=Paper=
{{Paper
|id=None
|storemode=property
|title=HealthNet: A System for Mobile and Wearable Health Information Management
|pdfUrl=https://ceur-ws.org/Vol-1075/08.pdf
|volume=Vol-1075
|dblpUrl=https://dblp.org/rec/conf/immoa/QuixBGHKLLQGJLMS13
}}
==HealthNet: A System for Mobile and Wearable Health Information Management==
https://ceur-ws.org/Vol-1075/08.pdf
HealthNet: A System for Mobile and Wearable Health
Information Management
Christoph Quix, Johannes Barnickel, Sandra Geisler, Marwan Hassani,
Saim Kim, Xiang Li, Andreas Lorenz, Till Quadflieg,
Thomas Gries, Matthias Jarke, Steffen Leonhardt, Ulrike Meyer, Thomas Seidl
UMIC Research Cluster at RWTH Aachen University, Germany
http://www.umic.rwth-aachen.de/
ABSTRACT Keywords
Medical health care is undergoing a significant change of Health Information Management, Data Acquisition, Data
paradigm. Moving health care from health centers to home Analysis, Body Sensor Network
environments poses new challenges for acquisition, manage-
ment and mobile exchange of information. The HealthNet
project at RWTH Aachen University has developed a proto-
1. INTRODUCTION
type which addresses these new challenges: a Body Sensor The demographic change with a growing population of
Network (BSN) collects information about the vital func- elderly people and the associated increase of health care re-
tions of a patient while she is in her home environment; lated expenses require new models for health care manage-
the integration of smart textile sensors increases the accept- ment. Moreover, there is a growing group of health-aware
ability of such technology; mobile communication and data people that would like to take more personal responsibility
management enables the exchange of health data between for their own health, e.g., by monitoring their vital parame-
patients and doctors; data stream mining techniques tuned ters during sport activities. New innovative technologies are
for mobile devices provide immediate feedback of the col- necessary to fulfill these new requirements. Mobile and re-
lected data to the user; and finally, advanced security and mote health monitoring has demonstrated positive influence
privacy features increase user acceptance and cope with le- on patients disease courses, especially for chronic diseases
gal requirements. This paper summarizes the challenges and [20, 13], and promises high cost reductions [16]. While vari-
achievements in the development of this prototype. ous systems have been proposed to measure the physiological
state of mobile users, most of these systems are restricted
to a certain set of sensors, or can monitor only a few vital
∗ parameters [6].
C. Quix, S. Geisler, X. Li, M. Jarke, and A. Lorenz are with
Informatik 5 (Information Systems) at RWTH Aachen Uni- In this paper, we describe an extendable and flexible mon-
versity, Germany (geisler,jarke,lixiang,lorenz@dbis.rwth- itoring system for the case study of physiological state moni-
aachen.de). toring of runners. The system has been developed in the con-
†
C. Quix and M. Jarke are also with the Fraunhofer Insti- text of the HealthNet project [14]1 , which addresses inter-
tute for Applied Information Technology FIT, St. Augustin,
Germany (christoph.quix,matthias.jarke@fit.fraunhofer.de). disciplinary challenges such as sensor network design, manu-
‡ facturing of smart textiles, information exchange, data min-
J. Barnickel and U. Meyer are with the IT Se-
curity group at RWTH Aachen University, Germany ing, security and privacy, and mobile communication. The
(barnickel,meyer@itsec.rwth-aachen.de). HealthNet project is part of the UMIC Research Cluster at
§ RWTH Aachen University which focuses at Ultra high-speed
M. Hassani and T. Seidl are with Informatik 9 (Data Man-
agement and Exploration) at RWTH Aachen University, Mobile Information and Communication systems support-
Germany (hassani,seidl@cs.rwth-aachen.de). ing the demands of future mobile applications and systems.
¶
S. Kim and S. Leonhardt are with the Philips Chair In the prototype developed by the HealthNet project team,
for Medical Information Technology (MedIT) at RWTH the vital functions of athletes (or patients) are monitored
Aachen University, Germany (s.kim,leonhardt@hia.rwth- by a BSN (e.g., ECG, skin humidity / temperature, activ-
aachen.de).
k ity) which are partly integrated into textiles. These sensors
T. Quadflieg and T. Gries are with the Institute for Textile
Technology (ITA) at RWTH Aachen University, Germany produce data streams that are integrated, consolidated, and
(till.quadflieg,thomas.gries@ita.rwth-aachen.de). aggregated in a device which acts as a peer in a network.
