BITalino: A Novel Hardware Framework for Physiological Computing

Hugo Pl

acido da Silva

, Jos

e Guerreiro

, Andr

e Lourenc¸o

, Ana Fred

and Ra

ul Martins

Instituto de Telecomunicac¸

oes, Instituto Superior T

ecnico

Avenida Rovisco Pais, 1, 1049-001 Lisboa, Portugal

Instituto Superior de Engenharia de Lisboa

Rua Conselheiro Em

ıdio Navarro, 1, 1959-007 Lisboa, Portugal

Keywords:

Biosignals, Instrumentation, Electromyography, Electrocardiography, Electrodermal Activity, Accelerometry.

Abstract:

Physical computing has spun a true global revolution in the way in which the digital interfaces with the real

world. From bicycle jackets with turn signal lights to twitter-controlled christmas trees, the Do-it-Yourself

(DiY) hardware movement has been driving endless innovations and stimulating an age of creative engineer-

ing. This ongoing (r)evolution has been led by popular electronics platforms such as the Arduino, the Lilypad,

or the Raspberry Pi, however, these are not designed taking into account the speciﬁc requirements of biosig-

nal acquisition. To date, the physiological computing community has been severely lacking a parallel to that

found in the DiY electronics realm, especially in what concerns suitable hardware frameworks. In this paper,

we build on previous work developed within our group, focusing on an all-in-one, low-cost, and modular

biosignal acquisition hardware platform, that makes it quicker and easier to build biomedical devices. We

describe the main design considerations, experimental evaluation and circuit characterization results, together

with the results from a usability study performed with volunteers from multiple target user groups, namely

health sciences and electrical, biomedical, and computer engineering.

1 INTRODUCTION

Nowadays, low-cost hardware is driving innovation in

ways never before seen; in his book ”The Medici Ef-

fect”, Frans Johansson states that ”When you step into

an intersection of ﬁelds, disciplines, or cultures, you

can combine existing concepts into a large number of

extraordinary ideas” (Johansson, 2006), which is par-

tially the secret behind this ongoing revolution. Phys-

ical computing has grown as a ﬁeld in its own right

(O’Sullivan and Igoe, 2004), and so far it has mostly

been characterized by the use of hardware designed to

deal with requirements that are not completely com-

patible with the needs of biosignal acquisition, such as

relatively high tolerance to noise, low sampling rates,

no need for galvanic isolation, among others.

While physical computing has the Arduino and its

successors and predecessors as stepping stones, the

physiological computing community has been mostly

lacking a comparable tool. Biosignals have very spe-

ciﬁc requirements, and many projects end up heav-

ily bounded by the high cost and limited access to

suitable hardware materials. Building on the guid-

ing principles of existing Do-it-Yourself (DiY) hard-

ware platforms, in this paper we focus on BITalino;

Figure 1: BITalino biosignal acquisition board.

a low-cost and highly versatile hardware framework

designed to allow anyone, from students to profes-

sional app developers, to create projects and applica-

tions with physiological sensors (Figure 1).

The rest of the paper is organized as follows: Sec-

tion 2 describes the background and related work;

Section 3 details the overall architecture of BITalino;

Section 4 presents the analog biosignal acquisition

blocks and their benchmarking; Section 5 highlights

the results of a usability study performed with multi-

ple potential target user groups; and ﬁnally Section 6

summarizes the main conclusions and future work.

246

Plácido da Silva H., Guerreiro J., Lourenço A., Fred A. and Martins R..

BITalino: A Novel Hardware Framework for Physiological Computing.

DOI: 10.5220/0004727802460253

In Proceedings of the International Conference on Physiological Computing Systems (PhyCS-2014), pages 246-253

ISBN: 978-989-758-006-2

 2014 SCITEPRESS (Science and Technology Publications, Lda.)

2 BACKGROUND

Several authors have delved into the problem of mak-

ing biosignal acquisition hardware tools accessible to

anyone. One of the most comprehensive work known

to date is the Open EEG Project

. This is a commu-

nity driven effort to make affordable hardware and

software tools available for anyone with interest in

working with Electroencephalography (EEG), how-

ever it has been mostly focused on EEG data and is

supported by technologies currently outdated.

