a first set of ±9 V 4. Isolate the first set from a second set of ±9 V if required

Selecting the power input, listed as the first requirement, is a convenient way to reliably use the device in different environments. If no external power is available, the scale uses its two internal 3 V CR232 batteries connected in series (see section 2.2). However, if it is necessary to save battery power or if the batteries are already depleted, the scale can be connected to the mains electricity network via a suitable 9 V transformer. Switching between these two power inputs can easily be implemented by re-using the existing single-pole triple-throw switch for unit selection. The input voltage is then either 6 V (internal) or 9 V (external). Since each acquisition circuit (BCG, ECG, EMG and COP) requires ±9 V, the input voltage has to be boosted to 9 V and/or inverted to -9 V. The fourth requirement, applicable only to the ECG and EMG circuits, ensures that the constructed device is safe for humans. Since these circuits are in direct contact with humans, isolated power supplies are needed for the output stages (see section 3.2). Isolation prevents ground loops through which dangerous currents can enter the system.

5.2

Circuit Design

The three final requirements from section 5.1 were converted into preliminary designs of separate circuits that could be used as desired for the different acquisition circuits. Subsections 5.2.1 and 5.2.2 focus on the boost and inverter circuits, respectively, while the final subsection examines the design of the isolation circuit.

5.2.1 Boost Circuit

If the internal batteries (BAT ) are selected, the 6 V input voltage has to be boosted to 9 V. The external power input of 9 V (PWR) does not require boosting, but to simplify the circuit and avoid an extra switch, it is nevertheless connected to the boost circuit. A schematic of the boost circuit is shown in figure 5.1.

Chapter 6

Experimental Verification

The sixth chapter of this thesis presents the results of the experimental verification of the different acquisition circuits presented in the previous chapters. Three circuits are discussed in the first two sections, with section 6.1 focusing on the BCG circuit and section 6.2 concentrating on the ECG and EMG circuits. The final section examines the results of a calibration test, involving both custom circuits and standard equipment.

6.1

BCG Circuit

The BCG circuit had to be tested to confirm that it could record the forces exerted onto the body by accelerating blood. At the time of the first BCG experiment, only two circuits (corresponding to two strain gauges) had been finished due to a limited number of components. For a first test, however, two circuits were deemed sufficient to provide a proof of concept. Strain gauges on opposite corners of the scale were thus plugged into the BCG circuits and Bayonet Neill-Concelman (BNC) cables were used to connect the circuit outputs to a combination of a data acquisition card (National Instruments Corporation, USA) and a computer running LabVIEW 8.2 (Version 8.2.1, National Instruments, 2007). As mentioned in subsection 1.3.1, an ECG signal is required for the correct interpretation of a BCG signal. A standard ECG monitor (Life Scope 7, Nihon Kohden Corporation, Japan) and four disposable Ag/AgCl electrodes in a Lead II configuration (right arm to left leg, see subsection 1.3.2) were used during this test. The ECG monitor was also connected to the computer through the same digital acquisition card.

The simple test protocol, involving a single subject, consisted of standing on the scale for about 25 s. Following data acquisition, the recorded signals were processed using MATLAB R2012a (Version 7.14.0, The MathWorks Inc., 2012). The signals from the two strain gauges were band-pass filtered between 4 Hz and 15 Hz, a bandwidth established after a short literature study and some additional tweaking [30]. The two force signals were then summed and inverted because the reaction force on the scale was recorded instead of the force corresponding to the acceleration of blood. The signal from the ECG monitor did not require filtering. Instead, a simple peak detection algorithm, followed by manual inspection, was used to detect and mark the R waves (see subsection 1.3.2).

Figure 6.1 on the next page presents a plot of the BCG and ECG signals during two heartbeats (two R waves). The different features of the BCG signal, relative to the R wave of the ECG signal, identify it as a correct BCG signal (also see the discussion in subsection 1.3.1). Unfortunately, a correct BCG signal could not be identified for every single heartbeat, but the achieved results were sufficient to prove that the BCG circuit was able to record a BCG signal.

6.2

ECG & EMG Circuits

The experimental verification of the ECG and EMG circuits required recording clean ECG and EMG signals from a person standing on the scale while connected to the relevant electrodes and circuits. The test set-up involved the same digital acquisition card and computer with LabVIEW

Chapter 7

Conclusions & Outlook

The final chapter first presents the conclusions of the thesis by summarizing the most import-ant achievements and results in section 7.1. Section 7.2 then outlines different possibilities to continue the efforts of this thesis.