Other trusted participants in the network are, for example,
other runners or trainers who want to observe the perfor-
mance of a runner. In a medical scenario, other peers in
the network might be doctors or nursing staff who monitor
the state of a patient while (s)he is at home. Furthermore,
data can be stored in a server system for long-term mon-
itoring and analysis. An intensive monitoring of vital pa-
rameters of patients is especially important after they have
1
http://dbis.rwth-aachen.de/cms/projects/UMIC/
healthnet
Proceedings IMMoA’13 361 http://www.dbis.rwth-aachen.de/IMMoA2013/
been released from hospital. Changes in environment and the information needs of the runners and the willingness
medication often result in expensive re-hospitalizations of to share information during training and competitions. All
patients which could be avoided by more detailed observa- interviews took about one hour, and were recorded for post-
tion of vital parameters [10]. Thus, both scenarios (sports interview analysis.
and medicine) share a common basis; in addition, specific
features like the identification of critical situations are rele- 2.1 Participants
vant for both cases, although the definition of a critical situ- The interviewed runners are male, three between 20 and
ation is different. Nevertheless, the same techniques for data 25 years old, and one between 30 and 35 years old. All
analysis can be applied. Furthermore, merging the acquired runners participate in competitions on national level. The
sensor data with additional information such as position, disciplines range from 3000 meters steeplechase to marathon
time, or weather conditions improves the expressiveness of distance, and triathlon. All interviewed persons do intensive
pure health data and can lead to new insights. training between 5 and 15 sessions per week.
Using a mobile communication infrastructure (e.g., UMTS,
LTE, or Wi-Fi), mobile devices can communicate with each 2.2 Information Requirements
other such that peers can easily exchange health informa-
tion. Especially, the mobility of patients is improved as 2.2.1 Personal Information Sources
detailed monitoring can now be performed at home: peri-
All interviewed persons consider their self-assessment as
odically or in the case of important events, the device sends
the most important source of information, which is even
the collected and pre-processed data to information systems
more reliable than any physiological measure. They stated
maintaining patient health records (e.g., hospital informa-
to ignore measured high peaks of their pulse if feeling good,
tion systems) which can be accessed by medical experts.
and also low measures if feeling bad. That means, that their
The main challenges in this project are
subjective rating of their state is more important for them
• the design and development of wireless medical sen- than an objective measurement. They treat many technolo-
sors, which are able to monitor vital functions of a gies as fun, which could be more interesting to increase mo-
person, tivation in mass sports.
• the integration of theses sensors into textiles and devel- 2.2.2 Medical Information Sources
opment of electronic units as textiles (e.g., conductive In addition to the self-assessment, all interviewed runners
paths) for unobtrusive and comfortable usage, were interested in heart rate (they all measure it in training).
• the integration of the data collected by various sensors The data could be used to trigger a notification if exceeding
in one data stream, and upper or falling below lower personal limits. Furthermore,
the sportspersons consider breathing rate and oxygen ab-
• the analysis, mining, and aggregation of the sensor sorption relevant to detect exhaustion in advance. Sweat
data to detect emergency events, to reduce commu- analysis could lead to an estimation of water balance of the
nication costs, and to predict near future. body, useful for reminding the runner to drink or to deter-
We addressed these challenges in the HealthNet project mine the amount of liquid to drink for convalescence after
and report in this paper our experiences in developing an a training session. Determining the blood sugar level could
integrated prototype. Sectioin 2 first describes the require- signal a low sugar level or hunger knock2 . Last but not least,
ments analysis which we have done with a group of athletes. all interviewed runners had measured lactate in the past. It
An overview of the prototype system and its architecture is probably the best indicator for the current personal fit-
is given in Section 3. The main components of the proto- ness. Noticeably, none of the interviewed runners associated
type are an intelligent T-Shirt with integrated conductive any value with blood pressure information, even if explicitly
leads/electrodes (cf. Section 4) and a Body Sensor Network asked by the interviewers.
(IPANEMA, Integrated Posture and Activity NEtwork by Most measurements listed above require settings incom-
Medit Aachen) which aggregates multiple data streams from patible with daily outside use. Some request tests in medical
a range of sensors (cf. Section 5). The data are transmit- labs, some include in addition blood analysis (such as blood
ted wirelessly over a Bluetooth interface to a mobile device sugar level, lactate) which is not compatible with mobile
for visualization and a first lightweight analysis (cf. Section use. All interviewed runners agreed that therefore it will be
6). A more detailed analysis of the data is done on a server challenging or impossible to apply these measurements in
which receives periodically or in case of peculiar events data their training sessions or competition.
from the mobile device. We also performed a case study in a
running event of which we will briefly summarize the results
2.2.3 Track, Time and other Information
in Section 8. Speed measurements have a strong influence on the run-
ning speed. All runners pointed out that speed measures
2. REQUIREMENTS AND USE CASES from cars or bicycles are not usable because of the mean-
ingless unit (mph, km/h) and low precision. They request
For gathering of requirements, four active runners on semi- measures of time needed for the last lap (on cycle tracks),
professional level were interviewed. All interviews were con- the last 400 meters or the last 1000 meters (all preferable in
ducted by two interviewers with one interviewee. The in- a unit of minutes:seconds) to adjust their personal running
terviews used a unique set of 14 questions regarding mobile speed accordingly.