With the advent of open source hardware plat-

forms such as the Arduino (Banzi, 2009; Buechley

and Eisenberg, 2008) or Raspberry Pi, efforts have

been made to extend their capabilities to biosignal

data acquisition. OLIMEX

created a low-cost sen-

sor shield for Electrocardiography (ECG) and Elec-

tromyography (EMG), Advancer Technologies has

devised the MuscleSensor

for EMG; the Pulse Sen-

sor

for Blood Volume Pulse (BVP) data acquisition

is yet another example. All of these are focused on

very speciﬁc modalities and limited to the actual ana-

log front-end for the sensors.

The main advantage of the Arduino-based ap-

proaches is the extremely low cost, given that from

$25 anyone can buy an Arduino board. Nonetheless,

they are designed to deal with very simple require-

ments for example in terms of sampling rate accuracy

and tolerance to noise. Previous work from our group

has actually attempted to use an Arduino Pro Mini

board for ECG data acquisition, however, experimen-

tal results clearly demonstrated several of these short-

comings (Alves et al., 2013).

More recently, Libelium

has proposed a multi-

modal sensor platform for e-Health, which includes

sensors for BVP, Blood Oxygenation (SpO2), res-

piration, temperature, ECG, glucose, Accelerome-

try (ACC), Electrodermal Activity (EDA), and blood

pressure. The advantage of this platform is the fact

that it integrates multiple sensors with embedded pro-

cessing algorithms. Still, most of the sensors do not

provide access to raw data, all the sensors are bundled

as a single unit (providing low ﬂexibility for custom

hardware conﬁgurations), and at ≈$500 it is not af-

fordable for everyone.

Despite the work developed so far, the physiolog-

ical computing community is still lacking adequate

hardware frameworks, leading our group to develop

preliminary work on hardware for multimodal biosig-

http://openeeg.sourceforge.net/doc/

http://www.olimex.com/SHIELD-EKG-EMG.html

http://www.advancertechnologies.com/

http://pulsesensor.com/

http://www.libelium.com/130220224710/

nal acquisition (Guerreiro et al., 2013). As high-

lighted in our previous work, further research on the

analog front-end of the voltage differential sensors,

namely the ECG and EMG, was needed, and several

usability issues related with the form factor ultimately

limited the scope of application of our initial proposal.

In this paper we address the main problems found

in the state-of-the-art, and present a completely re-

vised approach to BITalino, which has resulted in

the ﬁrst low-cost, all-in-one, ﬂexible and easy-to-use

hardware framework for physiological computing.

3 ANATOMY OF A BITALINO

3.1 Design Principles

We devised BITalino as an all-in-one ”Credit Card”

form factor that integrates multiple measurement sen-

sors for bioelectrical and biomechanical data acqui-

sition. The digital back-end is supported by a con-

trol block based on the ATmega328P microcontroller,

a power management block, and a communication

block that uses a Class II Bluetooth v2.0 module for

wireless data transfer to a base station (e.g. com-

puter, mobile phone, etc.). Two auxiliary connectivity

blocks introduced in the board enable RJ22 plugs to

be added to the device. The main speciﬁcations are

summarized in Table 1.

Table 1: BITalino speciﬁcations.

Sampling Rate 1, 10, 100 or 1000 Hz

Analog Ports 4 input (10-bit) + 2 input (6-bit)

Digital Ports 4 input (1-bit) + 4 output (1-bit)

Data Link Bluetooth (range up to 10m)

Actuators LED

Sensors EMG; ECG; EDA; ACC; LUX

Battery 3.7 V Lithium Ion

Weight 30 g / 1.06 oz

Size 105x60 mm / 4.13x2.36 in

By default, the system comes as a single board,

with its onboard sensors pre-connected to analog and

digital ports on the control block. Nonetheless, the

control, power, and communication blocks, as well

as the ﬁrmware are completely general purpose, en-

abling people to use only the digital back-end of the

BITalino with their own custom sensor and actuator

designs. Furthermore, each individual block can be

physically detached from the main board, allowing

people to use it in many different ways; in essence,

this architecture enables three conﬁgurations:

• Board: BITalino is used with no modiﬁcations,

enabling people to simply experiment with the

BITalino:ANovelHardwareFrameworkforPhysiologicalComputing

247

(a) Plugged

(b) Freestyle

Figure 2: Possible BITalino conﬁgurations.

onboard sensors for prototyping activities or real

time observation of multiple physiological phe-

nomena (Figure 1);

• Plugged: plugs are added to the BITalino and the

individual sensor blocks are separated from the

BITalino main board, leaving only the control,

power, communication and auxiliary connectiv-

ity blocks, and enabling people to interchangeably

use different sensor combinations (Figure 2(a));

• Freestyle: all the individual blocks are detached

from the BITalino main board, enabling people

to combine them in any way that best suits their

project ideas and applications (Figure 2(b)).

All the input channels selected by the user are

sampled by the MCU on the control block, and for-

matted into data packets of variable size (the extreme

cases are 8 bytes when all analog channels are se-

lected, and 3 bytes when no analog channel is se-

lected); a Cyclic Redundancy Check (CRC) code is

computed to identify and recover from data packet

inconsistency (e.g. due to loss of bytes in the com-

Figure 3: Data packets structure. The bit labels mark the po-

sitions corresponding to: the sequence number S; the CRC

code; each digital input channel D0 to D3, and each analog

channel A0 to A5.

munication), and a packet sequence number is also

added to help on the detection of loss of data packets.

Figure 3 shows the structure of the data packets sent

by the device. The conﬁgurable settings on BITalino

can be changed by sending 1 byte commands from

the base station to the device; Figure 4 summarizes

the command set that is interpreted by the device.

3.2 Performance Evaluation

We tested the BITalino to evaluate the performance

during real-time acquisition, namely the sampling

rate accuracy (or temporal uncertainty) and the dy-

namic speciﬁcations of the Analog-to-Digital Con-

verter (ADC) in what concerns the Effective Number

of Bits (ENOB), and the quality of the analog front-

end in terms of Signal-to-Noise Ratio (SNR), SNR

plus Distortion (SINAD) and Total Harmonic Distor-

tion (THD). In all of the tests we used an Agilent

33220A Function Waveform Generator.

To characterize the temporal uncertainty of each

version of the BITalino, a synthesized ramp wave with

a frequency of 1 kHz, 3 V

and offset of V

/2 was

acquired and its slope (

∆V

∆t

) was analysed and com-

pared with what was really injected. Table 2 summa-

rizes the temporal uncertainty results; only the sam-

pling rate Fs =1 kHz was tested as it is the most de-

manding.

Table 2: Sampling rate accuracy for each version of the

BITalino (Board, Plugged and Freestyle).

BITalino Fs (real) [Hz] Skew [%] Jitter [%]

Board 999.99 ± 0.025 0.00086 0.0025

Plugged 999.99 ± 0.025 0.00043 0.0025

Freestyle 999.99 ± 0.018 0.00129 0.0018

The ATmega328P features a 10-bit successive ap-

proximation ADC, which runs with input clock fre-

quency of 200 kHz ensuring maximum resolution,

and is connected to a 6-channel Analog Multiplexer

PhyCS2014-InternationalConferenceonPhysiologicalComputingSystems

248

Figure 4: Firmware commands.

with single-ended voltage inputs (0 − 3.3 V). As pre-

viously described, to evaluate the performance of the

ADC we used the SNR, THD, SINAD, and ENOB.

The SNR was computed as the ratio of AC sig-

nal power to noise power below half of the sampling

frequency (Equation 1); it is important to note that

the noise power excludes harmonic signals and DC.