7.1

Conclusions

Microgravity has several different effects on the human body, including postural instability, orthostatic intolerance and muscle atrophy. It is therefore essential to continuously monitor the health of astronauts: preflight, postflight and during the flight. The cardiovascular system and the vestibular system stand out as two important physiological systems, not only for astronauts, but also for patients in healthcare.

This thesis presented the electronic design and experimental verification of a device capable of recording signals related to cardiovascular health and postural analysis within the convenient form factor of a standard weight scale. First, an off-the-shelf weight scale was selected and disassembled. Next, simple electrodes and custom acquisition circuits were designed to record the following physiological signals.

• BCG: the forces exerted onto the body by the blood accelerating inside the body • ECG: the electrical activity of the heart

• EMG: the electrical activity of the (lower-body) muscles

• COP: the point location on the ground where the resultant force due to foot pressure is applied

The BCG circuit used a combination of built-in strain gauges, amplifiers and filters to record the desired forces, while the ECG and EMG circuits relied on an innovative two-electrode set-up. A preliminary design of a COP circuit was also developed, based on a combination of analog-to-digital conversion and analog-to-digital-to-analog conversion. Finally, a preliminary version of a dedicated power supply circuit was designed.

The BCG, ECG and EMG circuits were assembled onto a breadboard. The three completed ac-quisition circuits were tested with the scale and the physiological signals of several subjects were successfully recorded. The signals also corresponded to previous measurements with standard equipment.

7.2

Outlook

The next step towards cardiovascular health monitoring and postural analysis is the assembly and test of the COP and power supply circuits. In the future, extra functionality can also

7.2. OUTLOOK 27

be added to the system to improve both the acquisition of the physiological signals and the user-friendliness of the entire system. A short list of additions is shown below.

• A hand-held ECG electrode to simplify ECG measurements (e.g. wrist strap, finger strap or handlebar)

• Custom ECG and EMG electrodes with less noise pick-up

• A circuit to record the seismocardiography/seismocardiogram (SCG) signal (the accelera-tion of the breastbone caused by the beating heart [32]) and acquire more cardiovascular information

• Digital circuitry to enable LCD integration and improved data processing

A system as described in this work would be capable of obtaining cardiovascular information and information related to postural control, which has many applications in healthcare and space physiology. It could, for example, be used for in-home monitoring of the elderly or recovering patients. Also medical professionals could use the device to quickly assess cardiovascular para-meters or perform postural analysis. In the aerospace sector, one potential application is using the device in pre- and postflight tests to continuously monitor the recovery of the cardiovascular and vestibular system of the astronauts. The system could even be used during zero-g flights and on different planets to investigate the influence of the environment on cardiovascular health and postural control.

Bibliography

[1] N. Kanas and D. Manzey, “Basic issues of human adaptation to space flight,” in Space Psychology and Psychiatry, vol. 22 of The Space Technology Library, ch. 2, pp. 15–48, Springer Netherlands, 2008.

[2] M. F. Reschke, J. J. Bloomberg, D. L. Harm, W. H. Paloski, C. Layne, and V. McDonald, “Posture, locomotion, spatial orientation, and motion sickness as a function of space flight,” Brain Research Reviews, vol. 28, no. 1–2, pp. 102–117, 1998.

[3] A. Blaber, R. L. Bondar, and M. S. Kassam, “Heart rate variability and short duration spaceflight: relationship to post-flight orthostatic intolerance,” BMC Physiology, vol. 4, no. 6, 2004.

[4] J. Fritsch, J. Charles, B. Bennett, M. Jones, and D. Eckberg, “Short-duration spaceflight im-pairs human carotid baroreceptor-cardiac reflex responses,” Journal of Applied Physiology, vol. 73, no. 2, pp. 664–671, 1992.

[5] T. Stein, “Weight, muscle and bone loss during space flight: another perspective,” European Journal of Applied Physiology, vol. 113, no. 9, pp. 2171–2181, 2013.

[6] M. Nichols, N. Townsend, R. Luengo-Fernandez, J. Leal, A. Gray, P. Scarborough, and M. Rayner, “European Cardiovascular Disease Statistics 2012,” tech. rep., European Heart Network and European Society of Cardiology, 2012.