health monitoring, and eight questions regarding a station-
ary counterpart. The questions especially targeted the us- 2
a completely run out of energy, also known as “bonk” or
ability of smart phones as supporting device in runs and “hit the wall”
Proceedings IMMoA’13 372 http://www.dbis.rwth-aachen.de/IMMoA2013/
For uphill sections, the absolute distance and remaining 2.4.1 Live-sharing
distance of the uphill part are valuable for all interviewed Live-sharing information with others is considered a minor
runners. The gradient is less important because of the low issue by the interviewed runners. Together with personal
absolute number. trainers, post-processing (for long distance runs) and fre-
Other information, like weather conditions, weather fore- quent analysis after smaller sessions (e.g., in interval train-
cast or condition of the ground are important in preparation ing) was seen to be more important than live data transmis-
of training or competition; it is of no value while being on sion. One interviewee had the idea that the trainer might
the move. interrupt over-pacing of a runner in a hopeless intermedi-
ate state of a competition, especially if it is one in a row
2.2.4 Personalization
of competitions. All interviewed persons declined to lively
All interviewed persons request methods for personalisa- share personal or medical information with other external
tion of the measurements and accompanied items, such as persons like friends, training mates, online communities, or
frequency, upper/lower border. event organisers or competitors. It was only acceptable for
2.3 Mobile Monitoring notification in case of emergency.
To receive information from others, trainers and support-
2.3.1 Use of Technology ers call out time information and intermediate state of the
competition to the runner on track. The interviewed runners
All interviewed runners had applied technology for mon-
think that receiving more information, e.g., about personal
itoring heart rate; all interviewed runners knew technology
state of competitors, is rather distracting. One of the inter-
for gathering track data (i.e., GPS). None used other tech-
viewed persons stated a value of knowing intermediate state
nology, like step counters or sensors in shoes. Only one per-
of competition within the same age group, in particular if
son carries the mobile phone in training sessions, in a back
persons nearby are of the same or another group like the run-
pocket together with keys. They do their sports without
ner. Getting the positions of the team mates was considered
listening to music.
not interesting, neither in training nor in competition.
All interviewed runners track heart rate in training, only
As an open question, the interviewees brainstormed about
two do the same in competitions. Two do not track the
other ideas for valuable live-sharing of information. As a
heart rate in competition mainly because of loosing comfort,
result, it could be valuable for optimisation of the handover
i.e., chest belt slipping out of place and making the runner
in relays, especially in long distance relays. It would be of
feeling confined. The interviewed runners do not agree to
value to the successor to know the personal health state of
carry any additional device. In competition, none of them
the predecessor in order to adjust warming and preparation
would be willing to carry a mobile device.
phase. If the predecessor is in good shape, the estimated
2.3.2 Carrying a Mobile Device arrival time is earlier than if the person is in bad shape,
Carrying a mobile device while doing sports is considered influencing the point of time to start preparation.
burdensome. There must be a reasonable benefit from doing 2.4.2 Post-event sharing
so. It must not require any attention by the runner, it must
not swing (e.g., on a neck strap), it must not disturb the After a training session or competition, the runners were
rhythm of arms, legs or breathing (the latter nearly excludes open to share track and time data with team mates and on-
speech interfaces). The device must be lightweight, small, line communities, which is already implemented by portals
waterproof and shock resistant. The touch-sensitive surface, like http://www.gpsies.com.
if any, must come in a sweat resistant cover.
The shape and feeling of a watch was considered most
2.5 Persistent Storage
appropriate, as applied in current monitoring systems for Post-processing of the collected data is very important to
heart rate. It can be worn at one arm and operated with all interviewed runners. They asked to file all information
the hand of the other arm. If more functions are to be to a computer system for persistent storage. They all use a
integrated, the only sensible way of carrying a larger mobile kind of training diary, two use already computer applications
device seems to be a pocket at the arm. It supports a similar for this purpose.
way to operate it using the hand of the arm not carrying the
device.
2.5.1 Connecting with PC
The interviewed runners asked for easy connection with
2.3.3 Operating a Mobile Device the PC, and easy to handle download.
Operation of a 1-button-watch was considered sufficient;
nevertheless the operation of buttons of a mobile device were 2.5.2 Post-Processing
considered to require too much attention and too fine gran- All interviewed runners do intensive performance analy-
ular movements for hand and finger. A mobile device at the sis combining tracking data, time data, health information,
arm can be similarly operated by touch on the display. and comments on personal feeling. If applicable they com-
The interviewed persons see the problem with touching pare current data with past datasets for recurring events,
the display that it might get dirty and smeared by the run- competitions, tracks, or distances. The main goal of the
ner’s sweat, making checking current values from the display analysis is identification of flaws in performance (absolute
impossible. Because of disruption of rhythmic breathing, speed, endurance, power to go uphill) requesting updates of
speech-based operation is only considered feasible for a few the training method and plan.
short commands. The triathlete analyses shifts in performance of the single
disciplines, e.g., intensively training one discipline has con-
2.4 Sharing of Information tradictory influence on the performance in the other two.
Proceedings IMMoA’13 383 http://www.dbis.rwth-aachen.de/IMMoA2013/
One runner mentioned to use the post-processing also to In the current prototype, the BSN consists of an ECG
estimate lifespan of used hardware, e.g., professional running sensor, a combined temperature/humidity sensor, two 3D
shoes that loose suspension after 3000 km of use, demand- acceleration sensors, and a master node. The master node
ing for replacement to prevent damage from tendons and collects the data from the individual sensors and sends it
ligaments. to the smartphone. Conductive yarn acts as electrodes as
well as leads. The signals are received by the ECG sensor
2.6 Use Cases attached to the shirt. The sensor processes the ECG and
Based on the requirements analysis, several use cases were infers the current heart rate from it.