The THD was determined as the sum of the powers

of the harmonic components (spurs) divided by the

input signal power (Equation 2). We computed the

SINAD as the ratio of the amplitude of the fundamen-

tal input signal to the sum of all other spectral compo-

nents below half of the sampling frequency excluding

DC (Equation 3); it is important to note that the the-

oretical minimum for the SINAD is equal to the ideal

SNR (Equation 4). Finally, the ENOB was computed

as shown in Equation 5; a higher ENOB means that

voltage levels recorded in an analog to digital conver-

sion are more accurate.

The crosstalk between each channel was also

tested, and it is less than the ideal SNR (−61.96[dB])

for 10-bits resolution.

SNR(dB) = 10log

(1)

where P

is the signal power and P

is the noise power

(exclude harmonic signals and DC).

T HD(dBc) = 10log

(2)

where P

is the power of all the speciﬁed spectral

components (harmonics).

SINAD(dBc) = 10 log

+ P

(3)

SNR(dB)

ideal

= 6.02n + 1.76 = 61.96[dB] (4)

where n is the number of bits.

ENOB =

SINAD − 1.76

6.02

(5)

Table 3 shows the results of the ADC dynamic

speciﬁcations. Tests were performed by applying a

synthesized sine wave with a frequency of 15 Hz and

voltage range of 95% ADC full-scale range.

Table 3: Dynamic speciﬁcations of the ADC (15 Hz sine

wave; Fs = 1 kHz).

SNR [dB] SINAD [dBc] THD [dBc] ENOB [bit]

55.72 54.29 -59.80 8.73

We also evaluated the power consumption of

BITalino; tests with a LiPo battery of 850 mAh

showed the autonomy to be ≈ 17 hours in real-time

acquisition over Bluetooth with all sensors connected.

This enabled us to conclude that the system typically

has a ≈ 50 mAh load current.

4 ANALOG FRONT-END

4.1 Overview

The analog front-end integrates individual sensor

blocks for Electromyography (EMG), Electrocardiog-

raphy (ECG), Electrodermal Activity (EDA) and Ac-

celerometry (ACC); the board is also ﬁtted with light

sensor (LUX) and LED analog blocks, to enable syn-

chronization with third-party equipment (e.g. a com-

puter screen or video camera).

The ECG sensor can be used for 1-lead measure-

ment of the bioelectrical activity of the heart and de-

rived parameters (e.g. heart rate, heart rate variabil-

ity, etc.), works both with gelled and non-gelled elec-

trodes, and can be applied in any standard location

(e.g. chest, left/right hand palms, left/right ﬁngers)

(Silva et al., 2011).

BITalino:ANovelHardwareFrameworkforPhysiologicalComputing

249

Table 4: Onboard sensor speciﬁcations.

ECG EMG EDA ACC LUX

Principle V. Diff. V. Diff. Resist. MEMS Photo Trans.

Electrodes 2 or 3 3 2 - -

Bandwidth 0.5-40 Hz 10-400 Hz 0-3 Hz 0-50 Hz -

Input Imp. 100 GΩ @ 3 pF 100 GΩ @ 3 pF - - -

CMRR 110 dB 110 dB - - -

Range 0-3 mV 0-3.3 mV 0-1 MΩ ±3 G 360-970 nm

Gain 1100 1000 2 - -

The EMG sensor can be used for measuring the

bioelectrical activity of the muscles and derived pa-

rameters (e.g. onset, duration, etc.), and can be ap-

plied to any surface muscle; the standard recommen-

dations are for the sensor to be applied with the differ-

ential electrodes centered mid-way along the muscle

belly, with a 2 cm inter-electrode spacing, and aligned

with the muscle ﬁbers beam (Basmajian and De Luca,

1985).

The EDA sensor can be used for measuring skin

resistance; a typical use of this sensor is to assess the

sympathetic nervous system activity, in which case

the two sensor leads are applied at the hand palms

or feet, allowing the measurement of the variations in

the skin impedance originated by sweat duct secretion

activity (Boucsein, 2011).

Finally, the accelerometer can be used for mea-

suring biomechanical events (e.g. tilt, step counting,

fall detection, physical activity), and the light sensor

can be used for measuring ambient light or for op-

tical synchronization with external sources (a typical

example being the synchronization of BITalino with

content being presented in a computer screen).