[7] M. Alwan, D. Mack, S. Dalal, S. Kell, B. Turner, and R. Felder, “Impact of passive in-home health status monitoring technology in home health: Outcome pilot,” in 1st Transdisciplin-ary Conference on Distributed Diagnosis and Home Healthcare (D2H2), 2006, pp. 79–82, 2006.

[8] R. Sokwoo, Y. Boo-Ho, C. Kuowei, and H. Asada, “The ring sensor: a new ambulatory wearable sensor for twenty-four hour patient monitoring,” in Proceedings of the 20th An-nual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBS), 1998, vol. 20, pp. 1906–1909, 1998.

[9] J. S. Kim, Y. J. Chee, J. W. Park, J. W. Choi, and K. S. Park, “A new approach for non-intrusive monitoring of blood pressure on a toilet seat,” Physiological Measurement, vol. 27, no. 2, pp. 203–211, 2006.

[10] G. S. Chung, J. S. Lee, S. H. Hwang, Y. K. Lim, D.-U. Jeong, and K. S. Park, “Wakefulness estimation only using ballistocardiogram: Nonintrusive method for sleep monitoring,” in 32nd Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBS), 2010, pp. 2459–2462, 2010.

[11] I. Korhonen, J. Parkka, and M. Van Gils, “Health monitoring in the home of the future,” IEEE Engineering in Medicine and Biology Magazine, vol. 22, no. 3, pp. 66–73, 2003. [12] R. González-Landaeta, O. Casas, and R. Pallàs-Areny, “Heart rate detection from an

BIBLIOGRAPHY 29

[13] R. M. Wiard, O. T. Inan, B. Argyres, M. Etemadi, G. T. Kovacs, and L. Giovangrandi, “Automatic detection of motion artifacts in the ballistocardiogram measured on a modified bathroom scale,” Medical & Biological Engineering & Computing, vol. 49, no. 2, pp. 213– 220, 2011.

[14] O. T. Inan, D. Park, G. T. A. Kovacs, and L. Giovangrandi, “Multi-signal electromechanical cardiovascular monitoring on a modified home bathroom scale,” in 33rd Annual Interna-tional Conference of the IEEE Engineering in Medicine and Biology Society (EMBS), 2011, pp. 2472–2475, 2011.

[15] O. T. Inan, M. Etemadi, R. M. Wiard, L. Giovangrandi, and G. T. A. Kovacs, “Evaluating the foot electromyogram signal as a noise reference for a bathroom scale ballistocardiogram recorder,” in 21st IEEE International Symposium on Computer-Based Medical Systems, 2008, pp. 70–74, 2008.

[16] R. González-Landaeta, O. Casas, and R. Pallàs-Areny, “Bathroom scales as patient inter-faces for home health care,” MediTech, pp. 173–176, 2007.

[17] J. H. Shin, K. M. Lee, and K. S. Park, “Non-constrained monitoring of systolic blood pressure on a weighing scale,” Physiological Measurement, vol. 30, no. 7, pp. 679–693, 2009. [18] J. Gómez-Clapers, R. Casanella, and R. Pallàs-Areny, “Multi-signal bathroom scale to assess long-term trends in cardiovascular parameters,” in 34th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBS), 2012, pp. 550–553, 2012. [19] J. Williams, “Bridge circuits: Marrying gain and balance,” Application Note 43, Linear

Technology, 1990.

[20] A. Schmidt, M. Strohbach, K. van Laerhoven, A. Friday, and H.-W. Gellersen, “Context acquisition based on load sensing,” in Proceedings of the 4th International Conference on Ubiquitous Computing, 2002, pp. 333–350, Springer-Verlag, 2002.

[21] EMANT Pte Ltd, “Convert an electronic personal weighing scale to measure center of pressure (posturography).” http://www.emant.com/index.php?tid=100001, April 2013. [22] N. Goswami, P. G. Roma, P. D. Boever, G. Clément, A. R. Hargens, J. A. Loeppky,

J. M. Evans, T. P. Stein, A. P. Blaber, J. J. V. Loon, T. Mano, S. Iwase, G. Reitz, and H. G. Hinghofer-Szalkay, “Using the moon as a high-fidelity analogue environment to study biological and behavioral effects of long-duration space exploration,” Planetary and Space Science, vol. 74, no. 1, pp. 111–120, 2012.