identified which are described in this section. The use cases On the smartphone, a mobile application integrates the
are grouped in four categories: Sensor management, mobile health data with data measured by the phone, such as the
monitoring, sharing, and archiving. current GPS position. The mobile application also visual-
izes, stores, and analyzes the data. If enabled by the user the
2.6.1 Managing Sensors integrated data is sent via UMTS or Wi-Fi (IEEE 802.11)
The sensor managing use cases describe the setup, con- to a registry server which distributes it to registered third
figuration and maintenance of the set of sensors delivering parties, such as a trainer, a doctor, or a server analyzing
information to the system. The actor usually is the user. the data. The architecture also allows sending feedback and
In addition, other persons or organizations might perform results of the analysis of the data to the users smartphone.
the use cases as well, e.g., an emergency doctor who adds a
sensor after the user had an accident, or a physician who ad-
justs the upper border of a physiological parameter to raise
4. TEXTILE PLATFORM
notification earlier. The actor employs a plug-in / plug-off The state-of-the-art electrodes used for most medical ap-
mechanism to add or remove sensors to the network; this plications are, for example, disposable electrodes glued onto
should be as automatic as possible. The added sensors per- the skin. These electrodes are coated with electrolyte-gel
form registration and de-registration at the controlling com- to improve the conductivity. The advantages of those elec-
ponent of the sensor network. Configuration of the sensors trodes are low contact impedance and a fixed position. How-
should be also possible, so that user can adjust the prop- ever, they are not suitable for a continuous long-term mea-
erties of the sensor (e.g., sampling rate, sensor identifier, surement because the electrolyte-gel can dry and may also
measure unit, data transmission rate) to his/her personal cause allergic reactions. Moreover, the wires between elec-
needs. trodes and the sensor exacerbate the handling for untrained
users. To achieve the aim of a continuous and mobile mon-
2.6.2 Monitoring itoring system, another solution has to be found.
The monitoring use cases describe the use of a mobile sys- Textile electrodes could be a good alternative for the stan-
tem to monitor the health status. The user employs the sys- dard ones. They can be used for long-term measurements
tem for observing specific parameters, being informed about because they are not coated with electrolyte-gel. The yarn
the current status and alarming himself or another entity for the textile electrodes must possess high conductivity,
during a personal activity. The user can also turn off all good elastic behavior to assure a good skin conformance and
monitoring and notification functions by muting the device. it should be biocompatible due to the constant skin contact.
Another advantage of textile electrodes is that these elec-
2.6.3 Sharing trodes can be integrated into garments which lead to a very
The group of sharing use cases describes the information high mobility of the whole system and intuitive handling.
exchange between all parts of the system with external en- Mobility can be further increased by using textile integrated
tities (e.g., server or other users). It applies to sharing in- conductive paths instead of cables. A reversible interface is
formation while being mobile as well as sharing information necessary to remove the sensor node before washing. How-
from the other parts of the system like the archive. The ever, textile electrodes also have disadvantages: the contact
group contains: impedance is higher and movement causes motion artifacts.
Suitable yarns matching all requirements mentioned above
2.6.4 Archiving have been researched and tested. The best one was a silver-
The archiving use cases describe the use of and retrieval coated polyamide yarn. A circular foam padded textile elec-
from a persistent storage. The user employs a stationary trode with a radius of 2.5 cm was used. In addition to
device (such as a laptop or desktop PC) to search for infor- the ECG electrodes, the same material was also used to
mation of a specific type, date and time, activity, or value. manufacture the textile conductors (see Fig. 3). The tex-
The archiving use-cases are: tile conductors were applied to the outside of the T-shirt
with metal push buttons to connect both electrodes and
the sensor. Preliminary results with this T-Shirt show the
3. SYSTEM ARCHITECTURE suitability of textile electrodes for the application as ECG
The HealthNet prototype is based on a BSN integrated electrodes.
into a textile platform (i.e., T-shirt) measuring the phys-
iological state of a person. An overview of the system is
illustrated in Fig. 1. A registry server manages the commu- 5. BODY SENSOR NETWORK
nication between different peers in the network. The sensor Body Sensor Networks (BSN) usually consist of a varying
data is received by a smartphone via Bluetooth which sends number and diverse types of sensors. They are wirelessly
the data to other peers in the network. Other peers in the connected either to each other, called mesh network, or to
network are an advanced data mining & analysis service or a central master node, called star network. The acquired
other trusted parties such as trainers and doctors. data is then transferred over wide area networks (WAN) to
Proceedings IMMoA’13 394 http://www.dbis.rwth-aachen.de/IMMoA2013/
Light‐weighted Single‐stream Multiple Streams
Mining & Analysis Mining & Analysis
Master
node
Sensor
Mobile Trusted Parties
Smartphone Registry Server
Bluetooth IEEE 802.11 / IEEE 802.11 /
UMTS UMTS
Figure 1: Overview of the system architecture Figure 2: BSN node Figure 3: Sensor shirt
central data and health service providers for further process- Graphical User
Configuration
ing. This section focuses on the challenges in developing the Interface
medical sensors, connecting them in a BSN, and processing
the measured signals. Single-Stream
Prediction
Bringing health status monitoring to personal health care
environments presents a new set of challenges: devices have Data Data
Measure
to be small, unobtrusive and easy to handle. Preferably, Cache Window
they need no or only minor interaction and are connected via HealthNet
Data Incoming/
Outgoing Data
Sensor Transmission
wireless technology to the supervising medical professional Controller
Unit
or health care center.