Table 4 presents the speciﬁcations of the individ-

ual sensor blocks provided by default on the analog

front-end, and through which the users are able to ex-

periment with multiple biosignals.

4.2 Sensor Characterization

We characterized the dynamic speciﬁcations of the

ECG and EMG voltage differential circuits using a

synthesized sine wave with frequencies of 24 Hz and

55 Hz, respectively, 25 mV

and offset of V

/2.

The gain of each sensor was reduced to 100 (IN-

AMPs with unity Gain); this procedure ensured a de-

sirable output signal without saturation and ranged

between 0 − 3.3 V.

Table 5 summarizes the results of the dynamic

speciﬁcations of the the circuits of the three versions

of the BITalino (Board, Plugged and Freestyle). Fig-

ure 5 shows the frequency response of each circuit. In

the top plots, we show the time response of each cir-

cuit to a synthesized chirp wave with 1 second dura-

tion and frequencies ranging between 0 − 100 Hz (for

the ECG) and 0 − 500 Hz (for the EMG), 25 mV

and offset of V

/2. As expected, the output signal is

a chirp wave with attenuation at low and high frequen-

cies. The bottom plots show the frequency response.

Table 5: Dynamic speciﬁcations (ECG and EMG, Fs =

1 kHz). For the BITalino Plugged results,

∗

denotes a 10 cm

cable length while

∗∗

denotes a 100 cm cable length.

BITalino Sensor SNR[dB] SINAD[dBc] THD[dBc]

Board

ECG 37.38 37.11 -49.39

EMG 35.96 35.86 -52.03

Plugged

ECG

∗

39.39 36.58 -39.80

∗∗

34.09 32.70 -38.34

EMG

∗

34.96 34.67 -46.61

∗∗

34.49 34.21 -46.17

Freestyle

ECG 39.72 36.71 -39.72

EMG 32.03 31.95 -49.23

We also performed a transient analysis of the volt-

age differential circuits, which showed that the ECG

circuit has 2.337 seconds time delay, while the EMG

circuit has 0.146 seconds time delay. One should note

that the time delay is bounded by the time constant

(τ = RC) of the integrator network at the ﬁrst stage of

the biosignal ampliﬁer in both circuits.

To further reinforce the performance of each cir-

cuit, Figure 6 shows real-world data. For the ECG,

we used an off-the-person placement as described in

(Silva et al., 2013), while for EMG measurement we

placed pre-gelled electrodes ﬁxed over a muscle.

Only the voltage differential sensors were revised

in the scope of our work, and as such, we refer the

reader to the work by (Guerreiro et al., 2013) for the

characterization of the remaining sensor blocks.

5 USABILITY ASSESSMENT

5.1 Methodology

A major concern in our work was the ease-of-use and

adequacy of our hardware framework within poten-

tial target user groups, reason for which we have con-

ducted a usability study. We adopted the approach

PhyCS2014-InternationalConferenceonPhysiologicalComputingSystems

250

(a) Frequency response of the ECG sensor. Bandwidth be-

tween 0.5 − 40 Hz.

(b) Frequency response of the EMG sensor. Bandwidth be-

tween 10 − 400 Hz.

Figure 5: Time and frequency response of the voltage dif-

ferential sensors.

proposed by Brooke (Brooke, 1996) around the Sys-

tem Usability Scale (SUS). The SUS is composed

of a 10-item questionnaire, in which each question

is a statement that the user needs to rate on a ﬁve-

point scale that ranges from ”Strongly Disagree” to

”Strongly Agree”, and measures the subjective per-

ception of interaction with a system. The SUS items

have been developed in accordance with the effec-

tiveness, efﬁciency, and satisfaction criteria as de-

ﬁned by the ISO 9241-11, and it has been widely

adopted within the engineering and usability practi-

tioners community.

As proposed by Brooke, the SUS assessment was

performed after the volunteers had contact the system

and before any debrieﬁng or discussion. The partic-

ipants were requested to respond to all items, and to

mark the center of the scale for the items in which

they felt they were not able to respond. Furthermore,

(a) Example of an ECG signal acquired by the BITalino.