[23] K. Saladin, “The Circulatory System: The Heart,” in Anatomy & Physiology: The Unity of Form and Function, ch. 19, pp. 715–746, McGraw-Hill, 3rd ed., 2004.

[24] I. Newton, Philosophiae Naturalis Principia Mathematica. London, UK, 1687.

[25] A. Eblen-Zajjur, “A simple ballistocardiographic system for a medical cardiovascular physiology course,” Advances in Physiology Education, vol. 27, no. 4, pp. 224–229, 2003. [26] I. Starr, A. Rawson, H. Schroeder, and N. Joseph, “Studies on the estimation of cardiac

ouptut in man, and of abnormalities in cardiac function, from the heart’s recoil and the blood’s impacts; the ballistocardiogram,” American Journal of Physiology, vol. 127, no. 1, pp. 1–28, 1939.

30 BIBLIOGRAPHY

[27] W. R. Scarborough and S. A. Talbot, “Proposals for ballistocardiographic nomenclature and conventions: Revised and extended: Report of committee on ballistocardiographic terminology,” Circulation, vol. 14, no. 3, pp. 435–450, 1956.

[28] P. Paul and L. B. Symes, “Ballistocardiography in the Physics Classroom,” The Physics Teacher, vol. 46, no. 5, pp. 300–302, 2008.

[29] J. W. Gordon, “Certain molar movements of the human body produced by the circulation of the blood,” Journal of Anatomy and Physiology, vol. 11, no. 3, pp. 533–536, 1877. [30] O. T. Inan, M. Etemadi, R. M. Wiard, L. Giovangrandi, and G. T. A. Kovacs, “Robust

ballistocardiogram acquisition for home monitoring,” Physiological Measurement, vol. 30, no. 2, pp. 169–185, 2009.

[31] J. Alametsä, J. Viik, J. Alakare, A. Värri, and A. Palomäki, “Ballistocardiography in sitting and horizontal positions,” Physiological Measurement, vol. 29, no. 9, pp. 1071–1087, 2008. [32] K. Tavakolian, Characterization and Analysis of Seismocardiogram for Estimation of

Hemo-dynamic Parameters. PhD thesis, Simon Fraser University, 2010.

[33] M. B. Rappaport, H. B. Sprague, and W. B. Thompson, “Ballistocardiography: I. physical considerations,” Circulation, vol. 7, no. 2, pp. 229–246, 1953.

[34] B. Brown, R. Smallwood, D. Barber, P. Lawford, and D. Hose, “Electrophysiology,” in Medical Physics and Biomedical Engineering, Medical Science Series, ch. 16, pp. 511–547, Institute of Physics Publishing, 1999.

[35] K. Saladin, “Muscular Tissue,” in Anatomy & Physiology: The Unity of Form and Function, ch. 11, pp. 407–441, McGraw-Hill, 3rd ed., 2004.

[36] Y.-H. Kwon, “Force-plate issues.” http://www.kwon3d.com/theory/grf.html, 1998. [37] S. S. Hasan, D. W. Robin, D. C. Szurkus, D. H. Ashmead, S. W. Peterson, and R. G.

Shiavi, “Simultaneous measurement of body center of pressure and center of gravity during upright stance. Part I: Methods,” Gait & Posture, vol. 4, no. 1, pp. 1–10, 1996.

[38] R. González-Landaeta, O. Casas, and R. Pallàs-Areny, “Heart rate detection from an elec-tronic weighing scale,” in 29th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBS), 2007, pp. 6282–6285, 2007.

[39] O. T. Inan, M. Etemadi, A. Paloma, L. Giovangrandi, and G. T. A. Kovacs, “Non-invasive cardiac output trending during exercise recovery on a bathroom-scale-based ballistocardio-graph,” Physiological Measurement, vol. 30, no. 3, pp. 261–274, 2009.

[40] Tanita Corporation, “InnerScan Body Composition Monitor Model BC-551.” Manual, 2006. [41] Tanita Corporation, “Model comparison chart.” http://www.tanita.com/en/personal/

model-comparison-chart/, July 2013.

[42] O. T. Inan and G. T. A. Kovacs, “An 11 µW, two-electrode transimpedance biosignal amplifier with active current feedback stabilization,” IEEE Transactions on Biomedical Circuits and Systems, vol. 4, no. 2, pp. 93–100, 2010.