Preprocessing for
The IPANEMA BSN is designed to be easily modified for Multiple Streams
different application scenarios, e.g., cardio-vascular moni- Prediction
toring or hydration status monitoring [11]. It is small (68
x 42 mm, see Fig. 2), light (30 g) and wireless enabled. A
Figure 4: Architecture of the mobile application
sensor node consists of a base board which includes a low
power microprocessor (MCU, MSP430F1611, Texas Instru-
ments), power management circuitry, and a low power ra-
dio transceiver (CC1101, Texas Instruments). Modularity Data Window such that a single-stream prediction over a
is ensured by using a pair of connectors to attach different short timeframe is possible. The windows are implemented
sensor extensions. Two connectors (Samtech Inc.) enable as a circular data structure - if a window is full, the lat-
the use of digital (SPI, UART, I2C) sensors, five analog-to- est incoming data will flush the oldest data. Furthermore,
digital converter inputs and three interrupt capable inputs. the cache stores also all data (if desired by the user) such
The MCU is running at 8 MHz with an additional precision that the data can be uploaded to a server for detailed data
32.768 kHz crystal for the real time clock. It is powered by mining and analysis later on.
a lithium polymer battery which can be recharged over an The Data Transmission Unit (DTU) takes care of the
on-board MicroUSB connector. information exchange among different stakeholders. Four
The sensors of the current prototype produce a raw data methods of sending data to authorized entities have been
stream of about 14 kbit/s which is transmitted over a 433 implemented. Any external entity must prove eligibility to
MHz ISM band transceiver with a proprietary protocol. receive any data from the mobile application. The DTU
The network is structured in a star topology. The leaves supports three communication modes:
are formed by a flexible number of modules which can be
equipped with different types of sensors. The sensor data
is send over-the-air to a central master module. The main 1. Request-response: an external entity requests informa-
tasks of the master node include network management, data tion from the mobile application. The DTU retrieves
transfer to a mobile device and creating time synchroniza- the requested data from the cache and transmits the
tion beacons for the sensor nodes. response. This is for example done when a trainer
wants to see detailed data about a runner.
6. MOBILE APPLICATION 2. Time-based submission: A fixed interval after which
The goal in the design of the architecture of the mobile ap- a selected data set is sent, e.g., data is sent from the
plication was to have a very flexible and extensible system. runner to a trainer only every 10 seconds to reduce
As explained before, the HealthNet project is not limited required bandwidth and communication costs.
to a particular application domain, our solution should be
applicable in a healthcare domain as well as in a sports do- 3. Direct transmission: The relevant data is transmitted
main. To allow easy customization and adaptation to new directly to the receiver. This mode is used for audio
domains, we identified four main components for the mobile feedback from a trainer to a runner.
application on the smartphone (cf. Fig. 4).
The HealthNet Controller is the central unit for manag- Due to the modular design, peer mobile applications use
ing the set of active sensors, and notifying dependants if roughly the same architecture, with the only difference that
measures changed value or the composition of the network these applications receive data via the DTU and not from
changed. The Data Cache stores recent sensor data in a sensors.
Proceedings IMMoA’13 405 http://www.dbis.rwth-aachen.de/IMMoA2013/
On the user interface level, the data which is received from the devices, so that no data can be recovered wrongfully by
the sensors or other peers is managed according to the use someone who has physical access to a device. During com-
cases as described in section 2. munication between trusted devices, we do not rely on the
security mechanisms of the technologies used (e.g., UMTS,
6.1 Data Analysis LTE, WLAN) because the data must not be revealed to the
To get the maximum benefit of the HealthNet applica- network operators, and wireless technologies such as UMTS,
tion the measured data has to be analyzed to detect critical LTE, and WLAN typically only encrypt the air interface.
situation or events, and to make a short term predictions. Instead, all data transfers apply AES-128 encryption and
Data mining techniques in this context are restricted by two message authentication codes on the application layer.
important constraints: (i) the data is a continuous stream In wireless connections to trusted parties, all parties are
and has to be analyzed in real-time; persistent storage and identified using certificates with shared keys. The implemen-
long time series of data are not available as in classical data tation of the encryption is transparent to the application as
mining tasks, (ii) the resources (CPU power, battery life, standard interfaces of the Android SDK are used to imple-
memory) of the mobile device are very limited. ment secure storage and communication. In addition, we
To cope with the problem of limited resources, we devel- found that the authentication and encryption mechanisms
oped: (i) an adaptive technique for anytime classification, had no significant influence on battery life or performance
which is capable of both, classifying under varying time and of the handheld device.