(b) Example of an EMG signal acquired by the BITalino.

Figure 6: Samples of real-world data collected with the volt-

age differential sensors.

instead of thinking about the items for a long time,

the participants were asked to record their immediate

response to each item.

We enrolled volunteers from multiple target user

groups, namely: a) Biomedical engineering (BME):

users that are familiar with electronics, computer pro-

gramming and biosignals; b) Electrical engineering

(ECE): users that are mostly familiar with electronics

and computer programming; Computer science (CS):

users that are mostly familiar with computer program-

ming; and c) Health sciences (HS): users that are fa-

miliar with biosignals but non-proﬁcient in computer

programming nor electronics. Individual interviews

were performed with each user, and none of the vol-

unteers had previous experience with BITalino.

BITalino:ANovelHardwareFrameworkforPhysiologicalComputing

251

5.2 Experimental Results

A total of 100 volunteers were involved in the study

(BME: 25; ECE: 25; CS: 25; and HS: 25), with ages

rangin between 19 and 30 years old. Table 6 shows

the results for each SUS item in terms of mean and

standard deviation (µ ± σ), highlighting the best re-

sults in bold. The overall results suggest that our

framework is very easy to use, and that it has a fast

learning curve. Nevertheless, a few participants stated

that they would appreciate a more comprehensive

training before using the system.

Taking into account the SUS score interpretation

guidelines, BITalino not only exhibited above average

results, but it’s also an A grade system given that the

SUS scores across all groups stand above the 80.30

threshold (Brooke, 1996). As expected, BME users

presented the highest SUS scores, since they are more

familiar with the underlying principles and topics. In-

terestingly, CS users actually presented higher SUS

scores for the BITalino than ECE users; this may be

due to the fact that ECE users have a stronger back-

ground on hardware, leading their evaluation to be

more meticulous. Another interesting ﬁnding were

the high ratings of all users regarding the frequency

and ease of use of the system, which we interpret as

an indicator that a hardware framework such as the

BITalino is indeed necessary.

6 CONCLUSIONS

Affordable instrumentation has been a fundamental

problem in the ﬁeld of physiological computing for

many years, and while the state-of-the-art is rich in at-

tempts to provide sensors and systems, no single work

found to date has provided a comprehensive and reli-

able solution.

In this paper we presented the BITalino, a low-cost

all-in-one hardware framework that has a basic set of

multimodal sensors, enabling anyone to easily inte-

grate biosignal acquisition in their projects and appli-

cations. Our work builds on previous research from

our group to provide a novel purpose-built hardware

framework for physiological computing. Experimen-

tal results related both with the benchmarking of the

BITalino hardware blocks and usability assessment

demonstrate its suitability for the intended applica-

tion.

Future work will be mainly focused on the cre-

ation of a bootloader capable of enabling people to

upload custom ﬁrmware to the control block and also

on creating additional sensors for other modalities.

Table 6: Ranking for each SUS item in terms of mean

(µ) and standard deviation (σ). The rating is on a ﬁve-

point scale ranging from ”Strongly Disagree” to ”Strongly

Agree”.

Item

BITalino

BME ECE CS HS

I think that I would like to use this system frequently 4.48 ± 0.65 4.00 ± 0.75 4.12 ± 0.73 4.31 ± 0.68

I found the system unnecessarily complex 1.32 ± 0.48 1.73 ± 1.08 1.56 ± 0.96 1.42 ± 0.50

I thought that the system was easy to use 4.84 ± 0.37 4.73 ± 0.45 4.44 ± 0.71 4.62 ± 0.57

I think that I would need the support of a technical person to be able to use this system 1.60 ± 0.76 1.65 ± 0.85 1.60 ± 0.87 2.12 ± 1.03

I found the various functions in this system were well integrated 4.28 ± 0.68 4.19 ± 0.80 4.40 ± 0.58 4.08 ± 0.63