resource constraints, and incrementally learning from data
streams to adapt to possible evolutions of the underlying
data stream [15], (ii) and a novel in-network distributed 7. RELATED WORK
sensor data clustering technique that efficiently aggregates The interest in mobile healthcare applications started with
similar sensor readings using coordinators [8]. systems like [19, 12] supporting professionals (like physician,
Context prediction is an emerging topic in the field of nurse, therapist, or midwives) to enter, receive and exchange
data mining, e.g., predicting the location of mobile objects information about their patients. Systems for professional
was a frequently tackled subtask of mobile context predic- users in hospitals like [2] considered specific design aspects
tion in recent researches. For scenarios of managing health to support local mobility in the hospital by interconnecting
information of mobile persons, the prediction of the near fu- PDA, laptop and desktop computers. Examples of systems
ture health status of persons is at least equally important for non-professional users are the self-monitoring applica-
to predicting their location. A first method for predicting tion for overweight people [22], alcohol consumption moni-
the next health context of mobile persons equipped with tor [4], or dietary advisor [9]. The results of these studies
body sensors and a mobile device has been developed and point to a high degree of monitoring by those using a mobile
implemented [7]. The proposed PrefixSpan-based method monitoring device compared to other monitors. In differ-
searches for sequential patterns within multiple streaming ence to the aim of the HealthNet project, these systems are
inputs from the body sensors as well as other contextual not equipped for continuously monitoring vital parameter in
streams that influence the health context. silent mode.
Our main observation is: frequent sequential patterns ap-
pearing in rules containing multiple streams, are completely 7.1 Textile Sensor Platforms
built using frequent patterns existing in each single stream. A reasonable idea to integrate real-time monitoring into
Thus, predicted values were directly presented to the user daily life activities are the application of wearable or textile
in the mobile application using a light-weighted resource- sensor platforms. This section therefore reviews the integra-
aware algorithm that was implemented locally on the user’s tion of sensors into garments, such as sport shirts or similar.
mobile phone. More accurate predicted values were sent to In [5] two types of textile sensor platforms are distinguished:
the user from a multiple stream prediction algorithm which while textile sensors are realized by special yarns, non-textile
was implemented on a server using the preprocessed frequent or textile-integrable sensors are singular units which are ap-
patterns on each stream (cf. Fig. 4). plied to the garment, e.g., printed onto the textile. The
advantage of textile sensors is that these can be produced
6.2 Security and Privacy in one manufacturing process [17]. A disadvantage is that
A rigorous evaluation of security and privacy risks was current technologies for textile sensors have to be moistened
done, requirements were derived from it, and the implemen- to deliver acceptable results [5].
tation was developed accordingly [3]. The measured data is To integrate multiple vital parameters into one textile
kept confidential at all times: during collection, in storage, platform, several sensors are combined to form a sensor net-
and during transmission within and between all components work. Often, a master component controls the network and
of the system. To reduce the risk of data extortion from centralizes data acquisition, short-term storage and trans-
stolen devices, secure authentication methods are used both mission. These can be realized either wired or wireless.
for wireless links as well as user interfaces on the devices
themselves. Generally, data may only be read by persons 7.2 Textile-based Monitoring Applications
authorized by the user. Finally, no more data than required The MyHeart-project3 led by Philips was dedicated to the
for a given monitoring application shall be stored. prevention, diagnosis and therapy of cardiovascular diseases
Confidentiality during data collection is achieved by using The monitoring is based on sensors integrated into daily
ZigBee AES-128 encryption between the sensor nodes and life textiles, such as undergarment. A sensor shirt has been
the master node, and Bluetooth encryption E0 is used be-
tween the master node and the smartphone. Confidentiality 3
http://www.hitech-projects.com/euprojects/
during data storage is achieved using AES-128 encryption on myheart/
Proceedings IMMoA’13 416 http://www.dbis.rwth-aachen.de/IMMoA2013/
developed using conductive and piezoresistive yarn for mon- running shoes with an integrated step counter (with the
itoring of heart (ECG) and respiratory activity (impedance drawback of getting depended on the Nike’s brand), and
pneumography), core and skin temperature with non-textile that it uses the iPod instead of a mobile phone (with the
sensors and an accelerometer [1]. The shirt has been used same drawback of dependency).
for monitoring during outdoor activities and at home. A
proprietary user device or PDA is used for interaction [21]. 8. CONCLUSIONS AND LESSONS LEARNED
The respiratory sensing technology was also used in the
Wealthy project4 [18, 17]. For the data processing and trans- We implemented an end-to-end prototype for a runner
mission a relatively heavy and big Portable Patient Unit scenario (training and competition mode) with one or more
(250g) was connected with the sensors by wires. The data runners and a trainer. Case studies with the implemented
is transmitted from the PPU via GPRS to a central system prototype have been conducted during the Lousberglauf 2011
analysing and visualizing the data. & 2012 (a local running event in Aachen with about 2000
A project that supports medical treatment and behaviour participants). A team of five runners has been equipped with
of elderly people suffering from cardio-vascular disease is the sensors and smartphones. In addition, a trainer mon-
described in [23]. The system comprises a front worn array itored the performance of the runners using also a smart-
of body sensors, a user interaction system for a PDA for phone. Data communication and management did not cause
displaying information and entering simple answers and a any problems; the trainer could always see the position and
back-end system for professionals analysing data and pro- vital parameters of the runners. Due to excessive motion ar-
viding feedback. tifacts during running, we used standard electrodes for the
run. In the meanwhile, we did some additional measure-
7.3 Products for Sports Monitoring ments with a new version of the textile electrodes in a lab
environment on a treadmill which gave better results. We
Commercial products are available on the market in par-
also improved the algorithm for inferring the heart rate from
ticular to support ambitious sports(wo)men. The products
the raw ECG data, such that it is less sensitive to movement
do not aim on sophisticated measuring medical data. Usu-
artifacts. This improved the data quality in the second case
ally, it is considered sufficient to provide heart rate and calo-
study in 2012, but the data quality is still too low for deriv-
ries burned, and location and time related information. The
ing health-related advices.
often use wrist or chest bands.
We have shown in this project that health monitoring us-
A large set of wrist-mounted computers is available for
ing mobile wearable sensor networks is feasible. Data man-
example from Polar, ranging from low-end technology for be-
agement and analysis can be done in real-time although the
ginners to high-end systems for professionals like the RS8005 .
data is coming at a high frequency. Security and privacy
They receive body signals from chest straps, display and
issues have been addressed by implementing suitable en-
store the information on the watch, and allow for down-
cryption and authentication mechanisms into the applica-
loading and post-processing with the personal computer. A
tion. In another related project (Nanoelectronics for Mobile
similar system is the Garmin Forerunner6 . It monitors time,
AAL-Systems9 ), a similar approach for data management
distance, pace, heart rate and calories burned. As Garmin’s
has been developed in the context of Ambient Assisted Liv-
unique selling proposition, it additionally tracks the posi-
ing (AAL). Some results (e.g., the architecture of the mobile
tion of the sportsperson by the use of a high-sensitive GPS
application in Fig. 4) have been applied also in this project.
receiver built into the wrist watch. The GPS antenna is
However, we have seen that with the current technology,
partially integrated into the watchstrap. The heart rate is
problems like data management, analysis, security, and pri-
measured by the use of a chest strap. The system supports
vacy can be solved as mobile devices are powerful enough
different profiles, e.g. for swimming, cycling, and running of
in terms of CPU and communication bandwidth. The real
triathlons.
challenges are at the two ends of the data processing flow:
A more sophisticated system is the adidas miCoach7 . It
firstly, the sensor data must have very high quality to be use-
is an integrated system to plan, work-out, and analyse per-
ful in any kind of application (for sportspeople or patients),
sonal training. As the main part of the system, it combines
false alarms will be annoying, missed alarms might be fatal;
three components to support the work-out: an auditive dis-
secondly, the potential users have to be convinced about the
play (miCoach Pacer) for heart rate measures, speed and
usefulness of such technology. In our interviews, the sports-
distance which reacts to the speed; a bundle of a chest belt
people were sceptic about the benefit of such an application.
measuring heart rate and a wristwatch (miCoach Zone); an
The same applies also to elderly people who might be even
application running on the user’s mobile phone for coaching
more reluctant in wearing any device that monitors them.
(miCoach Mobile).
As a main advantage, the textile strap for monitoring the
heart rate can be replaced by two different bra’s (adidas Acknowledgements
supernova glide/sequence bra) or a shirt (adidas supernova This work was supported by the DFG Research Cluster
cardio shirt). Nike+8 is a training system similar to mi- of Excellence on Ultra High-Speed Mobile Information and
Coach developed by Nike and Apple. The main differences Communication UMIC (http://www.umic.rwth-aachen.de)
to miCoach is that Nike+ combines the features from Nike’s at RWTH Aachen University.
4
http://www.wealthy-ist.com
5
http://www.polar.fi/en/products/maximize_ 9. REFERENCES
performance/running_multisport/RS800CX [1] O. Amft and J. Habetha. Smart medical textiles for
6
http://www8.garmin.com/marathon/forerunner/ monitoring patients with heart conditions. In Smart
7
http://www.micoach.samsungmobile.com/
8 9
http://nikerunning.nike.com/ http://www.mas-aal.eu/
Proceedings IMMoA’13 427 http://www.dbis.rwth-aachen.de/IMMoA2013/
textiles for medicine and healthcare: Materials, [15] P. Kranen, E. Mller, I. Assent, R. Krieger, and
systems and applications. Woodhead Publishing Ltd., T. Seidl. Incremental learning of medical data for
2007. multi-step patient health classification. In C. Plant
[2] J. Bardram, T. A. K. Kjær, and C. Nielsen. and C. Bhm, editors, Database Technology for Life
Supporting local mobility in healthcare by application Sciences and Medicine, pages 321–344. World
roaming among heterogeneous devices. In L. Chittaro, Scientific Publishing, 2010.
editor, Proc. ACM MobileHCI’03, volume 2795 of [16] R. Litan. Vital signs via broadband: Remote health
Lecture Notes in Computer Science, pages 161–176. monitoring transmits savings, enhances lives.