I thought there was too much inconsistency in this system 1.28 ± 0.46 1.54 ± 0.90 1.48 ± 0.71 1.46 ± 0.58

I would imagine that most people would learn to use this system very quickly 4.60 ± 0.50 4.69 ± 0.55 4.72 ± 0.54 4.15 ± 0.78

I found the system very cumbersome to use 1.40 ± 0.58 1.65 ± 0.75 1.48 ± 0.82 1.69 ± 1.01

I felt very conﬁdent while using the system 4.32 ± 0.63 4.19 ± 0.69 4.32 ± 0.69 3.85 ± 0.78

I needed to learn a lot of things before I could get going with this system 1.84 ± 0.94 1.65 ± 0.98 1.52 ± 0.65 1.92 ± 0.80

SUS Score: 87.70 ± 9.87 83.94 ± 11.67 85.90 ± 10.68 80.96 ± 12.19

PhyCS2014-InternationalConferenceonPhysiologicalComputingSystems

252

ACKNOWLEDGEMENTS

This work was partially funded by Fundac¸

ao para

a Ci

encia e Tecnologia (FCT) under the grants

PTDC/EEI-SII/2312/2012, SFRH/BD/65248/2009

and SFRH/PROTEC/49512/2009, whose support the

authors gratefully acknowledge. The team would

also like to thank for their support in the validation

and speciﬁcation of requirements to: Prof. S

ılvia

Ouakinin, Marco Torrado and Susana Eus

ebio from

the Medical Faculty of the University of Lisbon; Prof.

Marta Aires-de-Sousa and Prof. Nuno Raposo from

the Cardiopneumology Department of the Health

School of the Portuguese Red Cross; and Dr. Andreas

Keil from the Center for the Study of Emotion and

Attention of the University of Florida.

REFERENCES

Alves, A., Silva, H., Lourenc¸o, A., and Fred, A. (2013).

BITalino: A biosignal acquisition system based on

the Arduino. In Proc. of the 6th Int’l Conf. on

Bio-Inspired Systems and Signal Processing (BIOSIG-

NALS), pages 261–264.

Banzi, M. (2009). Getting Started with Arduino. Make

Books.

Basmajian, J. V. and De Luca, C. J. (1985). Muscles

Alive: Their Functions Revealed by Electromyogra-

phy. Williams & Wilkins.

Boucsein, W. (2011). Electrodermal Activity. Springer.

Brooke, J. (1996). Sus: A quick and dirty usability scale.

In Jordan, P. W., Thomas, B., Weerdmeester, B. A.,

and McClelland, I. L., editors, Usability Evaluation in

Industry, pages 189–194. Taylor & Francis., London.

Buechley, L. and Eisenberg, M. (2008). The LilyPad Ar-

duino: Toward wearable engineering for everyone.

IEEE Pervasive Computing, 7(2):12–15.

Guerreiro, J., Martins, R., Silva, H., Lourenc¸o, A., and Fred,

A. (2013). BITalino: A multimodal platform for phys-

iological computing. In Proc. of the 10th Int’l Conf.

on Informatics in Control, Automation and Robotics

(ICINCO), pages 500–506.

Johansson, F. (2006). The Medici Effect: What Elephants

and Epidemics Can Teach Us About Innovation. Har-

vard Business Review Press.

O’Sullivan, D. and Igoe, T. (2004). Physical Comput-

ing: Sensing and Controlling the Physical World with

Computers. Thomson.

Silva, H., Carreiras, C., Lourenc¸o, A., and Fred, A. (2013).

Off-the-person electrocardiography. In Proc. of the

Int’l Congress on Cardiovascular Technologies (CAR-

DIOTECHNIX).

Silva, H., Lourenc¸o, A., Lourenc¸o, R., Leite, P., Coutinho,

D., and Fred, A. (2011). Study and evaluation of a sin-

gle differential sensor design based on electro-textile

electrodes for ECG biometrics applications. In Proc.

of the IEEE Sensors Conf., pages 1764 – 1767.

BITalino:ANovelHardwareFrameworkforPhysiologicalComputing

253