Springer, 2003. Technical report, Better Health Care Together, 2008.
[3] J. Barnickel, H. Karahan, and U. Meyer. Security and [17] M. Pacelli, G. Loriga, N. Taccini, and R. Paradiso.
privacy for mobile electronic health monitoring and Sensing fabrics for monitoring physiological and
recording systems. In IEEE International WoWMoM biomechanical variables: E-textile solutions. In Proc.
Workshop on Interdisciplinary Research on E-Health of 3rd IEEE/EMBS Intl. Summer School on Medical
Services and Systems, pages 1–6, 2010. Devices and Biosensors, pages 1–4. IEEE, 2007.
[4] R. L. Collins, T. B. Kashdan, and G. Gollnisch. The [18] R. Paradiso and D. De Rossi. Advances in textile
feasibility of using cellular phones to collect ecological technologies for unobtrusive monitoring of vital
momentary assessment data: Application to alcohol parameters and movements. In Proc. EMBC’06, pages
consumption,. Experimental and Clinical 392–395. IEEE, 2006.
Psychopharmacology, 11(1):73 – 78, 2003. [19] J. Pollard, S. Rohman, and M. Fry. A web-based
[5] A. Ehrmann, E. Flacke, A. Brücken, and A. Büsgen. mobile medical monitoring system. In Proc. Intl.
Intelligent umgarnt. Textile Sensorik. Physik in Workshop on Intelligent Data Acquisition and
unserer Zeit, 41(6):282–286, 2010. Advanced Computing Systems: Technology and
[6] Y. Hao and R. Foster. Wireless body sensor networks Applications, pages 32 –35, 2001.
for health-monitoring applications. Physiological [20] J. Riley and M. Cowie. A clinical perspective on
Measurement, 29:R27, 2008. remote monitoring of chronic disease. Eurohealth, 5(1),
[7] M. Hassani and T. Seidl. Towards a mobile health 2009.
context prediction: Sequential pattern mining in [21] R. Schmidt, J. Habetha, and M. Harris. Myheart -
multiple streams. In Proc. Intl. Workshop on fighting cardio-vascular diseases by prevention and
Managing Health Information in Mobile Applications early diagnosis. In L. Azevedo and A. R. Londral,
(HIMoA), pages 55–57, 2011. editors, HEALTHINF (1), pages 296–300, 2008.
[8] M. Hassani and T. Seidl. Resource-aware distributed [22] C. C. Tsai, G. Lee, F. Raab, G. J. Norman, T. Sohn,
clustering of drifting sensor data streams. In Proc. 4th W. G. Griswold, and K. Patrick. Usability and
Intl. Conf. on Networked Digital Technologies (NDT), feasibility of PmEB: a mobile phone application for
2012. monitoring real time caloric balance. Mob. Netw.
[9] S. Intille. Just-in-time technology to encourage Appl., 12:173–184, March 2007.
incremental, dietary behavior change. In Proc. [23] E. Villalba, M. Arredondo, A. Moreno, D. Salvi, and
AMIA’03, page 874. Wiley, 2003. S. Guillen. User interaction design and development of
[10] V. Jones, V. Gay, and P. Leijdekkers. Body sensor a heart failure management system based on wearable
networks for mobile health monitoring: Experience in and information technologies. In Proc. EMBC’06,
europe and australia. In Proc. Intl. Conf. Digital pages 400 –403, 2006.
Society (ICDS), pages 204 –209, 2010.
[11] S. Kim, L. Beckmann, M. Pistor, L. Cousin,
M. Walter, and S. Leonhardt. A versatile body sensor
network for health care applications. In Proc.
Intelligent Sensors, Sensor Networks and Information
Processing (ISSNIP), pages 175 –180, 2009.
[12] J. Kjeldskov and M. B. Skov. Supporting work
activities in healthcare by mobile electronic patient
records. In M. Masoodian, S. Jones, and B. Rogers,
editors, Computer Human Interaction, volume 3101 of
Lecture Notes in Computer Science, pages 191–200.
Springer Berlin / Heidelberg, 2004.
[13] C. Klersy, A. D. Silvestri, G. Gabutti, F. Regoli, and
A. Auricchio. A meta-analysis of remote monitoring of
heart failure patients. J. American College of
Cardiology, 54(18), 2009.
[14] P. Kranen, D. Kensche, S. Kim, N. Zimmermann,
E. Müller, C. Quix, X. Li, T. Gries, T. Seidl,
M. Jarke, and S. Leonhardt. Mobile mining and
information management in healthnet scenarios. In
Proc. Conf. on Mobile Data Management (MDM),
pages 215–216. IEEE, 2008.
Proceedings IMMoA’13 438 http://www.dbis.rwth-aachen.de/IMMoA2013/